van Inwagen, Peter

An Essay on Free Will

Oxford: Clarendon Press, 1983

Preface v-vi
I The Problems and How We Shall Approach Them 1
II Fatalism 23
III Three Arguments for Incompatibilism 55
IV Three Arguments for Compatibilism 106
V What Our Not Having Free Will Would Mean 153
VI The Traditional Problem 190
Notes 225
Index 246-8
van Inwagen, Peter

An Essay on Free Will

Oxford: Clarendon Press, 1983

Preface v-vi
I The Problems and How We Shall Approach Them 1
II Fatalism 23
III Three Arguments for Incompatibilism 55
IV Three Arguments for Compatibilism 106
V What Our Not Having Free Will Would Mean 153
VI The Traditional Problem 190
Notes 225
Index 246-8
 Vi PREFACE

There are, I think, only five twentieth-century works that

have had any very extensive influence on the present book:
C. D. Broad's "Determinism, Indeterminism and Libertarian-
ism", 1 R. E. Hobart's "Free Will as Involving Determination
Preface and Inconceivable Without It", 2 R. M. Chisholm's "Responsi-
bility and Avoidability", 3 Carl Ginet's "Might We Have No
Choice?' and Richard Taylor's Action ozdPurpose s. I discuss
This book considers several questions about free will, but the Hobart's article in Chapter IV. I do not discuss the other
bulk of it is addressed to the question whether free will and works, but their influence is pervasive. From the articles I have
determinism are compatible. Its answer is that, contrary to learned what arguments about free will are worth serious and
received opinion, they are not. extended consideration. Taylor's book has deeply influenced
There have appeared, in the last twenty years or so, a fair me; it is, I think, the source of a "picture" of human beings
number of books wholly or mainly about the problem of that is partly responsible for my most basic convictions about
free will and determinism. One should certainly mention free will.
M. R. Ayers's The Refutation of Determinism, Austin Farrer's Two works that have influenced me as regards special but
The Freedom of the Will, R. L. Franklin's Freewill and important points are Harry Frankfurt's "The Principle of
Determinism, Anthony Kenny's Will, Freedom and Power, Alternate Possibilities" 6 and G. E. M. Ansccanbe's inaugural
J. R. Lucas's The Freedom of the Will, A. I. Melden's Free lecture, Causality and Determination.'
Action, and D. J. O'Connor's Free Will. I have made no men- W. P. Alston read the book in manuscript. His comments
tion of these books in the body of the present work. This does have resulted in the rewriting of many confused and unclear
not reflect my judgement of the quality of these books, but passages.
rather the fact that the concerns of their authors are largely I have benefited from discussing the problem of free will
irrelevant to the topics I propose to discuss. I shall treat the and determinism with Keith Lehrer, Richard Taylor, Alvin
problem of the compatibility of free will and determinism Plantinga, and Carl Ginet. I have benefited from discussing
seriously and at length. The books I have listed either do not the problem of future contingencies with Margery Naylor.
treat this problem at all, or, at best, pass very quickly over Parts of this book have appeared elsewhere. I wish to thank
certain arguments the correct evaluation of which is essential to the editor of Noils for permission to reprint a few paragraphs
its solution. In particular, none of these books contains any- of "Laws and Counterfactuals" (1979), the editors of The
thing like an adequate discussion of the following argument: Philosophical Review for permission to reprint parts of
"Ability and Responsibility" (1978), D. Reidel Company
If determinism is true, then our acts are the consequences for permission to reprint parts of "The Incompatibility of
of the laws of nature and events in the remote past. But Free Will and Determinism" (Philosophical Studies, vol. 27,
it is not up to us what went on before we were born, and no. 3, 185-9, copyright @ 1975 by D. Reidel Publishing
neither is it up to us what the laws of nature are. There- Company, Dordrecht, Holland), the editor of Theoria for
fore, the consequences of these things (including our permission to reprint parts of "A Formal Approach to the
present acts) are not up to us. Problem of Free Will and Determinism" (1974), and the
The present book is essentially a defence of this argument Philosophy Documentation Center for permission to reprint
and an exploration of the consequences of supposing it to parts of "The Incompatibility of Responsibility and Deter-
be right, and, therefore, it owes little to any of the hooks I minism", which appeared in M. Brand and M. Bradie, eds.,
have mentioned. Action and Responsibility (Bowling Green: 1980).
 enduring philosophical interest, has been about this presupposition of the
Chapter I earlier debates about liberty and necessity. It is for this reason that
The Problems and How We Shall Approach Them nowadays one must accept as a fait accompli that the problem of finding
out whether free will and determinism are compatible is a large part,
1.1 There is no single philosophical problem that is "the problem of free perhaps the major part, of "the problem of free will and determinism".
will". There are rather a great many philosophical problems about free I shall attempt to formulate the problem in a way that takes account
will. 1 In this book I shall solve one of these problems and worry another at of this fait accompli by dividing the problem into two problems, which I
great length. will call the Compatibility Problem and the Traditional Problem. The
The problem I solve is the problem of "fatalism" or "future Traditional Problem is, of course, the problem of finding out whether we
contingencies". This problem will be dealt with in Chapter II, which is a have free will or whether determinism is true. But the very existence of
more or less self-contained essay. That chapter might have been left out of the Traditional Problem depends upon the correct solution to the
the book with almost no impairment of the argument of the remainder. It Compatibility Problem: if free will and determinism are compatible, and,
is included because it does, after all, bear on the question whether we have a fortiori, if free will entails determinism, then there is no Traditional
free will. And perhaps the fact that what it says is right will go some way Problem, any more than there is a problem about how my sentences can be
toward making up for its irrelevance to the other parts of the book. composed of both English words and Roman letters.
The problem of future contingencies is an old and honourable One of the main theses of this book is that the correct solution of the
philosophical problem, but it is hot one of the great central problems of Compatibility Problem does not imply the nonexistence of the Traditional
philosophy. The main topic of this book is one of the great central Problem; therefore my division of the problem of free will and
problems: the problem of free will and determinism. determinism into two is no idle exercise. But before I say more about this
division of the problem and about the ways in which I shall use it to
1.2 It is difficult to formulate "the problem of free will and determinism" organize this book, I shall explain what I mean by free will and
in a way that will satisfy everyone. Once one might have said that the determinism in sufficient detail to forestall certain possible
problem of free will and determinism — in those days one would have misunderstandings.
said 'liberty and necessity' — was the problem of discovering whether the
human will is free or whether its productions are governed by strict causal 1.3 In Chapter III, I shall formulate the thesis of determinism with as
necessity. But no one today would be allowed to formulate "the problem much precision as I am able to give it. Our present purposes will be served
of free will and determinism" like that, for this formulation presupposes by a short, preliminary account of what is meant by determinism.
the truth of a certain thesis about the conceptual relation of free will to Determinism is quite simply the thesis that the past determines a unique
determinism that many, perhaps most, present-day philosophers would future. But let us see what that might mean.
reject: that free will and determinism are incompatible. Indeed many Presumably, at any given moment there are many "possible futures",
philosophers hold not only that free will is compatible with many ways in which the world might go on. Or at
end of p. 1 end of p.2
determinism but that free will entails determinism. I think it would be fair least this is true if we understand 'possible' in a sufficiently liberal way:
to say that almost all the philosophical writing on the problem of free will there are certainly many "picturable" or "conceivable" or "consistently
and determinism since the time of Hobbes that is any good, that is of any describable" futures. But many of the futures that are possible in this sense
are impossible in another: they are physically impossible. For example, (1) if an event (or fact, change, state of affairs, or what have you)
though I can picture to myself what it would be like for there to be a total has a cause, then its cause is always itself an event (or what
eclipse of the sun this afternoon, though I can say without contradicting have you) and never a substance or continuant, such as a man;
myself that a total eclipse of the sun will be visible this afternoon, there is
an obvious sense in which this future I might imagine or describe is (2) if an event (or what have you) A was the cause of an event B, then it
physically impossible. To say this is not to say that it is contrary to the follows, given that A happened and given the laws of nature, that A
laws of nature — if we may allow ourselves this piece of terminology — "causally necessitated" B, that B could not have failed to happen;
that there should be an eclipse this afternoon, for the laws of nature do not
by themselves dictate when particular events like eclipses shall occur. 2 To (3) every chain of causes that has no earliest member is such that,
say that a "possible future" containing an imminent eclipse is physically for every time t, some event in that chain happens earlier than t.
impossible is rather to say that given the past, the actual past, and the laws
of nature, no eclipse will occur; that the past and the laws of nature It is easy to see why each of these premisses is necessary for the deduction
together rule out any possibility of an eclipse this afternoon. Those who, of determinism from the Principle of Universal Causation.
like me, do not object to talk of "possible worlds" may think of the matter Suppose that the Principle of Universal Causation is true and suppose
this way: while there are presumably possible worlds in which the laws of that premiss (1) is false. Suppose, that is, that the doctrine of immanent or
nature are the actual laws and in which there is an eclipse this afternoon, agent causation is true. 4 Suppose, to be more specific, that a certain
there is no possible world in which (i) the laws of nature are the actual change occurs in an agent, Tom, and Tom himself is the cause of this
laws, (ii) the past is the actual past, and (iii) there will be an eclipse this change, and no earlier state of affairs necessitated this change. Then the
afternoon. thesis of determinism is false. But our description of this case is internally
This example shows that there is a clear sense in which certain consistent, for it does not entail that any event is without a cause. In
"imaginable", "conceivable", or "consistently describable" futures are particular, the change in Tom has a cause: Tom himself. The cause of this
physically impossible. Determinism may now be defined: it is the thesis change, it is true, has no cause, but this does not entail the falsity of the
that there is at any instant exactly one physically possible future. 3 There Principle of Universal Causation, since its cause is not an event or change
must, of course, be at least one physically possible future; if there is more but a man, that is, a continuant. 5
than one, if at some instant there are two or more ways in which the world Suppose that the Principle of Universal Causation is true and suppose
could go on, then indeterminism is true. that premiss (1) is true and suppose that premiss (2) is false. That is,
Determinism in this sense must be carefully distinguished from what suppose that every event is caused by some earlier event or events but that
we might call the Principle of Universal Causation, that is, from the thesis these earlier causes do not necessitate, but merely produce, their effects.
that every event (or fact, change, or state of affairs) has a cause. It is far (That this supposition is consistent with our concept of causation — that is
from obvious what the logical relations that hold between these two theses to say, with the concept of causation, for every concept is the concept it is
are. I doubt, for example, whether the Principle of Universal Causation and is not some other concept — has been argued by Professor Anscombe
in her inaugural lecture. 6 I shall present
end of p.3
end of p.4
entails determinism. In order to deduce the latter from the former, we
should need at least three premisses: and defend similar arguments in Chapter IV.) I think it is easy to see that,
if these suppositions are correct, then, while every event has a prior cause,
the past nevertheless does not determine a unique future. (i) the phrase 'is a law of nature' is a real predicate: it is typically and
Suppose that the Principle of Universal Causation is true and suppose properly used in ascribing a certain property to certain objects
that both premiss (1) and premiss (2) are true and suppose that premiss (3) (unlike, say, 'exists', according to Kant, or 'is good', according to R.
is false. Then every event is caused by an earlier event that necessitates it; M. Hare);
nevertheless, determinism might be false, for as Lukasiewicz pointed out,
there might be a pair of times, t 1 and t2 , such that (i) a certain event A (ii) the objects that have this property are sentences or propositions (non-
happens at t2 , (ii) A is the final member of an infinite chain of causes, and linguistic entities expressed by sentences) or whatever it is that are
(iii) every member of this chain occurs later than t 1 . 7 In that case, we could the bearers of truth-value: anything that is a law is also either true or
consistently suppose that the past (up to and including ti) and the laws of false;
nature do not together determine whether A shall or shall not happen at t2 ,
and do not, therefore, determine a unique future. (iii) whether a proposition or sentence is a law is independent of what
It follows that to deduce determinism from the Principle of Universal scientists or others happen to believe or happen to have discovered: a
Causation we must assume (1), (2), and (3) as premisses. I doubt whether proposition, if it is a law, is unchangeably and objectively so, just as,
all three of these propositions are true. I am particularly doubtful about according to the prevailing view of mathematics, a proposition, if it
(2). 8 Therefore, I doubt whether universal causation entails determinism. is a theorem, is unchangeably and objectively so, whatever
I am uncertain what to say about the question whether determinism mathematicians or others happen to believe or happen to have
entails universal causation. Could there be, e.g., an explosion that was not proved.
caused by any earlier events but which was none the less inevitable, given
the past and the laws of nature? I think that anyone who answers Philosophers have on occasion proposed necessary conditions for a
immediately "Of course not!" reveals that a certain picture, or definition, proposition's being a law of nature. A law, for example, is supposed to be
or theory of causation has a very firm grip on him (which is not to say that true, to be contingent, to entail the existence of no particular (contingent)
he is wrong). But any real discussion of this question would lead us individual and to "support its counter-factuals" or "warrant inference to
needlessly, for we need not answer it, into a discussion of causation, subjunctive conditionals". But even if these conditions are individually
something I shall avoid whenever it is possible. necessary for lawhood, they are certainly not jointly sufficient. 9 Consider,
Having distinguished determinism from the Principle of Universal for example, the proposition that all men who are deprived of vitamin C
Causation, let us return to our examination of determinism. Our definition develop scurvy. This proposition is presumably both true and contingent.
of determinism presupposes some understanding of the notion of a law of Moreover, we may suppose that it "supports its counter-factuals": every
nature. I should like to define 'law of nature' in its turn, but I do not know man, past, present, or future, is such that if he were deprived of vitamin C,
how. I will mention, however, some constraints on an adequate definition he would develop scurvy. 10 But it hardly follows that this general
of this concept and some necessary conditions for its application. These proposition is a law of nature, for the fact that it supports its counter-
remarks, though they will not constitute factuals might depend on quite accidental circumstances. Suppose, for
example, that there is such a thing as "vitamin X", which could be used as
end of p.5 an effective replacement for vitamin C; but suppose that all the vitamin X
there is is
a definition of 'law of nature', will at least show how my use of this term

differs from that of some other writers. I should require any definition of end of p.6
'law of nature' to have the following three consequences:
locked in a vault on Mars. If that is true, then while our proposition does said is as much a part of a complete description of the world as are the
in fact support its counter-factual instances, its supporting them depends most ordinary factual statements of the geographer or the historian. And I
upon the accidental circumstance that the sole supply of vitamin X is think it is plain that what I have said I could not have said without
inaccessible to human beings. Or, again, we may imagine that all men are employing the concept of a law of nature, or, at least, without employing
such that they would develop scurvy if they were deprived of vitamin C, some essentially equivalent concept like the concept of physical necessity.
but that if an accident involving certain radioactive materials had Therefore the notion of a law of nature makes sense, even if no one knows
happened at a certain time and place, some of the witnesses would have how to explain it to one who has not yet acquired it. (How you and I did
had descendants whose bodies were capable of synthesizing vitamin C acquire it is a question for the epistemologist or the historian of science; I
and who would therefore not develop scurvy under any conditions of diet. am content to point out that we have it.) And, therefore, my definition of
It seems to me to be obvious that our concept of a law of nature entails determinism, though it may rest on an undefined concept, at least rests on
that the possession of lawhood by a proposition cannot depend on such an undefined concept we have.
accidental occurrences as these. Therefore, these examples show that the Let us now see what can be done about defining free will.
conditions we have been examining are not jointly sufficient for lawhood.
I know of none that are — except, of course, sets of conditions that are 1.4 I use the term 'free will' out of respect for tradition. My use of the term
trivially sufficient, such as sets involving the requirement that a law be a is not meant to imply that I think there is such a "faculty" as "the will".
physically necessary proposition — and neither, I think, does anyone else. When I say of a man that he "has free will" I mean that very often, if not
Nevertheless, 'law of nature' seems to be an intelligible concept and one always, when he has to choose between two or more mutually
we can't get along without if we wish to give a complete description of the incompatible courses of action — that is, courses of action that it is
world. Let me give a simple example of this that is quite independent of impossible for him to carry out more than one of — each of these courses
any problems about determinism. of action is such that he can, or is able to, or has it within his power to
I have recently read an article on the possibility of inter-sidereal carry it out. A man has free will if he is often in positions like these: he
travel in which the authors divide the unpleasant necessities of this sort of must now speak or now be silent, and he can now speak and can now
travel into two categories: those imposed upon the travellers by the remain silent; he must attempt to rescue a drowning child or else go for
ignorance of the designer of their vehicle, and those imposed upon the help, and he is able to attempt to rescue the child and able to go for help;
travellers by the laws of nature. This is an important distinction. he must now resign his chairmanship or else lie to the members, and he
Spaceships and other artefacts are doubtless never perfect. But certain has it within his power to resign and he has it within his power to lie.
disadvantages of intersidereal travel are not going to be removed by 'Free will', then, is to be defined in terms of 'can'. But how is 'can' to
technological advance as the corresponding disadvantages of inter- be defined? I am afraid I do not know how to define 'can', any more than I
continental travel were removed. Inter-continental travel, now a matter of know how to define 'law of nature'. Nevertheless, I think that the concept
hours, was once a matter of months or years. But intersidereal travel, if it expressed by 'can' in the examples given in the preceding paragraph — the
should ever come to pass, will always be a matter of years or centuries. No concept of the power or ability of an agent to act — is as clear as any
technological advance could ever change this unfortunate fact, for it is a philosophically interesting concept is likely to be. In fact,
consequence of the laws of nature. 1 1
end of p.8
end of p.7
I doubt very much whether there are any simpler or better understood
Now I have just said something about the way things are; what I have concepts in terms of which this concept might be explained. There are,
however, concepts with which the concept of human power or ability and the first is of no particular interest to us.
might be confused, either because they really are similar to the concept of (iv) The concept of the power or ability of an agent to act is not the
power, or because they are sometimes expressed by similar words. concept of causal power or capacity. We say that penicillin has the power
Perhaps what I say in the sequel will be clearer if I explicitly distinguish to kill certain bacteria, that a hydrogen bomb is capable of destroying a
the concept of power or ability from those concepts with which it might be large city, and that a certain computer can perform a thousand calculations
improperly conflated. 12 per second. (These are statements about capacities that may be unrealized.
(i) The concept of the power or ability of an agent to act is not the The vocabulary of our talk about the realization of causal capacities and
concept of moral or legal permissibility. I might say to a hard-hearted the vocabulary of agency similarly overlap: we talk about the action of
landlord, "You can't simply turn them out into the street", knowing full hydrochloric acid on zinc and the action of an automatic pistol.) But this
well that, in the sense of 'can' that is our present concern, he very well can. sort of talk is really very different from talk of the power of an agent to
The popular retort, "Oh, can't I? Just you watch!" is, I should think, act, despite their common origin in the technical terminology of medieval
usually a play on these two senses of 'can', though sometimes it may be an Aristotelianism. Perhaps the best way to appreciate this difference is to
expression of a genuine confusion between them. And in some few cases, examine some ascriptions of causal capacity to human agents and to
cases typified by the use of words like 'I can't go through with it', it may contrast them with ascriptions of ability to human agents.
be that no one, either agent or spectator, can say with any confidence We ascribe a capacity, rather than an ability, to an agent when we say
whether can is being used to express the idea of power or the idea of he:
permissibility.
(ii) The concept of the power or ability of an agent to act is not the can digest meat (being an omnivore);
concept of physical possibility, nor is power or ability entailed by physical has absolute pitch;
possibility. This can be shown by a simple example. Suppose I have been has normal colour vision;
locked in a certain room and suppose that the lock on the door of that can understand French.
room is a device whose behaviour is physically undetermined; it may
come unlocked and it may not: there is a future consistent with both the We ascribe an ability, rather than a capacity, to an agent when we say he:
actual past and the laws of nature in which an internal mechanism unlocks
the lock and another such future in which it doesn't. Then it is physically can cook meat;
possible that I shall leave the room. But it does not follow that in any can sing correctly;
relevant sense I can leave the room. can name the colours;
(iii) The concept of the power or ability of an agent to act is not the can speak French.
concept of epistemic possibility . Consider the sentence, 'Castro could have
arranged for Kennedy's assassination'. Clearly there are at least two things Despite their superficial similarity, there is all the difference in the world
that someone who spoke these words might mean by them. He might between the sort of property that the predicates in the first list ascribe to an
mean, 'For all we know, Castro did arrange for Kennedy's assassination', agent and the sort of property that those in the second list ascribe to an
or he might mean, 'Castro had it within his power to arrange agent. Consider, for example, the last item in each list. For a man to have
the capacity to understand French is for him to be such that if he were
end of p.9 placed in certain circumstances, which wouldn't be very

for Kennedy's assassination'. These senses are obviously quite different end of p.10
hard to delimit, and if he were to hear French spoken, then, willy-nilly, he Secondly, I do not mean to imply that this distinction is, at least in
would understand what was being said. But if a man can speak French, it any very straightforward way, supported by ordinary usage. There is
certainly does not follow that there are any circumstances in which he certainly nothing wrong with saying that someone is able to understand
would, willy-nilly, speak French. The concept of a causal power or French, and probably nothing wrong with saying that a certain king lacks
capacity would seem to be the concept of an invariable disposition to react the capacity to rule.
to certain determinate changes in the environment in certain determinate Thirdly, I have been making a conceptual distinction. No ontological
ways, whereas the concept of an agent's power to act would seem not to be conclusions should be drawn from the existence of this distinction.
the concept of a power that is dispositional or reactive, but rather the Nothing I have said entails that the abilities of agents are not in some
concept of a power to originate changes in the environment. 13 sense "reducible to" or do not "supervene upon" the causal capacities of
Three points need to be made about what I have said about capacities the agents — or of some parts of agents, such as organs, cells, or atoms —
and abilities. and their environment. Here are two analogous cases that may make this
First, it does not pretend to be an analysis of the distinction between point clearer: the concept of a number is not the concept of a set, but that
capacities and abilities. It is, rather, an argument by example for the does not mean that statements about numbers cannot be reduced to
existence of this distinction. I do not know how to give a general account statements about sets or that there are numbers in addition to sets; the
of it. (The reader may have noticed that I rarely attempt to give any concept of a mental event is not the concept of a physical event, but that
general account or analysis of a concept, being content, in problematical does not mean that statements about mental events cannot be reduced to
cases, to try to show that we have a concept answering to a certain statements about physical events or that there are mental events in
description and to try to distinguish it from other, similar concepts. In addition to physical events.
general, I am suspicious of philosophers' "analyses" of concepts, which I repeat: the purpose of this discussion of abilities and capacities,
seem to me to be only rarely correct and almost always tendentious. In the unlike the purposes of the discussions of most other philosophers who
course of this book, I shall frequently appeal to our understanding of have made this distinction, has been conceptual and not ontological. My
various unanalysed concepts — such as the concept of an agent's power to only purpose has been to state as clearly as I can what concept it is I shall
act — in order to convince the reader that one of my premisses is true. But be using 'can' and `able' and 'power' to express. It is this sense, of course,
note that if I were to offer a philosophical analysis of these concepts, I that these words bear in the above definition of free will. And it is free
should have to appeal to our pre-analytical understanding of them as part will as defined in the present section that I shall argue is incompatible with
of my argument for the correctness of my analysis. The appeal to intuition determinism as defined in the previous section.
must turn up at some level of discourse. I prefer making "nonce" appeals (v) The concept of the power or ability of an agent to act is not the
to intuition at specific points in the argument to making the very abstract same as the concept of a skill or an accomplishment. In the preceding
and general appeals to intuition that are inevitable when one is defending a discussion of abilities and causal capacities, I used the predicate 'can speak
philosophical analysis. This rationale for my procedure is, of course, self- French' as an example of a predicate that expresses the power of an agent
serving, since I almost never know of any plausible analysis of the to act. I did this because 'can speak French' stands in instructive opposition
concepts I employ. I should add that my definitions of terms — such as to the capacity-predicate 'can understand French'. But there is more to be
'determinism' — are not supposed to be analyses of concepts but said about 'can speak French'. Suppose that Jean-Paul, a valiant member of
explanations of my own technical terminology.) the Resistance, has been captured by the Germans and bound and gagged.
Can he speak French?
end of p. 1 1
end of p.12
Well, to be able to speak French is to be able to speak, and. he can't speak end of p.13
because he is gagged: so he can't speak French. On the other hand, if the
German commander ordered all prisoners who could speak French that we have free will. 14 (I should like to appropriate libertarianism' as a
brought before him, he would be unlikely to look approvingly on the name for the thesis that we have free will, but the incompatibilistic
action of the subordinate who produced only the ungagged French- associations of this term are too firmly ingrained in too many readers for
speaking prisoners ("For the others, Herr Oberst, cannot speak simpliciter, this to be wise. I have toyed with the idea of using ` elutherianism' for this
and, a fortiori, cannot speak French"). purpose, but reason has prevailed. I have finally and reluctantly settled on
Clearly there is a distinction to be made between a skill, `the free-will thesis'.) I object to these terms because they lump together
accomplishment, or general ability, on the one hand, and, on the other, the theses that should be discussed and analysed separately. Even having them
power to exercise it on a given occasion. This is true despite the fact that on hand is a permanent temptation to conflate the Traditional Problem and
the same words might be used in both kinds of situation (`can speak the Compatibility Problem. They are therefore worse than useless and
French'; 'can move her left arm'; 'can play the flute'). That is not to say that ought to be dropped from the working vocabulary of philosophers.
there may not be a close conceptual connection between the two. I should There is one other term that commonly figures in discussions of free
think, - in fact, that a statement ascribing a skill or other general ability to will and determinism that I shall avoid: 'contra-causal freedom'. This term
an agent is probably equivalent to some statement asserting that, under is highly ambiguous and, moreover, in accepting incompatibilism the
certain conditions, that agent has the power to perform acts that fall under believer in free will commits himself to accepting none of the things it
certain descriptions. But I shall not pursue this question, since it is not might mean.
relevant to our present concerns. It is plain that the 'can' that figures in 'Contra-causal freedom' might mean the sort of freedom, if freedom it
discussions of free will and determinism is not the `can' of skill: the thesis would be, that someone would enjoy if his acts were uncaused. But, as we
of determinism may or may not be relevant to the question whether shall see in Chapter IV, that someone's acts are undetermined does not
someone on a particular occasion can or cannot speak French; it is entail that they are uncaused. (This point was briefly touched on in Section
certainly irrelevant to the question whether that person is a French- 1.3.)
speaker. 'Contra-causal freedom' might mean the sort of freedom, if freedom it
would be, that someone would enjoy if in acting he violated the laws of
1.5 It is in these senses that I shall understand 'free will' and `determinism'. nature, that is, if he worked miracles — even if only small miracles, local
In Chapters III and IV, I shall argue that free will is incompatible with to the motor centres of his brain. Now I am not one of those philosophers
determinism. It will be convenient to call this thesis incompatibilism and who think that miracles are conceptually impossible. It seems to me that if
to call the thesis that free will and determinism are compatible God created ex nihilo a spinning object, then the proposition we call 'the
compatibilism . I have no use for the terms 'soft determinism', 'hard law of the conservation of angular momentum' would be false. Yet, it
determinism; and `libertarianism'. I do not object to these terms on the seems to me, it might be a law of nature for all that. I think I understand
ground that they are vague or ill-defined. They can be easily defined by the notion of a supernatural being, that is, the notion of an agent who is
means of the terms we shall use and are thus no worse in that respect than superior to and not a part of Nature (this enormous object that the natural
our terms. Soft determinism is the conjunction of determinism and sciences investigate), and I think that the falsity of a proposition counts
compatibilism; hard determinism is the conjunction of determinism and against its being a law of nature if and only if that falsity is due entirely to
incompatibilism; libertarianism is the conjunction of incompatibilism and the mutual operations of natural things, and not if it is due to the action of
the thesis such an "external" agent upon Nature. But it does not follow from this
perhaps rather quaint thesis about the concept of miracle that we can
end of p.14 end of p.15

perform miracles, for there is no reason to suppose we are supernatural be afraid of being obvious. Here is an argument that I think is obvious (I
beings. And even if we are supernatural beings, that we are is not a don't mean it's obviously right; I mean it's one that should occur pretty
consequence of the joint truth of the freewill thesis and incompatibilism. If quickly to any philosopher who asked himself what arguments could be
these two theses are true, then determinism is false, and, moreover, one's found to support incompatibilism):
free choices are undetermined: if I have a free choice between A and B,
then my doing A is consistent with the past and the laws of nature and so If determinism is true, then our acts are the consequences of the laws
is my doing B. Incompatibilism, therefore, entails that neither my freely of nature and events in the remote past. But it is not up to us what
doing A nor my freely doing B would "violate" a law of nature. It follows went on before we were born, and neither is it up to us what the laws
that it is sheer confusion to attribute a belief in contra-causal freedom, in of nature are. Therefore, the consequences of these things (including
the present sense, to the incompatib ilist who believes in free will. our present acts) are not up to us.
Finally, "contra-causal freedom" might be attributed to an agent if
that agent has it within his power to act contrary to the laws of nature; that I shall call this argument the Consequence Argument . In Chapter III,
is, if the agent is able to perform certain acts whose performance would be I shall argue for incompatibilism by presenting three very detailed
sufficient for the falsity of certain propositions that are in fact laws of versions of the Consequence Argument. Or, if you like — how does one
nature. But the incompatibilist, whether or not he accepts the free-will count arguments, anyway? — I shall present three detailed arguments each
thesis, does not believe in contra-causal freedom in this sense. Quite the of which is suggested by the Consequence Argument.
contrary: it is because the incompatibilist believes that we do not possess In Chapter IV, I shall examine three arguments for compatibilism:
this power with respect to the laws of nature that he believes free will the "Paradigm Case Argument", the "Conditional Analysis Argument",
requires indeterminism. The precise sense in which this is true will be and an argument I shall call the Mind Argument because it has appeared
evident from an inspection of the arguments for incompatibilism that will so often in the pages of that journal. 15
be presented in Chapter III. The Paradigm Case Argument attempts to establish the compatibility
Incompatibilism, therefore, may perhaps be described as the thesis of free will and determinism by an examination of cases of human action
that free action is "extra-causal"; to say it is the doctrine that free action is that, it is alleged, serve as paradigms for the teaching of the meanings of
"contra-causal" can only lead to confusion. words like 'free'.
The Conditional Analysis Argument maintains that statements
1.6 Incompatibilism can hardly be said to be a popular thesis among ascribing to human agents the power or ability to act otherwise are to be
present-day philosophers (the "analytic" ones, at any rate). Yet it has its analysed as disguised conditionals, which, when their disguise is removed,
adherents and has had more of them in the past. It is, however, can be seen to be compatible with determinism.
surprisingly hard to find any arguments for it. That many philosophers The Mind Argument proceeds by identifying indeterminism with
have believed something controversial without giving any arguments for it chance and by arguing that an act that occurs by chance, if an event that
is perhaps not surprising; what is surprising is that no arguments have occurs by chance can be called an act, cannot be under the control of its
been given when arguments are so easy to give. Perhaps the explanation is alleged agent and hence cannot have been performed freely. (In this brief
simply that the arguments are so obvious that no one has thought them statement of the Mind Argument I have left entwined three strands of
worth stating. If that is so, let us not reasoning that I shall disentangle in Chapter IV.) Proponents of the Mind
Argument conclude, therefore, that free will
end of p.16
1.7 If the arguments of Chapter III are correct, then incompatibilism is
is not only compatible with determinism but entails determinism. true. These arguments have premisses. Some of the premisses are more
I shall argue in Chapter IV that these arguments fail. There are, I controversial than others. That is, some of the premisses of Chapter III
think, no other arguments for compatibilism that need be taken seriously. will be accepted without question by the compatibilist and others he will
Let me give just one example of an argument that need not be taken want to argue about. Let us call the conjunction of these "controversial"
seriously. Some philosophers have urged that to suppose that free will and premisses P . Suppose I am willing to grant that if any of my premisses are
determinism are in conflict is to confuse compulsion with determination false, the false ones are conjuncts of P . The compatibilist and I will thus
by causal laws. For, these philosophers argue, to do something of one's agree that if compatibilism is true, then P is false. One compatibilist has
own free will is to do that thing without being compelled to do it, and to actually argued (in effect) that this proposition on which he and I agree
behave in accordance with a deterministic set of causal laws is not to be entails that I am begging the question against compatibilism by assuming
compelled. I reply that this argument confuses doing things of one's own the truth of P . 16
free will with having free will about what one does. These are not the It should suffice to point out that the situation in which this argument
same thing. The following case shows this. Suppose a certain man is in a places the compatibilist and me is a perfectly symmetrical one: I am in a
certain room and is quite content to be there; suppose, in fact, that he position to employ the same argument, mutatis mutandis , to prove the
wants very much to. remain in that room and that it would be difficult conclusion that his choice of premisses begs the question, and I should be
indeed to induce him even to consider leaving it. Then, I should think, he as well justified in employing the argument against him as he is in
remains in the room of his own free will. But we can with perfect employing it against me. Now why, I have asked myself uneasily, would
consistency go on to suppose that he has no free will about whether he anyone say something that can be so easily refuted? I think there are two
leaves the room: suppose that, unknown to him, the only door is locked possible answers. (i) The compatibilist thinks that compatibilism is prima
and that it is not within his power to open it. facie true and that the burden of proof lies on the incompatibilist. (When a
Now someone might want to say that our imaginary agent did not philosopher says, "The burden of the proof lies on you", he means, "You
remain in the room "of his own free will". I don't think this is right, but I must deduce your conclusion from the truths of immediate sensory
will not argue the point. For if it is true that our agent did not remain in the experience by means of an argument that is formally valid according to
room "of his own free will", then we cannot establish that a person does the rules of elementary logic; I on the other hand may employ any
something "of his own free will" by establishing that his doing it is dialectical tactic I find expedient".) (ii) He thinks that the premisses from
uncompelled. But the "compulsion" argument we have been considering which he derives compatibilism are more plausible than those from which
certainly does depend on the premiss that one can so establish that a I derive incompatibilism.
person has acted "of his own free will". Let us grant this premiss. The As to the first of these possibilities, I deny that compatibilism is
example of the man locked in the room shows that it does not follow from prima facie right and incompatibilism prima facie wrong. Quite the other
a person's doing something "of his own free will" that he can do way round, if you ask me. But I shall not assume that either of these
otherwise. And thus it does not follow from the undoubted fact that we propositions is prima facie right. I shall treat them as philosophical theses
often do things of our own free will that what I have called the free-will of equal initial plausibility, and this, it seems to me, is the only reasonable
thesis is true. And, therefore, the Compatibility Problem is not going to be way to approach the Compatibility Problem.
solved by jejune reflections on compulsion. In Section 4.3, I shall attempt to undermine the thesis that

end of p.17 end of p.18

the premisses of the compatibilist are prima facie more plausible than my right, it is none the less true that moral responsibility is possible only if we
own by comparing the premisses of one argument for compatibilism, the have free will.
Conditional Analysis Argument, with the premisses of one of the Now some philosophers will perhaps want to protest at this point in
incompatibilist arguments presented in Chapter III. I think that the the argument that while I may indeed have shown that in some sense free
unprejudiced reader — if such can be found: very likely anyone willing to will is incompatible with determinism, and while I may have shown that
read as far as Chapter IV will have brought strong feelings about the free will in some sense is logically necessary for moral responsibility, I
Compatibility Problem to his reading of this book — will agree that if the have not shown that there is any single notion of free will that has both
premisses of my argument and the premisses of the Conditional Analysis these features. Therefore, these philosophers may allege, I am not in a
Argument are considered independently of their implications for the position to say that considerations having to do with moral responsibility
Compatibility Problem, then the former will be seen to be considerably can be used to show that we ought to accept the doctrine called 'free will'
more plausible than the latter. 17 in Chapters III and IV and there shown to be incompatible with
determinism.
1.8 If the arguments of Chapters III and IV are cogent, then I shall have I shall meet this possible objection in two ways. First, I shall ask the
provided good reasons for thinking that free will and determinism are reader to examine the premisses of the arguments of Chapter III after they
incompatible. Suppose I am right. Suppose these theses are incompatible. have been rewritten according to the following rule: at each place at which
Which, if either, ought we to accept? As a first step towards answering the words 'free will' occur in the premisses of these arguments, replace
this question, I shall, in Chapter V, address the question, "What would it them with the words 'free will in just that sense offree will that is relevant
mean to reject free will?" I shall argue that such a rejection would have at to questions of moral responsibility'. I contend that the reader will find
least two interesting consequences. that the rewritten premisses are no less plausible than the original ones.
First, anyone who rejected free will could not consistently deliberate Secondly, I shall present an argument for the incompatibility of moral
about future courses of action. This is so, I shall argue, owing simply to responsibility and determinism that makes no mention whatever of free
the fact that one cannot deliberate without believing that the things about will, though it will be structurally identical with one of the arguments for
which one is deliberating are things it is possible for one to do. How the incompatibility of free will and determinism that occurs in Chapter III.
important one takes this consequence to be will, of course, depend on how I think this "direct" argument for the incompatibility of responsibility and
important one thinks consistency is. determinism will have the following feature: it will clearly be sound if and
Secondly, and, I think, much more importantly, to deny the existence only if the corresponding argument of Chapter III is sound. But if this is
of free will commits one to denying the existence of moral responsibility. true, then it seems very unlikely that it is only some sort of "free will" that
Until a short while ago, most philosophers would have taken this to be has nothing to do with moral responsibility that is shown in Chapter III to
obvious. But if any of these philosophers had been asked to defend this be incompatible with determinism.
obvious thesis, he would almost certainly have appealed to the following The principal conclusion of Chapter V will therefore be that to reject
principle: a person can be held morally responsible for what he has done free will — in just that sense of 'free will' in which we have earlier argued
only if he could have done otherwise. In a recent remarkable article, that free will is incompatible with determinism — is to reject moral
however, Harry Frankfurt has presented convincing counter-examples to responsibility.
this principle. I shall devote the bulk of Chapter V to showing that even if In Chapter VI, I shall discuss the Traditional Problem, that is, the
Frankfurt is problem of finding out whether determinism is true, or whether the free-
will thesis is true, or whether neither is true.
end of p.19
end of p.20 end of p.21

I shall proceed by asking what reasons we have for thinking that since there is such a thing as moral responsibility, there is such a thing as
determinism is true and what reasons we have for thinking that the free- free will. (Moreover, since free will is incompatible with determinism,
will thesis is true. determinism is false.) The first premiss of this argument is defended in
There are two sorts of reason for believing in determinism. First, one Chapter V in the way outlined above. In Chapter VI, we shall examine its
might believe in determinism because one believes that science has shown second premiss, and I shall defend my use of this argument against the
determinism to be true. I shall argue at length that science has, if anything, charge that for an incompatibilist so to argue amounts to his claiming to be
shown determinism to be false. Secondly, one might believe that able to prove that determinism — a thesis about the motion of matter in
determinism is a truth of reason, on the ground that it is a logical the void — can be shown to be false by a priori reflection on moral
consequence of the Principle of Sufficient Reason. I shall show that, responsibility. I shall also examine a condition of certain philosophers
whether or not determinism is a consequence of the Principle of Sufficient (their having fallen under the spell of "scientism") that makes it
Reason, the Principle of Sufficient Reason must be rejected, since it psychologically very difficult for them to believe that such "tender-
entails the collapse of all modal distinctions. minded" arguments as this could possibly provide one with good reason to
In addition to asking what reasons might be brought in support of reject determinism.
determinism, we shall also ask what reasons might be brought in support I shall finally address the question, "What would you say if, after all,
of a certain closely related but weaker thesis: the thesis that human beings, the progress of science did show that indeterminism was untenable?"
like typical digital computers, behave in ways that are entirely determined
by their past states and their current "input". It is important to consider this
thesis because: (i) it would seem to be incompatible with free will if and
only if determinism proper is; (ii) unlike determinism proper it is not in
even prima-facie conflict with current physical theory; and (iii) it is
commonly contended that the empirical study of human beings has
uncovered facts that strongly support it. I shall argue that this common
contention is sheer bluff.
The conclusion of the argument whose course is summarized in the
last few paragraphs is that neither physics nor pure reason supports
determinism, and, moreover, that the scientific study of human beings
does not support the thesis that the behaviour of human beings is "for all
practical purposes" determined.
What reasons can be brought in support of the free-will thesis? It
cannot, I think, be seriously maintained that we can know by some sort of
introspection that we have or that we do not have free will. And neither
can it be maintained that the empirical study of human beings is likely to
show us that we have or that we do not have free will. The only relevant
argument would seem to be this: if we do not have free will, then there is
no such thing as moral responsibility; therefore,
 Chapter II
Fatalism

2.1 Fatalism, as I shall use the term, is the thesis that it is a

logical or conceptual truth that no one is able to act otherwise
than he in fact does; that the very idea of an agent to whom
alternative courses of action are open is self-contradictory.'
The word 'fatalism' is used in philosophy in at least two other
senses: it is used (i) for the thesis that what is going to happen
is inevitable, and (ii) for the thesis that no one is able to act
otherwise than he in fact does. (This latter thesis is entailed
b y b u t does not entail what I am calling fatalism.) So long as
it is understood that neither of these theses is what I mean
by 'fatalism', no confusion will result from this plurality of
senses. But the idea of the inevitability of what is going to
happen is so commonly associated with the word 'fatalism'
that I feel I should say something about it.

2.2 Suppose a witch predicts that I shall drown within the

next twenty-four hours. She also predicts that I shall attempt
to evade this fate, and that my efforts will be in vain. Here are
two stories about how her prediction might come true.
(a) I determine to spend the next twenty-four hours at the
top of a high hill. But as I leave the witch's hovel, I am over-
powered by three assailants in the employ of an enemy of
mine, who, despite my struggles, carry me to a nearby pond
and hold me under water till I am dead.
(b) I determine to spend the next twenty-four hours at the
top of a high hill. While climbing the hill, I fall into a hidden
well and drown. Moreover, if I had simply gone about my
business and done nothing in particular to avoid drowning, I
should not have drowned.
In each of these stories, the witch's prediction, that my
efforts to avoid drowning would be in vain, came true. But
24 FATALISM FATALISM 25

there is an important difference between them. In story (a), I in story (a) had told me exactly how I should be drowned,
should have drowned no matter what I had done, and this is this would not have enabled me to escape drowning.
not a feature of story (h). And yet story (b) produces—in me Let us introduce some philosophical jargon. Let us say of a
at any rate--the sort of feeling one might express by saying, certain future event (an event that is in fact going to happen)
"Yes, that's what inescapable fate means". how is this feeling that it is strongly inevitable for me if it would happen no
produced? Let us examine a more artful story of the same matter what I did.' Similarly, a future state of affairs is strongly
kind. inevitable for me if it would obtain no matter what I did. Let
us say that a future event, or state of affairs, is weakly inevi-
SHEPPY: I wish now I'd gone down to the Isle of Sheppey when the
doctor advised it. You wouldn't 'ave thought of looking for me there.
table for me if both of the following conditions hold: (i) it is
not strongly inevitable, and (ii) if I tried to take measures to
DEATH: There was a merchant in Bagdad who sent his servant to market
to buy provisions and in a little while the servant came back, white
prevent it, then I should choose the wrong measures out of
and trembling, and said, Master, just now when I was in the market- ignorance, and it is strongly inevitable for me that I should be
place I was jostled by a woman in the crowd and when I turned I saw ignorant of the right measures.
it was death that jostled me. She looked at me and made a threaten- In story (a), my death by drowning at a certain moment
ing gesture; now, lend me your horse, and I will ride away from this
city and avoid my fate. I will go to Samarra and there death will not
was strongly inevitable for me at the time of the witch's predic-
find me. The merchant lent him his horse, and the servant mounted tion. In story (b), my death by drowning at a certain moment
it, and he dug his spurs in its flanks and as fast as the horse could gallop was weakly inevitable for me at the time of the witch's pre-
he went. Then the merchant went down to the market-place and he diction, at least assuming that nothing I might have done there-
saw me standing in the crowd and he came to me and said, Why did after would have been sufficient for getting a more detailed
you make a threatening gesture to my servant when you saw him this
morning? That was not a threatening gesture, I said, it was only a prediction from the witch. In Death's story, it was weakly in-
start of surprise. I was astonished to see him in Bagdad, for I had an evitable for the servant that he should meet her in Samarra
appointment with him to-night in Samarra. that night, at least assuming that nothing he might have done
SHEPPY: [With a shudder.] D'you mean there's no escaping you? before arriving in Samarra would have been sufficient for dis-
DEATH: No.'
covering that it was `Sarnarra' and not 'Bagdad' that was writ-
ten beside his name in her appointment book. (The "effect"
Death seems to imply that she and the merchant's servant of these stories, their "atmosphere" of inescapable fate, is
would have met that very night, no matter what he had done. not due solely to the fact that they are stories about men
But this is false. They would not have met that night if he whose death at a certain moment is weakly inevitable. These
had remained in Bagdad. (Otherwise, the point of the story is literary qualities also owe a great deal to the fact that their
lost.) But then why does Death's story have the effect on us protagonists' incomplete foreknowledge has magical or super-
of making her seem inescapable? Well, because her story leads natural sources, to the fact that their protagonists in fact
us to believe that whatever one attempts to do to avoid her attempt to escape death, and to the fact that the very measures
will be just the wrong thing. In this sense, she is inescapable. their protagonists take to escape death contribute to their
But notice that the fact that there's no escaping her depends deaths.)
on our inevitable ignorance of what she has got written down It is obviously possible, magical or supernatural prediction
in her little (presumably black) book. Such stories as story (b) aside, that some events be strongly inevitable, and possible
above and Death's story and the story of Oedipus depend for that some events be weakly inevitable. It is, for example,
their effect on the ignorance of their protagonist about the strongly inevitable for me and for anyone else that the sun
way in which a prediction will come true. Story (a) does not rise tomorrow. As to weak inevitability, suppose I am in a
depend on the ignorance of its protagonist. Even if the witch burning building—I do not yet realize it is burning, but shall
26 FATALISM FATALISM 27
in a moment—from which there are exactly two possible exits. that there was good reason to believe he should die tomorrow
And suppose that all the following propositions are true: at noon, then he would not die tomorrow at noon. Consider,
(i) if I do not try to leave the building, I shall be burned for example, Mergendus, who will die in a boating accident at
to death; noon tomorrow. Suppose, contrary to fact, that you or I were
to predict in Mergendus's hearing that he should die at noon to-
(ii) if I try to leave by the nearer exit, I shall be burned to morrow, and that Mergendus -he is a superstitious man—took
death; our prediction seriously enough to believe that his life was in
(iii) if I try to leave by the more remote exit, I shall succeed danger, and that he had better take care. What would happen?
and save my life; What I should expect would happen is that Mergendus would
(iv) I have no reason to think that either exit is more likely refuse to leave his house tomorrow (much less, go boating)
to lead to safety than the other, and have no way of and that, owing to this precaution, he would not die at noon
finding out if either exit is preferable; tomorrow. But, of course, there is no intrinsic absurdity in
supposing that what would happen is that Mergendus would
(v) if I believed I were in danger and saw two routes of
possible escape, I should always choose the nearer unless refuse to leave his house tomorrow and would be killed in his
I had some good reason to regard the more distant as own bed by a meteorite at just the moment at which he would
preferable. have drowned if he had gone boating. In fact, someone or other
has probably been in just this doubly unfortunate position:
If (i)-(v) arc true, then my death by burning is now weakly even if he had taken steps sufficient to avoid whatever it was
inevitable for me. that in fact caused his death at a certain moment, he would,
But if it is possible that some events be inevitable, in either none the less, have died at just that same moment because,
sense, for someone, could it be that all events are inevitable, to speak theromorphically, another death was lying in wait.
in either sense, for anyone or everyone? Let us first consider for him.
weak inevitability. Taken literally, this question must be There seems to be no conceptual absurdity in supposing
answered "No": it is certainly not weakly inevitable for me that this is always the case: it is conceptually possible that,
that I shall continue writing for at least the next five minutes, for every person, there is a moment such that he will die at
though, unless I am very much mistaken, I shall in fact do that moment and such that his death at that moment is weakly
this. A more interesting question is whether some important inevitable for him if it is not strongly inevitable. But we can-
subclass of the class of future events is weakly inevitable for not take seriously any suggestion that this conceptual possi-
everyone. For example, everyone will die at some particular bility is in fact realized. If it were realized, this would either
moment; could it be that, for every living person, there is be simply an enormous, meaningless multiple accident, an
some future moment such that it is weakly inevitable for him accident of more than astronomical improbability, or else the
that he shall die at that moment? result of the manipulations of some cosmic "puppet-master"
Let us look carefully at this question. Consider the class of like Death or Death's employers in Maugharn's play. And surely
people who will die at noon tomorrow. Many of these people, a belief in a personified Death or Fate, or whatever one might
unfortunately, are in excellent health, and no one could now want to call a cosmic puppet-master, is a belief for which
predict any of the fires, bathing accidents, acts of political 'superstition' is, if anything, too flattering a word.'
terrorism, and so on, that will cause their deaths. But this is Let us turn now to strong inevitability. Is there any reason
not sufficient for the weak inevitability of any of their deaths. to think that all future events, or some important and interest-
It seems enormously likely that at least one of these people is ing class of future events, are strongly inevitable for us? To
such that if now, contrary to fact, he were to come to think believe this is to believe that the future would be the way it is
28 FATALISM FATALISM 29

in fact going to be, even if we should choose to behave differ- staggeringly improbable aggregation of circumstances, or else
ently and no matter how we should choose to behave. To mere superstition. If one's belief is a belief in the strong inevi-
affirm this thesis is simply to deny the reality of cause and tability of all, or all of some important class of, future events,
effect. Thus baldly stated, the thesis that all future events (or then one's belief, if it is founded upon anything at all, is
all those of some important kind) are strongly inevitable has founded upon sophistry.
nothing whatever to recommend it. It is therefore not surpris-
ing that there exist philosophical arguments for it. I offer the 2.3 Let us now turn to fatalism proper. The fatalist is not an
reader two. One is a chestnut, and one is, as the White Knight "inevitabilist", strong or weak. The strong inevitabilist affirms
would say, my own invention. It may be regarded as a general- the counter-factual conditional
ization of the first.
If Caesar had taken ship for Spain on 14 March, Caesar
It's no good summoning a physician if you are ill, for a would have been murdered on 15 March
physician can't help you. For either you're going to re- but neither the fatalist nor the weak inevitabilist affirms this
cover or you aren't. If you aren't going to recover, the proposition. (Perhaps they don't deny it either; after all, for
physician can't help you. If you are going to recover, the all anyone knows, if Caesar had suddenly decided to leave for
physician can't help you, since you don't need help.' Spain on 14 March, a storm would have forced his ship back
to port, and the assassination would have gone through as
Let E be any event that might happen. Consider the planned. But the fatalist and the weak inevitabilist see no par-
theorem of logic `(q 3 p) v (r 3 —p)'. In virtue of this ticular reason to affirm this proposition.) The weak inevitabilist
theorem, one or the other of the following two prop- affirms the counter-factual conditional
ositions must be true:
If Caesar had taken seriously the soothsayer's warning,
If I try, by any means whatever, to prevent E, E will Caesar would have been murdered on 15 March
happen;
and the fatalist does not affirm this proposition, though per-
If I try, by any means whatever, to bring about E, E will haps he doesn't deny it; he simply sees no particular reason
not happen. to think it's true. What the fatalist does believe is that, since
Therefore, either E or the non-occurrence of E is strongly Caesar didn't take ship for Spain on 14 March, he couldn't
inevitable for me. have done this. And since Caesar didn't in fact take the sooth-
sayer's warning seriously, he couldn't have. And, of course,
Each of these arguments is sheerest sophistry. (I leave them to he believes this not because he thinks he knows some special
the reader to expose.) And, so far as I know, there is no reason facts about Caesar, but as an instance of a general thesis: it is
for thinking that all events, or all events of some important a logical or conceptual truth that if an agent in fact does some
kind, are strongly inevitable always and for everyone, other particular thing, then that thing is the only thing the agent
than the reasons, such as they are, that are supplied by soph- is able to do. And, as regards the future, the fatalist believes
istical arguments like these. that if an agent is in fact going to do some particular thing,
We may summarize our conclusions about the "inevitability then that thing is the only thing he can do, the only thing it
of what is going to happen" as follows. If one's belief that what is open to him to do.
is going to happen is inevitable is a belief in the weak inevita- Now if there is any good reason to think that fatalism is
bility of all, or all of some important class of, future events, true, it is a very important thesis. It seems to be a feature of
then one's belief is a belief in an accidental, meaningless, and our concept of moral responsibility that we hold a person
SO FATALISM FATALISM 31

morally responsible for the way he has acted only if we believe arguments for fatalism depend upon premisses that are among
he could have acted otherwise. And it seems to be a feature or are variants on the premisses of the two arguments I shall
of our concept of deliberation that we can deliberate about discuss.' Therefore, what I shall say will be directly relevant
which of various mutually exclusive courses of action to pur- to, if it does not actually refute, just about any argument for
sue only if we believe that each of these courses of action is fatalism.
open to us. Therefore, anyone who accepts fatalism must I shall not examine directly the classical sources of fatalism.
regard all ascriptions of moral responsibility as incorrect, and The meaning of Aristotle's famous passage (De Interpretatione,
must, on pain of self-contradiction, refrain from deliberating IX) is in dispute; the very structure of the Master Argument
about future courses of action.' But deliberation and the of Diodorus is a matter of scholarly conjecture. I, who am no
ascription of moral responsibility are extremely important; in historian, do not propose to undertake an investigation of what
fact it is hard to imagine what human life would be like with- these philosophers may have meant. If any of the blunders I
out them. But these very facts that show that fatalism is ex- shall doubtless make in the sequel could have been avoided
tremely important if there is any reason to think it true, also by more careful attention to Aristotle, Cicero, Epictetus, or
seem to show that there could not be any reason to think it their modern commentators, I expect someone will be good
true. If fatalism is true, then the ideas expressed by sentences enough to point this out.
like 'he alone is to blame for the accident' and 'I am trying to
decide whether to have my father declared mentally incom- 2.4 Fatalistic arguments typically depend on the notions of
petent' are conceptually defective. The former is, if fatalism truth and falsity. But what are truth and falsity? Truth and
is true, a straightforward conceptual falsehood, like 'he alone falsity arc properties. But properties of what? They are pro-
has trisected the angle'. The second, while it could be used to perties of propositions. I do not mean anything mysterious by
express a truth even if fatalism were true, would be like 'I am `proposition'. I use this word as a general term for the things
trying to devise a method for trisecting the angle': these sen- people assent to, reject, find doubtful, accept for the sake of
tences could be used to express a truth only by one who argument, attempt to verify, deduce things from, and so on."
believed a conceptual falsehood. (Some of the phrases in this list take more than one sort of
I think it is incoherent to suppose that any thesis could be object. One may, for example, reject not only propositions
true that has the consequence of rendering conceptually defec- but bribes. I hope no one is going to be difficult about this.)
tive sentences so utteil■, basic to human life as the sentences We have plenty of "specialized" words for propositions in the
about blame and deliberation mentioned above. But I shall language of everyday life, just as we have plenty of special-
not argue for this thesis. It could not, I think, be adequately ized words for human beings. On various occasions we call
argued for in print: philosophers who disagree about such propositions 'doctrines', 'theses', 'theories', 'premisses', 'con-
deep matters as these can hope to resolve their disagreement, clusions', 'theorems', 'views', 'positions', 'conjectures', 'state-
if at all, only in conversation. In any case, this is not a book ments', 'hypotheses', 'dogmas', 'beliefs', and 'heresies', just as,
of metaphilosophy but of metaphysics. In such a book as this, on various occasions we call human beings 'women', 'babies',
the wisest course is to look at various arguments for fatalism `thieves', `Trotskyites', 'Australians', and 'Catholics'.
and sec what can be made of them. It will probably not sur- It is thus uncontroversial that there are propositions. The
prise the reader to discover that I think that the arguments only question that could arise is: "What are propositions?"
we shall examine are houses built upon sand. But I shall not Many philosophers apparently think propositions are sen-
claim to show that fatalism is false, since, for all I shall show, tences, since they think sentences are what are true or false.
there may be arguments for fatalism that are not open to the But I can make no sense of the suggestion that propositions
objections I shall raise. Unless I am mistaken, however, most are sentences, and I shall not discuss it further. It is true that
32 FATALISM FATALISM 35
I am willing on occasion to speak of sentences being true or matters essentially, since they would none the less have been
false, but this is only shorthand. When I say that a given sen- in disagreement about the identity of Napoleon, and therefore
tence is true, I mean that the proposition—a non-sentence- saying distinct things.) More or less the same argument can be
that that sentence expresses is true. (To say that an English given in respect of the two occasions on which I believe I am
sentence expresses a given proposition is to say, roughly, that about to die: there are possible circumstances in which what I
the result of concatenating 'the proposition that' and that sen- believe on the earlier occasion is false and what I believe on the
tence denotes that proposition.) Similarly, when I say that a later occasion is true. But there are no possible circumstances
name is honourable, I mean that the individual or family that in which some one thing is both true and false. Of course some-
bears that name is honourable. I can no more understand the one might say that what I believed on the earlier occasion was
suggestion that a sentence might be true otherwise than in then false but later became true. But if that were right, I could
virtue of its expressing a true \proposition that I can understand say on the later occasion, "When I thought I was about to die
the suggestion that a name might be honourable otherwise twenty years ago I was then wrong. But what I then believed
than in virtue of its being borne by an honourable individual has now become true. When I look at my diary of twenty years
or family. ago and see the words am about to die', I am comforted in
There are philosophers who will demand at this point that my present affliction by the thought that what I wrote has
I state a "principle of identity" for propositions. I will not do become true and that, in consequence, nothing said in my
this. When one attempts to give a general way of determining diary is now false." And this would be an absurd thing to say.
whether two predicates applying to propositions are coexten- So the two madmen said, and I believed, two different
sive, a number of vexed questions arise. What I want to say things. That is, in each of our two cases, two numerically
about fatalism can be said without answering most of them. distinct propositions are involved. To take a more extreme
But there is one sort of question about propositional identity example, if someone asks me howl feel and I say, "I am tired",
that we must be able to answer. We might call questions of and, five seconds later I am again asked how I feel and again
this sort, 'questions about propositions expressed by sentences say, "I am tired", then I assert or express two distinct propo-
containing indexical expressions'. Suppose a madman says, "I sitions: at a certain moment, I express the proposition that
am Napoleon", and that a second madman, perhaps at another I am then tired; five seconds later, I express a distinct propo-
time and in another place, speaks these same words. Do these sition, the proposition that I am then tired."
madmen assert or express the same thing?—what we might call This is a very sketchy account of "propositions", but per-
'the proposition that one is Napoleon'. Or, again, suppose that haps it will do for our purposes. Truth and falsity are, as we
at a certain moment I believe I am about to die and that at said earlier, properties of propositions. There are, of course,
another moment, years later, I believe I am about to die. Do many other properties whose extensions comprise propo-
I believe the same thing to be true on these two occasions ("the sitions. Propositions, in addition to being true or false, may
proposition that one is about to die"), or two different things? be empirically verifiable, hard to understand, inexpressible in
It seems to me that only one answer is possible: the mad- the tongue of a certain tribe, and so on. But I take it that
men say, and I believe, different things, Take the former case. each of us knows what properties truth and falsity are. Any-
Suppose, to simplify the argument, that one of the madmen one sufficiently pervicacious to claim not to know, may per-
was Napoleon. Then what one said was true and what the other haps be helped by two famous passages from Aristotle's
said was false, and, therefore, by the non-identity of discern- Metaphysics (Ross's translation):"
ibles, what one said was not what the other said. (If we had sup-
To say of what is that it is not, or of what is not that it is, is false,
posed neither of the madmen to be Napoleon, then what each
while to say of what is that it is, and of what is not that it is not, is true
said would have been false; but this would not have changed . . . (r, ionb,)
34 FATALISM FATALISM 35

It is not because we think truly that you are white that you are white, incompatible courses of action, shaving and not shaving, both
but because you are white that we who say this have the truth. (0,1051b.) are open to me, is shown to be false.
Let us now turn to an argument for fatalism. This reasoning can, of course, be generalized. If it is correct,
then for every true proposition about the future there is a
2.5 The argument we shall now consider turns on the notion of similar argument, also sound, for the conclusion that I cannot
an agent's "ability to render a proposition false". This notion (nor can anyone else) render that proposition false. That is to
will be defined in Section 3.4. For the present, however, let say, if anyone is in fact going to act in a certain way then just
us suppose that we have a sufficient intuitive grasp of the that way is the only way he can act. Or, to put our conclusion
schema 's can render p false' to go on with. another way, fatalism is true.
Let us consider some proposition about the future—say the Various philosophers have found this way of arguing com-
proposition that I shall shave tomorrow morning. This propo- pelling. Such philosophers do not usually, at least in modern
sition—call it 'S', for short—is true if and only if I shall shave times, become fatalists. But they do think that this way of
during the morning of 4 June 1976. Since I shall shave during arguing forces us to choose between fatalism and some almost
this period (I might be wrong about this; but then I might be equally unattractive alternative, such as the recognition of
wrong about almost anything), S is true. That is, S is true now. some third truth-value in addition to truth and falsity. I cannot
But it would seem that if a proposition is true at some parti- myself say whether this is the case, for the above paragraphs
cular moment then it must be true at every moment. If this contain phrases I do not understand." Among them are: 'true
were not the case, then the following would be part of a con- at some particular moment', 'true at every moment', 'became
ceivable "history" of 8: S was false all through April 1902, true', 'remained true', 'is unchangeably true', and so on. That
but early in May it became true and remained true till Good is—and we must be very careful about this—I do not see what
Friday 1936, after which it was false until V-E day. .. . But these phrases mean if they are used as they are used in the
this sort of history is not conceivable, and, hence, if S is true, above argument for fatalism. If I were to say, "Municipal bonds
S is unchangeably true. We might put the matter this way: are a good investment", and someone replied, "That used to be
there are possible worlds in which S is always true and possible true but it isn't true any more", his words would be a model
worlds in which S is always false; but there are no possible of lucidity. But if someone were to speak to me as follows :
worlds in which S is at one time true and at another false. But Consider the proposition that municipal bonds are a good
if S is unchangeably true, then I cannot render S false, for the investment. This very proposition used to be true but is
same reason that, if a certain king is unchangeably powerful, no longer true
then I cannot render him helpless. But if I cannot render S
false, then I cannot refrain from shaving tomorrow morning; then I should have grave problems in understanding what he
for if I could refrain from shaving tomorrow morning, then I meant. Let us pretend that 'T' denotes in our dialect, yours
could render S false. and mine, the moment at which my imagined respondent
Now this argument, is sound only if a certain factual claim spoke. When he spoke the words 'the proposition that muni-
I have made—that I shall shave tomorrow morning—is true." cipal bonds are a good investment' he referred to a proposition
But even if I am mistaken in thinking I shall shave tomorrow that is true if and only if municipal bonds were a good invest-
morning, then, though the argument we have been considering ment at T. (I do not say the proposition he referred to is the
is not sound, an essentially identical argument for the con- proposition that municipal bonds were a good investment at
clusion that I cannot shave tomorrow morning is sound. Thus, T; whether this is the case is one of the questions about ident-
either I cannot refrain from shaving tomorrow morning or else ifying propositions that I don't have to answer.) If he had
I cannot shave tomorrow morning: my belief that, of two spoken these same words one day earlier, he would have

36 FATALISM FATALISM 37
referred to a proposition that is true if and only if municipal I have argued that, when, in ordinary speech, we appear to
bonds were a good investment one day earlier than T. What say of a certain proposition that it used to be true, we are in
could he mean by saying of such a proposition that it "was fact saying of a certain propositional name that it used to
once" true but "is no longer" true? My understanding of the denote, or of a certain sentence that it used to express, a propo-
ordinary use of 'was once true' and 'is no longer true' is of no sition that is true, just as when we appear to say of a certain
help to me. Normally, if I utter a certain sentence and am told, number that it used to be odd, we are in fact saying of a cer-
"That was once true",i 4 my respondent means something tain descriptive phrase that it used to denote a number that is
like: 'If you had used those same words on a certain earlier odd. If I am right, then our ordinary use of 'used to be true'
occasion, then you would have said something true' or 'The and similar phrases (of which a parallel account could be given)
words you have spoken used to express a true proposition'. will not enable us to understand the use made of such phrases
If anyone thinks this anad hoc description of what 'was once in the argument for fatalism that we are considering.
true' means, let him consider the following case, which raises What I have done so far is to argue that temporal qualifi-
exactly analogous difficulties. Suppose I say, "The number of cation of a copula connecting 'true' or 'false' and a name of a
committee members is odd, so the vote won't be a tie", and proposition cannot be explained in a certain way, namely, by
you, who know of a change in the structure of the committee reference to apparent instances of it in ordinary speech. But
reply, "It used to be odd, but it isn't any more". You are not it may very well be that there is some other way to make sense
saying that there is a certain number—twelve, say—that used to of such qualification. To make sense of this idea, it would be
be odd but isn't nowadays; rather, you are saying something sufficient to make sense of the open sentence:
like this: the phrase 'the number of committee members'used
to denote an odd number, but now it denotes a different (The proposition) x is true at (the moment) t,
number, one that isn't odd. I think that most philosophers will since all the other locutions involving temporal qualification
agree that we are on the right track in the "used to be odd" of the possession of truth by a proposition that arc required
case; I think we are also on the right track in the "used to be by the fatalistic argument we are examining can be defined in
true" case. terms of this sentence.
In any event, the analogy between the two cases is instruc- I know of only one explanatory paraphrase of this open
tive. Suppose someone were to say, 'There is a number that sentence that is worthy of serious consideration." It is this:
used to be odd", and I replied, as indeed I should, that I didn't
understand what he meant. My failure to understand him If someone were to assert x and nothing else at t, then
might or might not be justified, but, however this might be, what he asserted at t would be true.
clearly 1 should not be helped towards understanding him by Thus, for example, according to the proponents of this ex-
an explanation like this one: 'Look, you understand what I planation, the sentence
mean if I say that the number of committee members used to
be odd. But the number of committee members is a number; The proposition that Queen Victoria died in 1901 was
therefore you understand what it means to say there is a num- true in 1878
ber that used to be odd'. Similarly, my difficulties about sen- expresses a truth, since, if anyone had said in 1878, "Queen
tences like 'There is a proposition that used to be true' are not Victoria will die in 1901"—which are words that were suit-
going to be cleared up by explanations like: 'The proposition able in 1878 for asserting the proposition we now call 'the
that municipal bonds are a good investment used to be true. proposition that Queen Victoria died in 1901' •--he would have
So it is clear what it means to say that there is a proposition been right. (This sort of device can be applied in all manner of
that used to be true'. cases. For example, someone could say, "The number twelve
 FATALISM 39
38 FATALISM

is even in Tibet", and explain these strange words as just then proved fatalism? Not unless we can deduce 'I cannot
another way of saying, "If someone were to refer to the num- render S false' from 'S is unchangeably true'. Why should any-
ber twelve in Tibet, he would be referring to something even") one accept this consequence? Earlier I wrote:
Now I do not think that what we have been offered is a good But if S is unchangeably true, then I cannot render S false,
explanation of the meaning of `x is true at t' since I don't for the same reason that, if a certain king is unchange-
think this sentence means anything—just as I don't think 'The ably powerful, then I cannot render him helpless (p. 34),
number twelve is even in Tibet' means anything—and thus I
don't think that anything is or could be an explanation, good and this appears to be a conclusive argument. But does the
or had, of its meaning. But perhaps this is not terribly import- analogy really hold? When we say, "Edward III was powerful
ant, since we can always regard what we have been offered as in 1346", our use of the adverbial phrase 'in 1346' is trans-
a stipulative definition: it is, after all, a sentence containing parent; it does not need to be given a meaning. But when we
the proper variables free, and that is the only formal require- say, "S was true in 1346", the phrase 'in 1346' is not trans-
ment on the definiens of a stipulative definition. That is to parent; it needs to be given a meaning. Well, we gave it a mean-
say, we may regard 'x is true at t' as simply a convenient ab- ing in the only way in which anything can be given a meaning:
breviation, without antecedent meaning, for what has mis- by stipulation. having done this, we proceeded to define 'S is
takenly been offered as an analysis or explanation of `x is true unchangeably true' in such a way that this phrase would have
at t'." We now have a way of interpreting (apparent) temporal the same logical properties as 'Edward is unchangeably power-
qualifications of the ascription of truth-values to propositions. ful', provided 'S is true at t' had the same logical properties as
Let us return to our argument for fatalism and see whether its `Edward is powerful at t'. Do they have the same logical prop-
premisses appear plausible when the temporal qualifications erties? The answer to this question is not obvious. After all,
they contain are interpreted in this way. the logical properties of, for example, 'Edward III was power-
The crucial premiss in this argument seems to be the propo- ful in 1346' are determined by the rules, embedded in our
sition that S is unchangeably true. The sentence 'S is unchange- linguistic practice, for the use of the words it contains, and
ably true' can, I think, be paraphrased in terms of our "basic" this is not the case with 'S was true in 1346': its logical prop-
locution `x is true at t' in this way (by analogy to, e.g., 'God erties arc determined by the rules for the use of the words
is unchangeably powerful'): contained in its stipulated definiens. Perhaps the wisest course
we could take would be to refrain from framing our questions
(3x) (S is true at x) & — 0 (3 x) ( 3y) (S is true at x &—S in terms of the stipulatively defined sentence 'S is unchange-
is true at y) ably true' and, instead, frank& them in terms of the definiens
where '0' abbreviates 'it is possible that'. Expanding the sec- we have provided for this sentence. That is, instead of asking
ond conjunct of this sentence in terms of our stipulative whether 'I cannot render S false' follows from 'S is unchange-
definition of `x is true at t', we obtain: ably true', let us.ask whether it follows from the conjunction of
(a) — 0 ( 3x) ( 3y) (If someone were to assert S at x, then (3 x) (if someone were to assert S at x, then what he
what he asserted at x would be true & —If someone were asserted at x would be true)
to assert S at y, then what he asserted at y would be true). with proposition (a). Informally, suppose S has these two
Now this sentence seems to express a truth: consider the con- features: (i) there is a time such that if anyone were to assert
junctive open sentence got by dropping (3 x) ( 3y)' from S at that time, his assertion would be true, and (ii) there could
it; there is no possible world in which there exists a pair of not be a pair of times such that, if someone were to assert S
moments of time that satisifies this open sentence. Have we at one of these times, he would say something true, and if
40 FATALISM FATALISM 41

someone were to assert S at the other of these times, he would be denied that this sentence expresses a truth. But if the argu-
say something false. Does it follow from this supposition that ment we are considering is to avoid the fallacy of equivocation,
I cannot render S false? Well, perhaps it does; but I don't then 'yesterday' must be used in the same sense in the argu-
see any reason to think so. Suppose someone said yesterday ment's first and second sentences. And if its first sentence is to
(Thursday) that I should shave on Saturday morning, and that make sense, then 'yesterday' must have in that sentence a sense
what he said on Thursday is true. It is, of course, quite im- of the sort we have stipulated for temporal adverbs that qualify
possible for this to be the case and for it to be the case that copulae flanked by 'true' or 'false' and propositional names,
he should say tonight, "You are going to shave tomorrow or some other purely stipulative sense. I shall ignore this sec-
morning", and be wrong. Moreover, it is simply not possible ond possibility, since I do not know what other sense might
that he should have said on Thursday "You are going to shave be given to these adverbs. Therefore, in evaluating this argu-
on Saturday" and have been right and it be the case that I ment we must interpret
shall not shave on Saturday. But, I think, nothing of interest It's not now up to you what was the case yesterday
about my free will with respect to shaving on Saturday follows;
the argument-form in accordance with our stipulation. And I see no reason to
p think that this sentence, so interpreted, expresses a truth. In
—o(p & g) particular, I see no reason to assent to
hence, —0q
If S was true yesterday, then it's not now up to me
is notoriously fallacious. But perhaps clever fatalists do not whether S was true yesterday,
rely on this fallacious form of reasoning. The clever fatalist
may claim to reason as follows: though I see every reason to assent to, for example,
Look, you admit that S was true yesterday. But it's not If S was believed yesterday, then it's not now up to me
now up to you what was the case yesterday: it is not whether S was believed yesterday.
within your power to change the fact that S was true In the second of these sentences, 'yesterday' has the sense it
yesterday. And it is a logical consequence of S's having has in everyday usage; in the first it has a peculiar sense stipu-
been true yesterday that you will shave tomorrow. There- lated by the fatalist, just as 'in Tibet' has a special sense in
fore, it is not within your power to change the fact that `The number twelve is even in Tibet' if this peculiar sentence
you will shave tomorrow. If I am making use of any is interpreted as was suggested earlier.
"modal" argument-form, it is this: If we expand the first of these sentences in accordance with
our stipulation, we obtain:
—'P p
the fact that q is a logical consequence of the fact that p If (if someone had asserted S yesterday, then he would
hence, —P q, have asserted something true), then it's not now up to
where 'P' abbreviates 'it is within one's power to change me whether it is the case that (if someone had asserted
the fact that'.' S yesterday, then he would have asserted something true).
Now I think the antecedent of this conditional is true. But I
I am not convinced by this argument. Its proponent says,
"But it's not now up to you what was the case yesterday", and see no reason to think that its consequent isn't false. That is,
I see no reason to think that the truth-value of
this has the ring of an extremely plausible assertion. If the ad-
verb 'yesterday' that occurs in 'It's not now up to you what If someone had asserted S yesterday, then he would have
was the case yesterday' has its usual sense, then it can hardly asserted something true

4 2 FATALISM FATALISM 43
isn't now up to me, for it is up to me just in the case that I can only powers one has are—of logical necessity--powers that one
shave tomorrow and can refrain from shaving tomorrow. And in fact exercises. Suppose I say that fatalism is false since I
I see no reason to doubt that both these things are within my have powers I never in fact exercise, and the fatalist replies,
power. Moreover, even if I am wrong about this and I am either "Very well then, if you have such powers, lees see you exer-
unable to shave or unable to refrain from shaving, I do not see cise one". This is not perhaps a very convincing argument. In
that the fatalist has ever suggested any logical or conceptual fact, it depends on the same fallacy as an infamous sophistical
reason why this might he so. Someone might, I suppose, offer demand sometimes made by not very able Berkeleians: 'You
the following argument: 'If S was true yesterday and if you say that objects continue to exist while they're not being ob-
have it within your power to render S false, then you have the served; very well then, let's see one'.
power to make it the case that S was at one time true and at To recapitulate: Either the use that the fatalist makes of
another time false, which is impossible'. But such a conse- "temporal" qualifications of the possession of truth and falsity
quence doesn't follow. What follows is that it is within my by a proposition is meaningless, or else this use must be ex-
power to make it the case that S was always false. When I say plained by a stipulation like the one I have suggested. If it is
this, I am talking the fatalist's language. If it sounds strange explained by such a stipulation, then sentences like 'Every
to say that I now have it within my power to make it the case proposition is, if true, unchangeably true' express, in his usage,
that a certain true proposition was always false, that is the truths. (And there is no other usage in which they express
fatalist's fault. It is he, after all, who has invented a strange anything.) But the proposition expressed by this sentence is
sense for the word 'always', a sense such that, using the word quite consistent with its being the case that there are many
in that sense, I can truly describe my ability to refrain from true propositions about, for example, what will happen
shaving tomorrow morning as an ability to make it the case tomorrow that I have it within my power to render false. Or,
that the true proposition S "was always" false. at least, I have never seen any compelling argument for the
I realize that neither the fatalist nor the philosopher who conclusion that this is not the case.
feels drawn to fatalism is likely to be convinced by these argu-
ments. Richard Taylor has responded to them in conversation 2.6 Richard Taylor has offered a puzzling and ingenious argu-
along these lines: 'You say you have the ability to render false ment for fatalism." His argument may be put as follows:
the proposition that you will shave tomorrow—that is, to
make it the case that this proposition has always been false— Suppose I am a naval commander who is deliberating about
even though this proposition is in fact true. And you say there whether to issue order 0 or to refrain from issuing order a
are ever so many true propositions about the future that you Suppose that, under the conditions that in fact prevail, my
have this sort of power over. Very well then, let us see you issuing 0 would result in there being a naval battle tomorrow,
exercise this power that you claim: pick any true proposition while my refraining from issuing 0 would result in there being
about the future, and then so act that this proposition has no naval battle tomorrow.
always been false'. But this is an illegitimate demand. I claim to We shall show that either it is not within my power to issue
have a certain power and the description I give of this power 0, or else it is not within my power to refrain from issuing 0.
depends for its application on this power's not being exercised. (i) Suppose no naval battle will occur tomorrow, Then a
(The description is 'the power to render certain always true condition necessary for my issuing 0 is absent: namely, a naval
propositions always false'.) What Taylor demands is that I battle tomorrow. For, since my issuing 0 is a condition suf-
show that I have such a power by exercising it. If this sort of ficient for there being a naval battle tomorrow, then a naval
demand were legitimate, fatalism could be established very battle tomorrow is a condition necessary for my issuing 0. If
easily. Fatalism may be looked upon as the doctrine that the someone feels it is odd to talk of my now doing something
44 FATALISM FATALISM 45

"in the absence of", for example, a naval battle tomorrow, we is open to him; that at most one is such that he can pursue it;
may agree with him. Let us say that our use of 'in the absence that it is not up to him which if any of these courses he shall
of is an extension of normal English usage. We may in our pursue.
present usage say I am now writing in the absence of an earth-
quake in London tomorrow, provided there will be no earth- The crucial premiss in Taylor's argument is Principle (A). If
quake in London tomorrow. As regards the past, we may say we accept this principle, then we can derive fatalism by a much
I am now writing in the absence of a German invasion of the simpler argument than Taylor's. For it is obvious that when-
British Isles in 1940. While there is an element of artificiality ever I am not performing a certain act, then there is absent a
in this way of speaking, the truth-conditions for assertions condition necessary for my performing it: namely, my per-
that involve it are none the less intelligible enough. The first of forming it." And, of course, whenever I am performing a
our illustrative sentences expresses just the same proposition certain act, there is absent a certain condition necessary for
as, or a proposition necessarily equivalent to, that expressed by my not performing it: my not performing it." Thus, Principle
(A) leads directly to the collapse of the distinction between
I am now writing and there will be no earthquake in what one does and what one can do: one who accepts Prin-
London tomorrow. ciple (A) has already got fatalism in his pocket and need not
The second illustrative sentence expresses the same proposition shop for it in Taylor's elaborate naval bazaar, diverting though
as, or a proposition necessarily equivalent to, that expressed by the wares offered there may be.
Is Principle (A) true? The question is complex, for Prin-
I am now writing and the Germans did not invade the ciple (A) is ambiguous. We shall see that there are two ways
British Isles in 1940. it might be interpreted. Interpreted in one of these ways, it
Now consider the following principle: is obviously true but does not yield Taylor's conclusion; in-
terpreted in the other way, it indeed yields Taylor's conclusion,
(A) No agent is able to perform an act in the absence of a though there is no reason to think it is true.
condition necessary for its accomplishment. Let us call an ability-sentence a complete, grammatical,
Taylor says of this principle, "This is no law of logic, and in declarative sentence that consists of a subject term that denotes
fact cannot be expressed even in the contemporary modal or purports to denote a person (`I'; 'Richard Taylor'), followed
logics, but it is nonetheless manifestly true" (p. 58). It follows by an "ability-copula" such as 'can', 'am able to', or 'has it
from (A) and from what we have already established—that within his power to', followed by any string of words.
a condition necessary for my issuing 0 is absent—that I cannot Ability-sentences containing adverbs or adverbial phrases
issue 0. are often ambiguous. Consider:
(ii) Suppose a naval battle will occur tomorrow. Then a I can refrain from talking at any time.
condition necessary for my refraining from issuing 0 is absent:
namely, the non-occurrence of a naval battle tomorrow. But Does the adverbial phrase 'at any time' modify the verb-phrase
then, by principle (A), I cannot refrain from issuing 0. 'refrain from talking' or does it modify 'can'? In the former
Therefore, either I cannot issue 0 or else I cannot refrain case, the sentence means something like 'It is within my power
from issuing 0. That is to say, it is not up to me whether I to keep a vow of perpetual silence'; in the latter case, some-
issue 0. This argument, of course, can be generalized. If it is thing like 'At any given moment, I can at that moment refrain
sound, then, for any case of an agent who is deliberating about from talking'.
which of various courses of action to pursue, we can show by In normal English usage, this ambiguity is usually not very
a similar argument that at most one of these courses of action important. It is resolved by such factors as intonation or one
46 FATALISM FATALISM 47

of the syntactically possible interpretations being wildly in- But this is not quite satisfactory, since the propositions ex-
appropriate. If there were some real possibility of confusion, pressed by these sentences seem to have unwanted entailments,
the speaker might reposition the adverbial phrase: as, for example, that I can understand English. Let us, there-
fore, regard our first explanation of our use of brackets as
I can, at any time, refrain from talking. official, and the second and third explanations as being offered
It will be convenient for us to introduce a more explicit device for the sake of such intuitive value as they may have.
than these for the disambiguation of ability-sentences. Let us Let us now return to principle (A). Let us look at some in-
use round brackets in this fashion: stances of this principle. First a "trivial" instance:
I can (refrain from talking) at any time; (a) I cannot move my finger when my finger is not moving.
I can (refrain from talking at any time). This sentence may be disambiguated as follows:
The import of the round brackets is this: in the first of these (b) I cannot (move my finger) when my finger is nonmoving;
sentences, 'at any time' must modify 'can', since both `can' and
`at anytime' are outside the brackets; in the second sentence, 'at (c) I cannot (move my finger when my finger is not moving).
any time' must modify `refrain from talking', since both these If an English speaker were to utter sentence (a), he would
phrases are inside the brackets. Roughly speaking, the presence most naturally be taken to mean (c). But (b) is apossible read-
of a pair of disambiguating brackets in an ability-sentence ing of (a), though what is expressed by (b) would normally
"forces" an adverb or adverbial phrase to modify a verb or be expressed by 'Whenever my finger is not moving, I lack
verb-phrase that is on the "same side" of the pair of brackets. the ability to move my finger'. Obviously, if someone uttered
Or, if the reader is willing to include in his ontology ab- sentence (c), or uttered (a) meaning (c), he would be right.
stract entities called "acts" and is willing to recognize 'refrain- But what about (b)? Could a person utter this sentence and
ing from talking' and 'refraining from talking at any time' as say something true? Well, consider the following case. My
phrases naming acts, then he may regard the round brackets finger is paralysed, but it is important to me that it move, so
as marking the boundaries of names of acts. Thus, on this way I am continuously straining to move it. During intermittent
of looking at matters, the above pair of sentences may be read: and unpredictable intervals I become able to move my finger
I can perform the act refraining from talking at any time; and then, of course, it begins to move; then it again becomes
paralysed and ceases to move. In that case, if I were to utter
I can perform the act refraining from talking at any time. (b), I should say something true. There is therefore a semantic
Perhaps even the reader whose ontology is not so copious as difference between (b) and (c); sentence (a) is not merely
to include such acts will regard this explanation as intuitively syntactically ambiguous. But if (a) is an instance of (A), then
useful if ontologically dubious, like an explanation of the it would seem to follow that (A) is ambiguous. Is it (b) or (c)
derivative as a quotient of infinitesimal quantities. that is an instance of (A)?
A third way of looking at this pair of sentences is to regard A similar point can be made about
them as marking the same distinction as: (d) I cannot issue 0 in the absence of a naval battle tomorrow.
I can obey the command 'Refrain from talking' at any
time; Does Taylor intend this sentence to mean (e) or (f)?
I can obey the command `Refrain from talking at any (e) I cannot (issue 0) in the absence of a naval battle
time'. tomorrow.
 FATALISM 49
48 FATALISM

I cannot (issue 0 in the absence of a naval battle is entailed by, the proposition that I cannot issue 0 without a
(f)
tomorrow). naval battle tomorrow resulting. And the following inference
hardly seems valid:
Which of these sentences is an instance of (A)? Obviously this
question has no answer. For (A) itself is ambiguous. We must I cannot issue 0 without a naval battle tomorrow re-
distinguish between: sulting;
(A)1 No agent is able to (perform an act) in the absence of a There will be no naval battle tomorrow;
condition necessary for its accomplishment, hence, I cannot issue 0.
and Of course, if fatalism is true, then this inference is valid, as is
(A)2 No agent is able to (perform an act in the absence of a the inference about the movement of my finger we looked at
condition necessary for its accomplishment). earlier. But I am not arguing that fatalism is false; rather that
a certain argument does not show that it is true.
In which of these ways shall we read (A)? Let us try it both It seems, therefore, that, while A(2) is true, there is no
ways. Suppose (A) means A(2). From A(2) we may deduce (f), reason to think that it can be used to prove fatalism. What
though there seems to be no reason to think we can deduce about A(1)? A(1) does indeed entail (e), and hence fatalism.
(e). Does it follow from (f) together with the proposition that &fit is there any reason to think that A(1) is true? I see none.
there will be no naval battle tomorrow that I cannot issue 0? The only argument Taylor" offers in support of (A) does not
I see no reason to think it follows. Certainly not just any in- produce—in me, at any rate—any tendency to accept A(1). His
ference of this form can convincingly claim to be valid. For argument consists simply in exhibiting instances of situations
example, the inference in which he is unable to perform some act and in which a con-
I cannot (move my finger in the absence of my finger's dition necessary for his performing that act is absent and in
moving); which his inability to perform that act is a result of the absence
My finger is not moving; of that condition:
hence, I cannot move my finger I cannot, for example, live without oxygen, or swim five
miles without ever having been in water, or read a given
does not seem to be valid. In fact, it seems to be invalid. For page of print without having learned Russian, or win a
its premisses are true, but its conclusion certainly seems to be certain election without having been nominated, and so
false. As one philosopher (an unlooked-for ally) has put the on. (p. 58.)
matter:
But this argument is invalid. Since my being able to perform
The statement "I can move my finger," as well as the statement "I can a certain act is a necessary condition for my performing it, any
hold my finger still," are both true (though their joint truth obviously
does not entail that I can do both at once). This I take to be quite cer- condition necessary for my being able to perform that act is a
tain . . . if there is any philosophical theory implying that one or the condition necessary for my performing it. Therefore, since
other of these statements must be false, then that theory is doubtful. 2 ' there are indeed conditions necessary for my being able to per-
form various acts, there are conditions that are necessary for
Is there then some special reason, some reason having to do my performing various acts that are also necessary for my
with naval battles and orders, to suppose that the proposition being able to perform those acts. What Taylor has done is
that I cannot issue 0 follows from (f)? This would be hard to simply to list certain conditions necessary for his being able
maintain. Proposition (f) is the same proposition as, or, at least, to perform certain acts, conditions that are a fortiori necessary
50 FATALISM FATALISM 51

for his performing those acts. But such a list lends no support Candidate (i) must be provided with some informal qualifi-
to A(1). One might as well try to show that one cannot live cation if it is even to be plausible. Obviously 'sentence' must
in America if some condition necessary for one's living in be understood to mean 'declarative sentence', where, one
California is absent by arguing: hopes, the class of "declarative" sentences of English is one
that can be specified by purely syntactical means. But even if
I cannot, for example, live in America, if I do not live this can be done, a more serious problem remains: (i) seems to
in the Western Hemisphere, live upon dry land, live north be a report of a fact about English that is at best contingently
of the Equator ... true. And this would be the case even if 'true' in (i) were re-
I conclude that Taylor's attempt to establish fatalism is a fail- placed with 'necessarily true'. After all, 'or' and 'not' might
ure. Principle A is ambiguous. On one interpretation, A(2), it have meant something other than what they in fact mean. Of
is true but there is no reason to think it entails fatalism; on course, (i) might be elaborated to meet this difficulty, but the
the other, A(1), it entails fatalism but there is no reason to objection might be elaborated, too. On the whole, I think it
think it true. would be more profitable for us to turn our attention to (ii)
and (iii), than to play modification-and-counter-example
2.7 Discussions of the "Law of the Excluded Middle" bulk with (i).
large in most treatments of fatalism. But I have not discussed There are two main differences between (ii) and (iii): (ii)
this "law" at all, and it does not appear as a premiss in either contains `false' and (iii) does not, and (iii) contains 'denial'
of the fatalistic arguments we have examined. It would be and (ii) does not. But what does 'false' mean if not 'not true';
possible to maintain, however, that these two arguments do what is falsehood but the complement of truth in the domain
make covert use of the Law of the Excluded Middle. My for- of propositions? And what is the "denial" of a proposition if
mulation of Taylor's argument, for example, depends for its not the proposition that it is not true? Some philosophers, I
validity on the validity of the inference-form know, think that 'true' and 'false' are like 'transparent' and
`opaque': just as there are visible objects that are neither trans-
p q parent nor opaque, so there are propositions that are neither
—p. J q true nor false. It has been argued, for example, that the propo-
hence, q sition that the present king of France is bald is neither true
which certainly seems to depend, in some sense, on the Law nor false. These well-known arguments, however, establish
of the Excluded Middle. And perhaps it is arguable that the at best that it is not clear whether 'the proposition that the
argument discussed in 2.5 depends, again "in some sense", on present king of France is bald' denotes anything, given the
the Law of the Excluded Middle. political conditions that actually prevail in France. But if this
But what is the Law of the Excluded Middle (hereinafter description does denote something—presumably it denotes the
TEMP)? Here are three candidates: proposition that someone is now the only king of France and
is bald, if it denotes anything—what it denotes is, I should
(i) If p is any English sentence, then the sentence that re- think, either true or false; if it denotes nothing, then no
sults from writing p and then writing 'or it is not the
case that' and then once more writing p, expresses (as an counter-example to the thesis that falsehood is, in the domain
of propositions, the complement of truth, has been produced.
English sentence) a true proposition;22
(The thesis that propositions "about the future" are neither
(ii) Every proposition is either true or false; true nor false we shall consider presently.)
(iii) For every proposition, either that proposition is true or If, as I have argued, 'false' means 'not true', and the denial
its denial is true. of a proposition is the proposition that it is not true, then (ii)
52 FATALISM FATALISM 53

and (iii) come to much the same thing. I shall treat them as By way of commentary on this argument: (a) it is formally
equivalent and equally good expressions of LEM. valid, being an instance of the form `p q;ra- s; q vs;hence
Is LEM true? Various reasons for saying it is not have been p V r'; 23 (b) the sentence displayed to the right of '(1)' is formed
offered. One class of interesting reasons for rejecting LEM by filling the blank in 'The proposition that . . . is true if and
arises in physical science, from considerations involving the only if . . Such a sentence expresses—as a sentence of English
physical interpretation of the formal mathematical theory of and relative to a given context of utterance (I shall from now
quantum mechanics; another class of reasons arises in pure on leave it to the reader to supply such qualification as this)—
mathematics from considerations involving non-constructive a true proposition provided the sentence that fills the blank
descriptions of infinite sets. I have nothing interesting to say expresses any proposition. (I shall presently discuss the conten-
about these reasons, which are, in any case, of no direct rel- tion that 'I shall kill myself' does not express any proposition.)
evance to those questions about LEM that arise in discussions If anyone denies this, I do not understand what he means by
of fatalism. (They may, of course, be of indirect relevance; if `true'. The same remarks apply mutatis mutandis to the sen-
they are persuasive, they may lead the erstwhile champion of tence to the right of '(2)'. (c) Therefore—leaving aside the
LEM to mistrust his intuitions concerning the application of question whether 'I shall kill myself' and its negation express
this principle even in cases involving only macroscopic objects propositions-.(4) is true if (3) is true.
and constructive predicates.) Is (3) true? It is certainly the case that if I were to utter
What is of direct relevance to the problem of fatalism is the the sentence displayed to the right of `(3)' above, I should
question whether LEM "applies" to propositions "about the say something true. (Or at least this is the case if we suppose
future". Does this principle give us licence to say that, for I utter this sentence in a context in which my uttering it con-
example, the proposition that I shall one day kill myself (K) stitutes my asserting that I either shall or shall not kill myself,
is either true or false? Let us say that it does, and then ask and not a mere phonetic exercise. But such a context is easily
whether the principle, so interpreted, is true. It will, I think, imagined: suppose you and I are discussing the advisability of
be sufficient to ask whether K is either true or false. If K is my buying a certain type of life insurance, and I utter this
neither true nor false, then LEM is false; if K is either true or sentence as a preface to an argument by cases.) But if that is
false then—I should think—just any proposition "about the the case, then how could (3) be anything but true? Could it
future" is either true or false, at least assuming it involves only be the case that there is a certain sentence s such that (i) if I
assertions about macroscopic objects and properties express- were to utter s I should in uttering s say something true, and
ible by means of constructive predicates. (ii) the proposition expressed by s is not true? To answer
The question whether K is either true or false. is just the "Yes" to this question is surely to contradict oneself.
question whether it is the case that either I shall kill myself Therefore, (4) is true unless 'I shall kill myself' fails to ex-
or I shall not. For consider the argument: press a proposition, unless there is no such proposition as the
proposition that I. shall kill myself, just as there is no such
(1) The proposition that I shall kill myself is true if and only proposition as the proposition about kangaroos I just now
if I shall kill myself asserted. How could there be no such proposition? Surely
(2) The proposition that I shall kill myself is false if and only someone may believe or assert or even know that I shall one
if 1 shall not kill myself day kill myself. But if a sentence can be meaningfully con-
catenated with 'he believes that . . .' or 'he knows that . .
(3 Either I shall kill myself or I shall not
)
or can be used as the vehicle of an assertion, then that sentence
hence, (4) Either the proposition that I shall kill myself is expresses a proposition. That is what it is for a sentence to
true or it is false. express a proposition.
54 FATALISM

Perhaps the opponent of LEM will want to protest: "I grant

you that it is possible for someone to utter, say, 'he will some-
day kill himself' and make an assertion. What I am saying is Chapter III
that the assertion he makes, or the proposition he expresses,
is neither true nor false". But to say this is simply to grant the Three Arguments for Incompatibilism
premiss of my argument—that it is possible to make assertions
about the future—and to deny the conclusion: that any such
assertion is either true or false. And that would be to deny 3.1 The main contested question in current discussions of
that my argument is valid without attempting to find a flaw free will is not, as one might expect, whether we have free will.
in it. It is whether free will is compatible with determinism. It seems
There is a good deal more that could be said about LEM, to me that free will and determinism are incompatible, and in
but I shall not say it. I have attempted in 2.7 to show that this chapter I shall try to demonstrate this incompatibility.
LEM is true because I think it is true and because certain Discussions of this question are usually not on a very high
writers on fatalism have denied it, thinking that such a denial level. In the great majority of cases, they are the work of corn-
constitutes the only escape from fatalism. But since there are patibilists and consist to a large degree in the ascription of
no known compelling arguments for fatalism, this is not the some childish fallacy or other to incompatibilists (conflation
case. Such writers are like atheists who become Parmenideans of "descriptive" and "prescriptive" laws; failure to distinguish
in order to deny the premiss of St. Thomas's First Way. To between causal necessity and compulsion; equation of free-
enter into an extended debate about LEM would be as profit- dom and mere randomness). Donald Davidson places himself
less an undertaking as writing a treatise proving that some in this tradition when he writes:
things are in motion. I shall not be directly concerned with [arguments for the incompatibility
of freedom and causal determination], since I know of none that is more
than superficially plausible. Hobbes, Locke, Hume, Moore, Schlick, Ayer,
Stevenson, and a host of others have done what can be done, or ought
ever to have been needed, to remove the confusions that can make deter-
minism seem to oppose freedom.'

It is not my purpose in this book to defend any previous writer

against a charge of fallacious argument. My own arguments
will be explicit, and any fallacies they commit should be cor-
respondingly visible. (It is doubtful whether anyone has ever
been seduced by the fallacies with which incompatibilists are
customarily charged; if anyone indeed has achieved such a level
of philosophical incompetence, I, at least, fall short of it.) Now
the line between arguing for a certain view and ascribing fal-
lacious reasoning to the opponents of that view, while clear
enough in theory, is often blurred in practice. Entangled with
the charges of fallacy and confusion made in the writings of
the philosophers Davidson mentions, there are positive argu-
ments for the compatibility of free will and determinism. In
so far as these arguments can be disentangled from diagnoses
56 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 57

of ills I am not heir to—if, indeed, these ills exist at all—I shall objection to one of the arguments, then I do not propose to
examine them in Chapter IV.' hide behind the others. In fact, I think this would be imposs-
ible: I am quite sure that any specific and detailed objection
3.2 In Chapter I, I mentioned the following simple argument to one of the arguments can be fairly easily translated into
for incompatibilism, the Consequence Argument: specific and detailed objections to the others; and I think that
any objection to one of the arguments will be a good objection
If determinism is true, then our acts are the consequences
of the laws of nature and events in the remote past. But to that argument if and only if the corresponding objections
to the others are good objections to them. I may be wrong
it is not up to us what went on before we were born, and
about this. Perhaps one of the arguments fails in some way to
neither is it up to us what the laws of nature are. There-
which no defect in either of the others corresponds; indeed,
fore, the consequences of these things (including our
perhaps all the arguments are defective, but for, loosely speak-
present acts) are not up to us.
ing, different reasons. I hardly think this is the case, however.
This, I think, is a good argument. But I must admit it's rather We shall see.
sketchy. The present chapter is an attempt to fill in the details There is a second dialectical advantage in my presenting
of this sketch in three different ways. (I have called this chap- three arguments, provided I am right in thinking that they
ter "Three Arguments for Incompatibilism", but the principle stand or fall together. If I relied on only one of these argu-
of individuation for arguments is far from clear; I might have ments, a compatibilist might reason as follows: "Since the
called it "One Argument for Incompatibilism Done Three conclusion of this argument is fundamentally wrong, the argu-
Ways".) These three arguments, or versions of one argument, ment contains a fundamental mistake. I admit I can't put my
or whatever they are, are intended to support one another. finger on this mistake, though I'm sure I could uncover it if I
Though they have essentially the same point, they are very were willing to take the time and trouble. But life is short ..."
different in structure and vocabulary. I intend to exploit these Now I am not at all unsympathetic to such an attitude, at least
differences as follows. If anyone has a vague feeling of unease provided that the philosopher who expresses it does not claim
about one of the arguments—because it's not stated in anything to be an expert on the problem of free will and determinism.
like ordinary language, say, or because its logical structure is Life is short. But I do think that this attitude is a less reason-
rather complicated and may serve to conceal some muddle— able one to express in the face of three arguments that seem
he may find, I hope he will find, that this feeling of unease intuitively to have the same point and which are yet very dif-
does not arise when he contemplates one or both of the other ferent in the vocabulary they employ and in their logical
two arguments. This, I think, will tend to show that the cause structure. Surely if there were some fundamental mistake
of his feeling of unease is an accident of a particular mode of common to all three arguments, it would be at least likely to
presentation of the features common to the three arguments reveal itself in one of them, however well hidden it was in the
and does not reside in these common features themselves. Of others?
course, it does not follow logically that this is the case: the Each of the three arguments has its peculiar virtues and vices.
cause of the feeling of unease that someone feels when he con- The virtue of the first argument is its vocabulary, which is close
siders one of these arguments may be present but inoperative to that of traditional discussions of the free-will problem. Its
when he considers another, owing to the fact that this cause, vice is its extremely complex structure. The virtue of the sec-
which may or may not be a real defect in the arguments, is ond argument is its extremely simple structure. Its vice is its
masked by features peculiar to the other. vocabulary, which is radically different from the traditional
I intend these remarks to apply only to cases of "a vague vocabulary of arguments about free will. The third argument is
feeling of unease". If a philosopher has a specific and detailed an intermediate case as regards complexity oflogical structure
58 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 59

and departure from traditional terminology. Its peculiar logical A proposition in this sense is true if it contains the actual
vice is a dialectical virtue: while the first two arguments are world; the denial of a proposition is its complement on the
valid in first-order extensional logic, the third employs special set of all possible worlds; the conjunction of two propositions
modal principles. This is a logical vice because I endorse no is their intersection; one proposition entails a second if the
extensional semantics for the modal operator involved in the former is a subset of the latter. I am not wedded to this model.
argument, and, therefore, my contention that the modal prin- It has features that unfit the "propositions" it models for cer-
ciples I employ in the argument are valid rests only on intuition. tain tasks traditionally assigned to things called 'propositions'.
But the use of a modal operator is a dialectical virtue, since For example, in Section 2.4, "propositions" were introduced
many people seem to think that a modal fallacy in some way as things that could be accepted, asserted, proved, denied, and
underlies incompatibilism, and this charge is best evaluated in so on. But the objects of such activities can hardly be modelled
its application to an explicitly modal argument. on sets of possible worlds, for on that model, for example,
the proposition that some Albanian barber shaves all those
3.3 In this section, and in the following section, I shall explain and only those Albanians who do not shave themselves is
what I mean by certain terms that will be used in the first identical with the proposition that there are Greek geometers
argument for incompatibilism. who know how to trisect the angle: each is the empty set. Yet
I shall begin by attempting to say what I mean by deter- if these propositions are objects of assertion or denial, surely
minism. In order to define determinism I shall need three they must be distinct propositions. Still, this model yields all
subordinate notions: the notion of a proposition (and allied the features of propositions that I need for my present pur-
notions such as truth, denial, conjunction, and entailment) poses, and shows, assuming the consistency of the notions of
and the notion of the state of the entire physical world at an set and possible world, that these features are consistent.
instant, and the notion of a law of nature. Let us now turn to the state of the entire physical world
Propositions (that is, non-linguistic bearers of truth-value) (hereinafter, the state of the world) at an instant. I shall leave
were introduced in Section 2.4. I have little to add to what I this notion largely unexplained, since my argument is very
said there and in 2.7 about propositions, truth, and falsity. I nearly independent of its content. Provided the following two
shall assume only that propositions have the following four conditions are met, one may flesh out 'the state of the world'
properties. (i) To every possible way the world could be, there in any way one likes.
corresponds at least one proposition, a proposition that is (i) Our concept of state must be such that, given that the
necessarily such that it is true if and only if the world is that world is in a certain state at a certain instant, nothing follows
way. Since there are doubtless ways things could be that are about its state at any other instant: if x and y are any "states",
too complex to be described in any natural language, this and some possible world is in x at t i and y at t 2 , there is a
means that there are propositions that cannot be expressed in world that is in x at t i and not in y at t 2 . For example, we must
any natural language. (ii) Every proposition is either true or not choose a concept of state that would allow as a part of a
false. (iii) The conjunction of a true and a false proposition is description of the momentary state of the world, the clause,
a false proposition. (iv) Propositions obey the Law of Contra- `. . . and, at T, the world is such that someone's left hand will
position with respect to entailment. That is, for every x and be raised 10 seconds later than T'.
every y, if x and y are propositions, and if it is impossible for The theory of relativity has the consequence that this notion
x to be true and y false, then it is impossible for the denial of —the state of the world at an instant—has no application to
y to be true and the denial of x false. things as they are, and is perhaps even incoherent. But there
One "model" for propositions that gives them these features is a relativistically acceptable concept—the state of things on
is this: propositions are identified with sets of possible worlds.' the surface of a light-cone—that could be used in its place. This
60 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 61

refinement, however, could be implemented only by tedious been any human beings or other rational animals. Law of
and philosophically irrelevant (in the present context) elabor- nature, at least in my usage, is no more an epistemological
ation of definitions and arguments that are already rather more term than is star. (Ontologically speaking, a star is a material
elaborate than I like. I shall therefore ignore relativistic con- body: some material bodies have the feature being a star and
siderations in the sequel. The second and third arguments will some don't, and which do and which don't is a matter utterly
be similarly unsophisticated. independent of the present state of astronomical knowledge
(ii) If there is some observable change in the way things are and of the history of astronomy. The stars would be just as
—if a white cloth becomes blue, a warm liquid cold, or if a they are even if there had never been any human beings or
man raises his hand—this change must entail some change in other rational animals.)
the state of the world. That is, our concept of state must not Despite the parallels I want to draw between the stars and
be so theoretical, so divorced from what is observably true, the laws of nature, I want also to insist on an important dif-
that, for example, it be possible for the world to be in the ference between our powers with respect to these two classes
same state at t 1 and t 2 , despite the fact that someone's hand is of objects. It is quite conceivable that human power will grow
raised at t i and not at t2 . It is arguable that this requirement is in- to the extent that we shall one day be able to alter the stars
compatible with our interpreting state as quantum-mechanical in their courses. But we shall never be able to do anything
state, for it is arguable that there exists a moment t and poss- about the laws of nature. There are presumably many propo-
ible worlds w 1 and w 2 such that w i and w 2 are in the same sitions that are in fact true, but which it is within our power
quantum-mechanical state at t, although a certain cat is, in to falsify. Probably the proposition that no one ever has read
w 1 , alive at t and, in w 2 , dead at t. 4 But I shall say nothing or ever will read all of the Oxford English Dictionary aloud is
about the deep and difficult problems this contention raises. true. I should think, however, that someone could falsify this
Having defined—or, at any rate, discussed—the notions of proposition if he were willing to devote a large part of his life
proposition and state of the world at an instant, we shall now to this pointless task. But suppose someone has set out to fals-
combine them. Let us say that a proposition expresses the state ify the Principle of the Conservation of Angular Momentum.
of the world at t provided it is a true proposition that asserts That is, suppose someone is attempting to construct a piece
of some state that, at t, the world is in that state. (We may put of laboratory apparatus the behaviour of which would violate
the matter this way in terms of our "sets of possible worlds" this principle. If the principle is a law of nature, he cannot
model of propositions: a proposition p purports to express the succeed. If he can succeed (even if he doesn't), that is, if he has
state of the world at t if there is some state such that p con- it within his power to succeed, then it is not a law. We might
tains all those and only those worlds that are in that state at even imagine that the principle is in fact true but is not a law
t; p expresses the state of the world at t if it purports to ex- simply because someone can falsify it. Suppose that a certain
press the state of the world at t and is, moreover, true.) physicist designs a certain piece of apparatus and that all com-
Finally, I need the notion of a law of nature. I have no idea petent physicists agree that if this piece of apparatus were built
how to explain this term, much less define it. But I can say and put into operation it would violate the Principle of the
what I do not mean by it: as I said in Section 1.3, it is not an Conservation of Angular Momentum. Suppose a respected firm
epistemological term. Ontologically speaking, a law of nature of engineering contractors, having examined the physicist's
is a proposition: some propositions have the feature being a specifications, state that it would be possible, given "the state
law of nature and some don't, and which do and which don't of the art", to build the device. Suppose the physicists and
is a matter utterly independent of the present state of scientific engineers are right. But suppose that actually to construct the
knowledge and the history of scientific knowledge. The laws device would require an enormous expenditure of resources,
of nature would be just as they are even if there had never a fact that results in its never being constructed. We may
62 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 63

consistently add to these suppositions the supposition that is one. No law of nature, however, is such that anyone can
angular momentum is conserved, that is, that the Principle of render it false: the schema, 'If P is a law of nature, then no
the Conservation of Angular Momentum is true. But this prin- one can render P false' has no counter-instances. Or put the
ciple would simply not be a law of nature if such a device matter this way: it is true and trivial, and not what I am argu-
could be constructed: if human beings can (have it within ing for, to say, "It is impossible for there to be a person x and
their power to) conduct an experiment or construct a device a proposition y such that x can bring it about that y is a law
that would falsify a certain proposition, then that proposition of nature and y is false". Call this 'the de dicto principle'. It
is not a law of- nature. A law of nature must be immune to is also true to say, "It is necessary that, for every person x and
such possible disc onfirmation. This of course is consistent with every proposition y, if y is a law of nature, then x cannot
our saying that for any given law we could conceive of an render y false". This is at least less trivial and it is what I have
experiment that would disconfirm it. been arguing for. Call this 'the de re principle'. From the de
The conclusion of the above argument may be restated this re principle and 'the proposition that momentum is conserved
way: the laws of nature impose limits on our abilities: they are is a law of nature', we may validly deduce, `Feynman cannot
partly determinative of what it is possible for us to do. And render the proposition that momentum is conserved false'.
indeed this conclusion is hardly more than a tautology. The But from this premiss and the de dicto principle, no such con-
oddness of denying it can be brought out if we think of some- clusion follows. Note that such applications of modus ponens
one ordering a subordinate to violate a law of nature. Suppose to instances of the de re principle may yield contingent truths,
a bureaucrat of the future orders an engineer to build a space- since the proposition that a given proposition is a law of nature
ship capable of travelling faster than light. The engineer tells may very well be a contingent truth. Feynman, for example,
the bureaucrat that it's a law of nature that nothing travels may be unable to render the proposition that momentum is
faster than light. The bureaucrat concedes this difficulty, but conserved false because this proposition is a law of nature.
counsels perseverance: "I'm sure", he says, "that if you work But, for all I know, there are possible worlds in which he exists
hard and are very clever, you'll find some way to go faster and in which this proposition (though perhaps true) is not a
than light, even though it's a law of nature that nothing does." law of nature and in which he is able to render it false.
Clearly his demand is simply incoherent. If we interpret law of nature' very broadly, there seem to
I should point out that the above conclusion does not rest be exceptions to the de re principle, and these must be dealt
on the premiss that the laws of nature are true propositions. I with. Consider psychological laws, including laws, if such there
am not saying that it is impossible for us to alter the laws of be, about the voluntary behaviour of rational agents. If there
nature owing simply to the fact that it is logically impossible are such laws, it is at least arguable that they should be in-
to cause any proposition simultaneously to satisfy the con- cluded among the "laws of nature"; rational agents are, after
ditions 'x is a law of nature' and `x is false' (laws of nature all, in some sense part of "nature". But it is hard to see how
being by definition true). 5 If this argument were valid, one to avoid the conclusion that, if we have free will, we have it
could derive fatalism from the premiss that it is logically im- within our power to act differently from the way such laws
possible to cause any proposition simultaneously to satisfy say we shall act. Let us look at a particular case.
the conditions 'x is a true proposition' and 'x is false'. But, as Suppose psychologists discover that no one who has received
I said in Chapter II, many true propositions are such that one moral training of type A in early childhood ever spreads lying
can render them false: the schema 'If P is a true proposition, rumours about his professional colleagues. Suppose you and
then no one can render P false' has many counter-instances, I in fact received such training. Does it follow that we can't
at least in my opinion. The instance got by replacing 'P' with engage in this odious activity? I don't see why it should be
`the proposition that no one ever reads all of the OED aloud' supposed to follow. (Mark Twain: "I am morally superior to
64 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 65

George Washington. He couldn't tell a lie. I can and I don't.") Thus, on our "sets of possible worlds" model for propositions,
Suppose further that you and I are in fact able to spread lying a law of nature is any set of worlds that has as a subset the
rumours about our colleagues. Does it follow that a statement set of all worlds in which the laws of nature are the same as
of the regularity we have imagined Psychologists to have dis- those of the actual world, or, as we might say, are nomologi-
covered is, though true, not a law? Well, suppose the exist- cally congruent with the actual world.
ence of this regularity is a logical consequence of some well- We may now define 'determinism'. We shall apply this term
confirmed theory of human moral development that has great to the conjunction of these two theses:
explanatory and predictive power. In that case, it would cer- For every instant of time, there is a proposition that ex-
tainly be very tempting to call this statement a `law'; I should presses the state of the world at that instant; 6
hardly want to counsel resistance to this temptation. "But
If p and q are any propositions that express the state of
why", someone may ask, "does this regular pattern of be-
haviour occur if people don't have to conform to it?" Note the world at some instants, then the conjunction of p
that the only people in a position to depart from it are those with the laws of nature entails q.
who have in fact had training of type A. Perhaps it is just these This definition seems to me to capture at least one thesis
people who see the point in not spreading lying rumours. To that could properly be called 'determinism'. Determinism is,
come to see the point in not exercising an ability one has is intuitively, the thesis that, given the past and the laws of nature,
not to lose that ability. there is only one possible future. And this definition certainly
So it seems at least plausible to suppose that the de re prin- has that consequence. It also has the consequence that the
ciple might be false if law of nature' were interpreted broadly future determines a unique past. This consequence, however,
enough. I shall simply narrow the interpretation of law of does not trouble me, The only physical theories that are known
nature' by fiat: "laws of nature" in the sequel shall be by to be deterministic "from-past-to-future" (two-particle classi-
definition propositions that apply non-vacuously to things cal mechanics and certain mathematically similar theories) are
that are not rational agents. (Things such as teacups, electrons, also known to be deterministic "from-future-to-past". There
and galaxies.) For such laws, I maintain, the de re principle are "theories" in a certain broad sense of the word that are
holds good. This stipulation has an important consequence. It only "one-way" deterministic—"theories" describing the be-
may well be that, for all that is said in this book, human be- haviour of certain Turing machines, for example—but I should
haviour is wholly predictable on the basis of laws that are be at least mildly surprised to see any plausible physical theory
about the voluntary behaviour of rational agents. Moreover, I that had that feature. But if anyone is really troubled about
see no reason to think that such predictability would be in- this, he may add a suitable "later than" clause to the defi-
compatible with free will. nition. Such an addition will not affect the use I shall make
Let us make one further stipulation about the laws of nature: of this definition in what follows.
the logical consequences of any set of laws of nature are also The reader will note that the horrible little word 'cause' does
laws. This stipulation produces what is in some ways a rather not appear in this definition. Causation is a morass in which I
artificial notion of a law of nature. It has, for example, the for one refuse to set foot. Or not unless I am pushed. Certain
consequence that the conjunction of Snell's Law with the arguments for the compatibility of free will and determinism
Principle of the Conservation of Angular Momentum (assum- will force me to say something about the relation between
ing these propositions to be laws) is a law. If anyone is troubled determinism and "universal causation". (See Section 4.4.)
by this stipulation, which I adopt only because it will simplify
the statement of my argument, he may read 'the laws of nature' 3.4 Determinism is a thesis about propositions, but the free-
in the sequel as 'the logical consequences of the laws of nature'. will thesis is a thesis about agents. If we are going to investigate
66 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 67

the conceptual relations between these two theses, we shall created the impression that I believe that human beings can
do well to state the free-will thesis as a thesis about agents somehow enter into causal relations with propositions.? But
and propositions. I propose to do this by devising a way to being able to render a proposition false is nothing so meta-
describe our powers to act—and, by acting, to modify the physically exotic as that; to be able to render a proposition
world—as powers over the truth-values of propositions. This false is to be able to arrange or modify the concrete objects
can be done as follows. Consider the propositions I should that constitute one's environment—shoes, ships, bits of sealing
express if I were to utter any of the following sentences at wax—in a way sufficient for the falsity of that proposition.
the present moment: But how shall we understand this sufficiency? We might
understand 'sufficient' to mean 'logically sufficient'. That is,
(a) 27X 15='405; we might understand 's can render p false' to mean 'It is
(b) Magnets attract iron; within s's power to arrange or modify the concrete objects
(c) Mary Queen of Scots was put to death in 1587; that constitute his environment in some way such that it is
not possible "in the broadly logical sense" that he arrange or
(d) I have never read The Teachings of Don Juan; modify these objects in that way and p be true'. 8 For example,
(e) No one has ever read all of Hume's Enquiry aloud; I can, according to this proposal, render false the proposition
The cup on my desk has never been broken. that this cup is never broken, since I can break this cup—at
(f) least if I have free will—and it is not possible that I break this
(All these propositions are, I think, true.) There is at least one cup and this cup never be broken. If I could move my hand
important and interesting difference between the relations I faster than light, then I could, in the sense proposed, render
bear to (a)-(c) and those I bear to (d)-(f). The difference I false the proposition that nothing ever travels faster than light,
have in mind might be described in various ways: there is since it is not possible that I should move my hand faster
nothing I can do, or ever could have done, about the fact that than light and nothing ever travel faster than light.
(a)-(c) are true, and this is not the case with (d)-(f); the truth I do not believe that this proposal exactly captures the in-
of (a)-(c) is something it is not and never has been within my tuitive notion of being able to render a proposition false, that
power to change, though the truth of (d)-(f) is something that is, the notion of "having control over" the truth-value of a
it is within my power, or once was within my power, to change; proposition. Let us suppose that in 1550 Nostradamus pre-
(a)-(c) are true and I do not have, and never have had, any dicted that the Sphinx would endure till the end of the world.
choice about this, but, though (d)-(f) are true, this is some- And let us suppose that this prediction was correct and, in
thing I have a choice about, or is something I once had a choice fact, that all Nostradamus's predictions were correct. Let us
about. (In making these assertions about the difference be- also suppose that it was within Gamal Abdel Nasser's power
tween (a)-(c), on the one hand, and (d)-(f), on the other, I to have the Sphinx destroyed. Then, I should think, it was
assume I have free will. If I don't, then this apparent differ- within Nasser's power to render false the proposition that all
ence between the two sets of statements is illusory.) Nostradamus's predictions were correct. But this would not
I shall mark this distinction by using an idiom I introduced be the case according to the definition proposed in the pre-
in Section 2.5: I can render, or once could have rendered, all ceding paragraph, since it is possible in the broadly logical
of ( d)-( f) false ;I cannot render, and never could have rendered, sense that Nasser have had the Sphinx destroyed and yet all
any of (a)-(c) false. I rather like the name being able to render Nostradamus's predictions have been correct. That is, there
false for this relation that I bear to (d)-(f) and do not bear to are possible worlds in which the proposition that all Nostra-
(a)-(c). But I admit that this name could be misleading. In fact damus's predictions were correct is true and in which Nasser
it has been misleading. My use of this phrase has on occasion hadi the Sphinx destroyed: worlds in which Nostradamus did
68 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 69

not predict that the Sphinx would endure till the end of the execution of a sentence of death upon a certain criminal, such
world and made no other predictions that would have been a hand-raising being the sign, according to the conventions of
falsified by Nasser's destruction of the Sphinx. the judge's country, of a granting of special clemency. Let us
The best way to rule out such counter-examples would seem further suppose that the judge—call him `J'—refrained from
to be to "build the past into" our definition. More precisely, raising his hand at T, and that this inaction resulted in the
we may define 's can render p false' as follows: criminal's being put to death. We may also suppose that J was
unbound, uninjured, and free from any paralysis of the limbs;
It is within s's power to arrange or modify the concrete that he decided not to raise his hand at T only after a suitable
objects that constitute his environment in some way such period of calm, rational, and relevant deliberation; that he had
that it is not possible in the broadly logical sense that he not been subjected to any "pressure" to decide one way or
arrange or modify those objects in that way and the past the other about the criminal's death; that he was not under
have been exactly as it in fact was and p be true. the influence of drugs, hypnosis, or anything of that sort; and,
I believe that this definition captures the notion that is sug- finally, that there was no element in his deliberations that
gested by the words 'being able to render a proposition false', would have been of any special interest to a student of abnor-
except, perhaps, for the case of false propositions about the mal psychology. I shall argue that, despite all these advantages,
past.' For example, it has the consequence that I can render J could not have raised his hand at T if determinism is true.
the proposition that Socrates died of old age false, since it is My argument for this conclusion will take the form of
not possible that the past should have been exactly as it in comments on the premisses of an "argument" in the logic-text
fact was and Socrates have died of old age. There are three sense: a numbered sequence of propositions, all but the last of
reasons why this feature of our definition need not trouble us, which are the argument's premisses and the last of which is its
however. First, it would be easy enough to remove it by some conclusion. One critic has supposed that my use of an argu-
ad hoc fiat. Moreover, the first argument for the incompati- ment in the logic-text sense constituted my attempting to
bility of free will and determinism will involve only true propo- provide a "formal proof" for a philosophical thesis, and has
sitions. Finally, the 'can render false' idiom will not appear in derided the very possibility of such an enterprise." I applaud
the conclusion of the argument. Thus the odd consequences his values while deploring his exegesis. Formal proofs of philo-
of the definition can affect the conclusion of the argument sophical theses are not to be had and I should be a fool to
only if they result in some defect (such as falsity) in at least attempt any. My critic has mistaken what is really no more
one of its premisses. than a bookkeeping device for an argument, and, in fact, for
a proof. The numbered sequence of propositions below is not
3.5 Now the first argument. I shall imagine a case in which a my first argument for incompatibilism. My first argument for
certain man, after due deliberation, refrained from performing incompatibilism, rather, takes the form of a commentary on
a certain contemplated act. I shall then argue that, if deter- the premisses of the "argument in the logic-text sense". My
minism is true, that man could not have performed that act. argument, in fact, is conterminous with Section 3.5 of this
Because this argument will not depend on any features peculiar book. I shall distinguish my argument from the numbered
to our imagined case, the incompatibility of free will and sequence of propositions it discusses by calling the former
determinism in general will be established, since, as will be the First Argument and the latter the First Formal Argument.
evident, a parallel argument could easily be devised for the These are mere convenient labels and should be understood
case of any agent and any unperformed act. in the light of the present paragraph.
Let us suppose that there was once a judge who had only In the First Formal Argument and in my subsequent com-
to raise his right hand at a certain time, T, to prevent the mentary upon it, I shall use 'T o ' to denote some arbitrarily
 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 71
70

chosen instant of time earlier than J's birth, 'P o ' to denote a If the premisses of the argument were necessary truths, then
proposition that expresses the state of the world at T o , 'P' to its conclusion would be a necessary truth. But its conclusion is
denote a proposition that expresses the state of the world at false in any possible world in which determinism is true and J
T, and 'I,' to denote the conjunction into a single proposition raised his hand at T. Of course, in such worlds at least one of
of all the laws of nature. All these symbols are to be regarded the premisses will be false. This is a simple consequence of the
as "rigid designators". Thus, if I discuss certain counter-factual formal validity of the argument. Take, for example, a world in
situations or. unrealized possibilities or such, and if I use, for which L is the conjunction of all laws of nature, determinism is
example, 'P' in the course of this discussion, I mean 'P' to true, and J raised his hand at T. In such a world, (5) is false,
designate a proposition that in fact expresses the state of the since its consequent is false and its antecedent true. The ante-
world at T, whether or not it would express the state of the cedent of (5) is true in such a world because J could in that
world at T if those situations obtained or those possibilities world have done something incompatible with the truth of the
were realized. (false) conjunction of Po and L: raising his hand. He could have
The First Formal Argument consists of seven propositions, done this in that world simply in virtue of the fact that he did.
the seventh of which follows from the first six: So I don't say that all the premisses of the First Formal
Argument are necessary truths. But I do say that they are true
(1) If determinism is true, then the conjunction of P o and L in all possible worlds in which the story we have told about J
entails P is true. That is, the premisses of the First Formal Argument,
(2) It is not possible that J have raised his hand at T and P and hence its conclusion, follow from our story about J. The
be true story of J is, or we are pretending it is, true. But it's obviously
(3) If (2) is true, then if J could have raised his hand at T, only contingently true: there are possible worlds in which J
J could have rendered P false raised his hand at T; there are possible worlds in which J never
existed at all. And in many such worlds, (7) will be false and
(4) If J could have rendered P false, and if the conjunction at least one among (1)-(6) will therefore be false. And this is
of P. and L entails P, then J could have rendered the just the way things should be, for the conclusion of the First
conjunction of P. and L false Formal Argument oughtn't to be a logical consequence of
(5)
If J could have rendered the conjunction of P o and L incompatibilism alone.
false, then J could have rendered L false Let us now examine the premisses of the First Formal
(6) J could not have rendered L false Argument. The preceding paragraph should make it clear
that in this examination we are entitled to draw upon any
( 7 ) If determinism is true, J could not have raised his hand facts about J and his situation that were presented in the
at T. story we told about J.
That (7) follows from (1)-(6) can easily be established by Premiss (1). This premiss follows from our definition of deter-
truth-functional logic. Note that all the conditionals that occur minism and our specifications of the designations of 'P.', 'L',
in (1)-(7) are material conditionals: the 'could have' that and 'P'.
occurs in them is merely the past indicative of 'can'. ("John
hasn't read this letter, has he?" "I don't know, but I hope Premiss (2). The symbol 'P' is our name for the proposition
you've kept it under lock and key. You know what a dedicated that expresses the state the world was in fact in at T, a time
snoop he is. If he could have read it, he did read it.") at which J's hand was not raised. It is therefore impossible for
One critic has supposed that the premisses of this argument P to be true if J's hand was raised at T, or, indeed, if things
are supposed to be necessary truths." But this is not the case. were in any way different at T from the way they actually were.
72 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 73

Premiss (3). The clause `J could have raised his hand at T' is For all I know, the conjunction of these two propositions is
ambiguous. Using the system of disambiguating brackets in- true. At any rate, let us assume it is true. Given that it is true,
troduced in Section 2.6, we may represent this ambiguity as it seems quite clear that I can render it false if and only if I
follows: this clause could mean either `J could have (raised can visit Alaska. If, for some reason, it is not within my power
his hand at T)' or 'J could have (raised his hand) at T'. I mean ever to visit Alaska, then I cannot render it false. This is a
this clause to be understood in the former sense." quite trivial assertion, and the general principle of which it is
As to the truth of (3): it is obvious that if (2) is true and an instance is hardly less trivial. And it seems incontestable
if J could have raised his hand at T, then there is a certain that premiss (5) is also an instance of this principle.
"arrangement or modification of the concrete objects consti- The general principle of which (5) is an instance need not
tuting J's environment"—J's hand rising at T—such that (i) it be defended only by an appeal to intuition. Let us assume
is not possible that J should arrange things this way and P be that the antecedent of this principle is true. That is, let us
true, and (ii) J could have arranged things this way. assume the truth of both:

Premiss (4). This premiss is an instance, allowing for a shift of (a) q is a true proposition that concerns only states of affairs
tense, of the following general principle: that obtained before s's birth,
If s can render r false, and if q entails r, then s can render and
q false.
'P' for 'r', and 'the conjunction of P o (b) s can render the conjunction of q and r false.
(Substitute `.1' for
and L' for 'q'.) This principle is a trivial truth. For if q entails We shall proceed to derive the consequent of the principle.
r, the denial of r entails the denial of q. Thus anything suf-
Let W (for 'was') be the conjunction into a single proposition
ficient in the broadly logical sense for the falsity of r is also
of all true propositions about the past. Then we have, from
sufficient for the falsity of q. Therefore, if there is some (b) and from the definition of 's can render p false' that was
arrangement of objects that s can produce, which is such that given in Section 3.4:
s's producing it would be sufficient for the falsity of r, there
is some arrangement of objects—the very same one—that s can (c) There is a possible arrangement of objects a such that s
produce, which is such that his producing it would be suf- can bring it about that a is realized and such that the
ficient for the falsity of q. conjunction of W with the proposition that a is realized
Premiss (5). This premiss is an instance of the following general (call this conjunction 'W & a') entails the denial of the
principle: conjunction of q and r ("d (q & r)").

If q is a true proposition that concerns only states of It follows from (a) that W entails q and thus that W &a entails
affairs that obtained before s's birth, and if s can render q. From this and (c) it follows that W & a entails the conjunc-
the conjunction of q and r false, then s can render r false. tion of q and d (q & r). And from this it follows that W & a
entails the denial of r. And, therefore, since s can bring it about
Consider, for example, the propositions that a is realized, s can render r false.
The Spanish Armada was defeated in 1588
Premiss (6). This premiss would seem to be an obvious con-
and sequence of what we said about our powers with respect to
Peter van Inwagen never visits Alaska. the laws of nature in Section 3.3. But a compatibilist might
74 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 75

reject this contention. I can imagine a compatibilist arguing that does justice to our pre-analytic conviction that no
as follows: one can render false a law of nature, then, I predict, I
shall be able to show, according to the strict terms of
Suppose determinism is true and suppose I am not in that definition, that one of the other premisses of the
fact going to raise my hand one minute from now. It argument—I expect it would be either (4) or (5)—is false
follows that there is a certain possible arrangement of if compatibilism is true."
objects—any arrangement that includes my hand's being
raised would do—such that it is not possible in the broadly But now, I think the compatibilist is doing no more than call-
logical sense that I should arrange objects in that way ing our attention to the fact that, if compatibilism is true, then
and the past have been exactly as it was and L be true. some premiss of our argument is false. That is, he is calling our
But since I can raise my hand one minute from now, I attention to the fact that the argument is valid.
therefore can render L false, though, of course, I am not The reader who accepts the characterization of can render
going to do so. If this result sounds queer, that's not my false that was presented in Section 3.4 will perceive a certain
fault; the queerness derives entirely from your definition ironic consequence of the compatibilist's argument (which is
of 's can render p false'. perfectly correct) for the conclusion that compatibilism entails
the falsity of premiss (6): it is the compatibilist, and not the
Well, there is obviously some sense in which I can't render the incompatibilist, who believes in "contra-causal freedom".
laws of nature false: I have no choice about what the laws of
This completes my presentation of the First Argument.
nature are; there's nothing I can do about them. There are
many propositions whose truth-values are within my power, While this argument contains a great deal of detail, the general
but surely the laws of nature are not among them. Let us there- idea behind it is a simple one. Consider any act that (logically)
fore simply set aside the definition of 's can render p false' someone might have performed. If it should turn out that this
that was given in Section 3.4 and ask: aren't the premisses of act was incompatible with the state of the world before that
the First Formal Argument evidently true, however 's can person's birth taken together with the laws of nature, then it
follows that that person could not have performed that act.
render p false' may be defined? Isn't our pre-analytic under-
standing of the notion of one's power over a proposition suf- Moreover, if determinism is true, then just any deviation from
ficiently clear that we can simply see that the premisses of the the actual course of events would be incompatible with any
First Formal Argument are true? The unregenerate compati- past state of the world taken together with the laws of nature.
bilist is likely to respond to these questions this way: Therefore, if determinism is true, it never has been within my
power to deviate from the actual course of events that has
I'm not willing to grant that. The premisses of the First constituted my history.
Formal Argument may be plausible at first glance, but Some philosophers seem to think the statement, 'if an act
that can be said of many false propositions. I can't say was incompatible with the state of the world before a person's
for sure which of your premisses is false since, now that birth taken together with the laws of nature, then it follows
you have set aside the definition given in Section 3.4, that that person could not have performed that act' must be
I'm not clear about what you mean by can render false. based on some sort of muddle. But if one examines an actual
But I am confident that for any reasonably precise speci- case in which a certain act is "ruled out" by the state of the
fication of this notion—one, that is, that is as clear as that world before someone's birth and the laws of nature, then this
given in Section 3.4—I shall be able to show that at least conclusion does seem to follow, and it is very hard indeed to
one of your premisses is false when it is interpretedaccord- see what muddle one's conviction that this is so is based on.
ing to that definition. If you should devise a definition Let us ask, for example, whether I could have visited the star
76 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 77

Arcturus half an hour ago. My having visited Arcturus at that It is not possible that I have visited Arcturus at T and P
moment seems to be ruled out by the state of the world before be true
I was born and the laws of nature. Let us consider the state If (2) is true, then if I could have visited Arcturus at T,
of the world one minute before I was born. At that moment, I could have rendered P false
I was approximately 3.6 X 10' 7 metres from Arcturus. The tem- If I could have rendered P false, and if the conjunction
poral interval separating the moment one minute before I was of Po and L entails P, then I could have rendered the con-
born from the moment that occurred one half hour before junction of Po and L false
now is about 1.16 X 10 9 seconds. It is a law of nature—or so
we believe at present; let's suppose we're right—that no two If I could have rendered the conjunction of P o and L false,
objects have a relative velocity greater than 3 X 10 8 metres per then I could have rendered L false
second." It follows, by simple arithmetic, that I did not visit I could not have rendered L false
Arcturus one half hour ago. I could not have visited Arcturus at T.
Here we have a case in which the proposition that I did
not do a certain thing is deducible from the state of the world This seems to be a perfectly cogent and unexceptionable argu-
before I was born taken together with a law of nature. And ment for the conclusion that I could not have visited Arcturus
it certainly seems to follow from the fact that this deduction at T. Anyone who thinks he can demonstrate that one of the
is possible that I could not have done this thing. It is at any premisses of the First Formal Argument is false, must either
rate true that I could not have done this thing (could not show that his argument does not also "demonstrate" the falsity
have visited Arcturus one half hour ago). Is there some further of the corresponding premiss of the "Arcturus" argument, or
fact, beyond the fact of the deducibility of my non-visit to else he must accept this conclusion and explain why the ap-
Arcturus from the state of the world before I was born and parent truth of the premisses of the "Arcturus" argument is
the laws of nature, to which we should need to refer to justify only apparent. Perhaps someone will be able to do one of these
our belief that I couldn't have visited Arcturus one half hour things, but this project does not look very promising to me.
ago? I don't see what this further fact could be. Is there some On page 74, I said that "the unregenerate compatibilist"
feature of this "deduction of non-performance" that is not a was likely to maintain that if premiss (6) of the First Formal
universal feature of deductions of non-performance, which Argument is true, then either premiss (4) or premiss (5) is
we have capitalized upon? What is it, then? In order to drive false, or, at least, is inconsistent with compatibilism. (Which
this point home, I shall construct an argument parallel to the of (4) and (5) is false, according to this compatibilist, would
First Formal Argument for the conclusion that I could not depend on exactly how one defines 'can render false'.) This
have visited Arcturus one half hour ago. Let us use 'T' to thesis about (4) and (5) is quite implausible when it is ap-
designate the moment of time that occurred one half ,hour plied, mutatis mutandis, to the corresponding premisses of
ago. Let `P' designate the proposition that I did not visit the "Arcturus" argument. Perhaps the reader will find it
Arcturus at T. Let 'P o ' designate the proposition that expresses instructive to reflect on this implausibility and to ask himself
the state of the world one minute before my birth. (Note that whether some of it doesn't rub off on the compatibilist's thesis
Po entails the proposition that at that moment Arcturus and I concerning the First Formal Argument.
were separated by a distance of about 3.6 X 10" metres.) Let us look at premiss (4). Suppose I could have rendered
Let `L' designate the proposition that nothing travels faster P false, that is, could so have arranged things that I have visited
than 3 X 10 8 metres per second. We may now argue: Arcturus at T. If Po and L entail that I did not visit Arcturus
at T, then I could so have arranged things that the conjunction
(1) The conjunction of P o and L entails P of these two propositions is false: my so arranging things as
78 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 79

to visit Arcturus at T is logically sufficient for the falsity of on sets of possible worlds. We saw that various notions inti-
this conjunction. mately related to the notion of a proposition could be defined
Let us look at premiss (5). P. is a proposition about the —rather than simply grasped intuitively—and defined in a very
arrangement of the furniture of the world before I was born— pleasing and elegant way if propositions were taken to be, or
and therefore before I was capable of arranging things. L is a to differ only in irrelevant ways from, sets of possible worlds.
proposition whose truth imposes an upper limit on the relative The Second Argument differs from the First Argument by
velocities of physical objects at any time. Surely if I could exploiting the intimate connection between propositions and
ever have done something sufficient for the falsity of the con- sets of possible worlds. This exploitation takes the form of
junction of these two propositions, it would be just this: get- the elimination of all references to propositions: in the Sec-
ting two objects to exhibit a relative velocity incompatible ond Argument we shall discuss the problem of the compati-
with the truth of L. bility of free will and determinism in terms of possible worlds
These arguments for the truth of (4) and (5) may be com- "directly".
bined in the following intuitive way. I believe that one minute A good deal of the discussion that preceded the First Argu-
before my birth the star Arcturus and I were separated by ment will apply to the Second Argument. The reader may find
a distance of about 3.6 X 10 17 metres. I believe that between a second reading of Section 3.3 helpful.
the moment one minute before my birth and the moment The concept of a possible world is an extremely useful one.
that occurred one half hour ago, there elapsed approximately Yet a considerable number of philosophers who seem to be
1.16 X 10 9 seconds. I believe that no two physical objects ever otherwise intellectually responsible have taken to amusing
move toward each other with a velocity greater than 3 X 10' themselves and their graduate students by engaging in what
metres per second. Now suppose someone trustworthy—God, can only be described as Philistine sneering at this harmless
say—tells me that, while these beliefs of mine are true, I should and fruitful notion. Well, as poor Paley said, who can refute a
be wrong to infer from them that it was not within my power sneer? Nevertheless, while some of our discussion up to this
to have visited Arcturus one half hour ago. For, God says, this point has involved loose talk about, or loose talk couched in
was within my power. What could I conclude from this revel- terms of, possible worlds, the Second Argument will involve
ation? Only one conclusion seems possible: it was within my quantification over them that is meant to be taken in full onto-
power to do something that has in fact never been done, logical seriousness. Therefore, I had better say a few words
namely to travel (relative to a certain object) at a speed greater in defence of these inoffensive objects. Since it is better to be
than 3 X 10' metres per second. If God should also tell me right than original, what I say will be largely cribbed from
that it is a law of nature that nothing exceeds this velocity, Alvin Plantinga's brilliant book, The Nature of Necessity."
then I could only conclude that it was within my power to There is one canard that must be disposed of before I say
violate a law of nature, that is, to work a miracle. This might anything else. Some philosophers have supposed that talk of
be an unpalatable conclusion, but not so unpalatable as the possible worlds is supposed to provide us with a device for
conclusion that it was within my power to have been born defining modal concepts in terms of non-modal concepts.
over a year earlier than I in fact was. Having supposed this, these philosophers proceed to point out
that possible world is itself a modal concept. But this is simply
3.6 The structure of the First Argument was complex indeed. silly. No one, so far as I know, has ever supposed that one
The structure of the Second Argument will be simplicity it- could grasp the notion of a possible world if one had no prior
self. The First Argument made extensive use of the notion of grasp of the notion of possibility. (One might as well say that
a proposition. Our preliminary discussion of propositions Frege believed he could explain truth and falsity by reference
touched upon the idea that propositions could be modelled to the objects the True and the False.)
80 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 81

What are possible worlds? Possible worlds are members of bilities are like the possibility that Socrates teach Plato:
the class of ways things might be or possible ways things might there are possibilities they neither include nor preclude.
be arranged or, simply, possibilities." We quantify over such If a possibility fails to have this feature it is comprehensive.
objects when we say, "There are three ways in which the Earth That is, a comprehensive possibility is a possibility that,
could be destroyed", or "There are still a few possibilities for every other possibility, either includes it or precludes it.
that we haven't investigated". Let us call such objects possi- A comprehensive possibility is a possible world: 'compre-
bilities.' Possibilities are designated by phrases consisting of hensive possibility' and 'possible world' are, in my usage,
`the possibility that' followed by a sentence, with main verb stylistic variants.
in the subjunctive mood, that expresses a possibly true propo- I shall assume without argument that there are compre-
sition. (Phrases like 'the possibility that 7 + 5 equal 13' thus hensive possibilities. If there are any, there is more than one.
fail to designate anything, there being no such possibility.) (To say there is exactly one comprehensive possibility is
Possibilities are either realized or unrealized. For example, Spinozism.") I shall assume without argument that it is false
the possibility that Socrates teach Plato is realized, but, if that all comprehensive possibilities are unrealized. But if any
things had gone differently, it would have been unrealized. comprehensive possibility is realized, exactly one is, since it
To say that a possibility is unrealized is obviously not to say is an obvious consequence of our definitions that any two
that it doesn't exist. For there obviously exist unrealized possi- comprehensive possibilities preclude each other. The realized
bilities, such as the possibility that Socrates never have met comprehensive possibility is the actual world: 'the realized
Plato. (To say that there exist no unrealized possibilities is to comprehensive possibility' and 'the actual world' are, in my
adopt Spinozism, a doctrine equivalent to the doctrine that a usage, stylistic variants.'
proposition is true if and only if it is necessarily true.) Since possible worlds are possibilities and possibilities are
But not just any possibility is a possible world, for not just abstract objects, possible worlds, including the actual world,
any possibility is sufficiently "filled out" to deserve to be are abstract objects. Therefore, what philosophers sometimes
called a world. In order to say which possibilities are possible call "the world", what, indeed, I have called "the world" in
worlds, we shall need two definitions. Let us say that a possi- this book, is not the same object as the actual world. The
bility includes a second possibility if it is impossible for the world is the universe, or the cosmos, or what Professor Geach
former to be realized and the latter unrealized. (Moreover, has called "the upper limit of the series: the solar system, the
every possibility includes itself.) Let us say that a possibility galaxy, the system of galaxies . .." If things were different in
precludes a second possibility if it is impossible for them both some minor way, if, say, I had not shaved this morning, then
to be realized. (Moreover, no possibility precludes itself.) Thus, of course the world would still have existed, albeit it or a few
the possibility that Socrates teach Plato includes the possibility of its components would have had different features. But if I
that 2 + 2 equal 4, the possibility that Socrates exist, the had not shaved this morning, the abstract object that is in fact
possibility that Plato exist, and the possibility that Socrates the actual world would not have been actual, though it would
teach someone. It precludes the possibility that Socrates fail still have existed. If I had not shaved this morning, the actual
to exist and the possibility that Socrates teach no one. It world would have been an unrealized possibility, and some
neither includes nor precludes the possibility that Socrates other possible world (some other comprehensive possibility)
teach Aristotle or the possibility that the most famous snub- would have been actual (realized).
nosed Greek philosopher teach Plato. 1° For every possi- Now it may be a bad idea for me to use the word 'world' in
bility, there is exactly one possibility it both includes and is this double sense. Perhaps I should call the world 'the cosmos'
included by: itself. (The preceding sentence is a statement or 'the universe'. Or perhaps I should refrain from calling com-
of the principle of individuation for possibilities.) Most possi- prehensive possibilities 'worlds'. But I like the word 'world'
82 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 83

as a name for the cosmos better than its pompous Greek and 3.7 The Second Argument, like the First Argument, is a com-
Latin alternatives. And to call possible worlds 'comprehensive mentary on the premisses of an argument in the logic-text
possibilities' outside contexts in which I am specifically dis- sense. I shall call the "logic-text" argument 'the Second Formal
cussing what I take to be their nature—that is, in contexts in Argument'. The vocabulary of the Second Formal Argument
which I am using them as devices for presenting propositions is that of the first-order predicate calculus with identity, sup-
about free will and determinism—would be an inadvisable de- plemented by four non-logical constants. These constants, with
parture from customary philosophical terminology. 21 At any suggested English readings, are:
rate, the construction of my sentences will remove all possi-
bility of ambiguity. If I say "the world", I must mean the A: the actual world;
cosmos, for there are many comprehensive possibilities. If I Sxy : x shares a slice with y;
talk of "worlds", I must mean comprehensive possibilities, for Nxy : x is nomologically congruent with y;
there is only one cosmos. If I say "the actual world", I cannot
mean the cosmos, for 'actual' in my usage means 'realized' Hxy : x has access to y.
and it is possibilities not cosmoi that can be said to be realized We shall also employ a one-place defined predicate, 'D':
or unrealized. Dx= df (3y) (Sxy) & (y) (Syx & Nyx. y = x).'
We need one more definition and then we shall be ready to
present the Second Argument. We say that a proposition is `Dx' is read, 'x is deterministic'.
true at, or in, a given world if that proposition would be true The intended sense of 'A' is evident from its suggested
if that world were actual." An important special case of the English reading. This term is to be understood as short for the
notion of the truth of a proposition at a world is the notion definite description 'the actual world': 'A' is not to be taken
of the truth of a singular existential proposition at a world. as a proper name for, or a "rigid designator" of, the world
Applied to this special case, the above definition tells us that that happens to be actual.
a given object—the Eiffel Tower, say, or the number four— The predicate 'S' represents the symmetrical, reflexive, and
exists at, or in, a given world provided that object would exist non-transitive relation that has the class of all possible worlds
if that world were actual. 23 It is an obvious consequence of as both its domain and its range and which two worlds bear
this definition that there are objects that exist in more than to each other if they are indistinguishable at some instant. One
one world. For the proposition that Socrates had a long, might want to respond to this explanation by asking, "Indis-
straight nose, while false, is false only contingently. Therefore tinguishable with respect to what?", for it is not clear whether
there are possible worlds, none of them actual, at which there is any such thing as absolute-indistinguishability-at-an-
Socrates had a long, straight nose. I am aware that there is instant. Perhaps w 1 and w 2 are indistinguishable at t if, for
supposed to be something called "the problem of trans-world every property x and object y, y exists and has x in w 1 at t if
identity", but, though I have assiduously attempted to find and only if y exists and has x in w 2 at t? But suppose there
out what this problem is, I have never succeeded. To those are not only well-behaved properties like being blue or being
who think there is such a problem, I make a gift of a rather radioactive; suppose there are also such unruly properties as
more pressing problem, the problem of trans-propositional being grue and having resigned the Presidency five years since.
identity: how can the proposition that Nixon is a villain and A moment's reflection should show that if these properties and
the proposition that Nixon is an honest man both be about their relatives indeed exist, then two worlds are indistinguish-
the same man, when one is about a villain and the other is able, by the above definition, at some instant only if they are
about an honest man? indistinguishable at all instants, a result that obviously renders
,

"being indistinguishable at an instant" a worthless notion.

84 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 85

If this problem has no solution, then, I think, determinism Or if we are willing to think of a possible world (strictly
is an incoherent doctrine. Let us, therefore, simply assume speaking, to think of the universe that exists in that world) as a
there is some solution or other to it. Our remarks in this chap- compact sequence of instantaneous three-dimensional "slices",
ter will be sufficiently general that they will be consistent with then we may say that the indistinguishability relation holds
just about any specification of what is meant by two worlds between two worlds just in the case that they have a slice in
being indistinguishable at an instant. This is essentially what common—hence, the suggested English reading of 'S'.
we did in Section 3.3 when we left the concept of "state" an Let us now turn to 'N'. I mean this predicate to represent
open concept. Indeed, if we had the concept of "state" that the equivalence relation that holds between two worlds just
would be required for a full specification of the meanings of in the case that the laws of nature are the same in each. An
the terms employed in the First Argument, we could easily alternative reading of 'Nxy' is 'what is physically necessary and
define 'S': impossible in x is what is physically necessary and impossible
in y', provided these modal terms are not taken as meaning
Sxy= d f (3 t) (the state that x is in at t = the state that `what is now physically impossible or necessary, given the
y is in at t). actual past'. (For example, it is, I suppose, now physically
(The definiens of this definition contains a specimen of a cer- necessary that a total eclipse of the Sun be visible, weather
tain sort of loose talk I have indulged in before, and, owing permitting, in Siberia on 9 March 1997. But I should think
to its convenience, shall continue to indulge in. A possible there are possible worlds—though perhaps only worlds in
world is an abstract object, and does not change with the pass- which the "entire past" was different—that bear the relation
age of time; or, at least, it changes only in the "Cambridge" I intend 'N' to express to the actual world, and in which no
sense. When I talk of the state that a possible world w is in at eclipse will then be visible.) Rather, these modal terms must
time t, I am to be taken as talking about the state that, at w, be taken to refer to what is "timelessly" physically necessary
the world—the cosmos, the universe—is in at t.) and impossible.
I must be satisfied to leave the content of the indistinguish- Examples of worlds that do not bear this relation to the
ability-at-a-time relation pretty much open, pending the for- actual world would probably be controversial. But I should
mulation of a theory of properties that allows us satisfactorily think that there are probably possible worlds in which the
to distinguish between being grue and having only six days fundamental constants of nature—Planck's Constant, the speed
to live on the one hand, and well-behaved properties like being of light, the charge on the electron, and the universal constant
blue and having only six moving parts on the other. But I of gravitation—have values different from their actual values.
offer the reader two ways of looking at this relation that seem Whether there are worlds in which the laws of nature have a
to me to have some heuristic value. different structure from their actual structure—worlds, say, in
Let us imagine a Leibnizian God, who somehow "stands which the energy associated with a photon is not a function
outside" all possible worlds and is able somehow to "examine" of its wavelength, or worlds in which the force of mutual
them individually, sub specie aeternitatis. Presumably, such a gravitational attraction between two particles is inversely pro-
God would be able to restrict his examination of a world to portional to the cube of the distance between them—is some-
(focus on, as it were) flit way that world is at a single instant of thing I should not care to speculate about. (There do seem to
time. If we find this way of speaking intelligible, then we may be deep conceptual reasons for the structure exhibited by
say that the indistinguishability relation holds between two physical law. But we view these matters through a glass darkly.)
worlds if there is some instant t such that, if God were to exam- Let us now examine our defined predicate, 'D'. This predi-
ine each of these worlds "as it as at t", he could observe, on the cate represents a property of some possible worlds, the
basis of these examinations alone, no difference between them. property of being deterministic, which may be informally
86 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 87

characterized as follows: a world is deterministic if that world "accessibility" relation holds between worlds and worlds.) In
itself is the only world that both shares a slice with it and has order to show what is meant by saying that a person "has
the same laws of nature it does. access to" some merely possible world—we may take it to be
Perhaps an example will make the point of this definition true by definition that everyone has access to the actual world
clear. Let w be some possible world that shares with the actual —I shall first give some translations from ordinary talk about
world, A, a slice taken at the instant Harold's eye was pierced abilities to "access" talk. We translate, 'Napoleon could have
by a Norman arrow. It may share indenumerably many other defeated Wellington at Waterloo' as 'Napoleon had access to
slices with A; it shares at least that slice. And let us suppose some possible world in which he defeated Wellington at Water-
that in A and w the laws of nature are the same. There are two loo'. We translate 'It is within my power to keep the money
possibilities: A and w may be the same world or they may be I found and within my power to return it' as 'I have access to
distinct. But A and w can plausibly be called 'deterministic' some world in which I keep the money I found and I have
only if they are identical. For suppose they are not identical: access to some world in which I return it'. Non-actual worlds,
let us say that w is one of those worlds in which a thermo- we remember, are unrealized possibilities. Thus "access" talk
nuclear war was fought in 1966. If there is a world that has all is a way of organizing our talk about unexercised abilities by
the properties we have ascribed to w, it would be odd to say reference to unrealized possibilities: an unexercised ability is
that anything that could reasonably be called 'determinism' treated as an ability to realize some unrealized possibility. This
is true. The actual world is a world in which a certain situation seems to be a harmless and quite intelligible rewriting of ordi-
in 1066 did not precede a thermonuclear war by nine hundred nary ability-talk. Like our earlier treatment of unexercised
years. But in w, a world having exactly the same laws of nature, abilities as abilities to render true propositions false, it is some-
precisely the same situation was followed, after nine hundred what artificial. The artificiality has in both cases the same
years, by a thermonuclear war. In other words, if our descrip- excuse: if we are to investigate the conceptual relations be-
tion of w is consistent—that is, if w exists—then, though there tween free will and determinism, it is hardly to be supposed
was in actuality no thermonuclear war in 1966, such a war that we shall succeed if the vocabulary we use to state the
was a possibility relative to the laws of nature and the state thesis of determinism and the vocabulary we use to state the
of the world in 1066. But 'determinism' must, if violence is thesis of free will have no elements in common. Therefore, if
not to be done to every traditional association that word has, determinism is formulated as a thesis about possible worlds (or
be used to refer to the thesis that there are no such alternative about propositions) the best plan would seem to be to try to
possibilities. Let us, therefore, understand by 'determinism' formulate the free-will thesis as a thesis about possible worlds
the thesis that the actual world is "deterministic": according (or propositions). Of course one might apply this strategy in
to determinism, every world distinct from the actual world reverse: the free-will thesis might be formulated as a thesis
either differs from it at every instant, or, if it differs from the about acts, and determinism as a thesis about events, acts being
actual world at only some instants, is governed by a different treated as a species of event. I prefer the present course. Here
set of laws of nature. is a good rationalization of my choice (that is, a good reason
for it that did not in fact influence my choice): the action-
3.8 So much for determinism. Now for free will. We shall theory industry has generated so much controversy about acts
talk of a person's abilities in terms of which worlds he "has and events that a good many philosophers would be more
access to". (Thus, our relation having access to bears little re- interested in whatever I had to say about acts and events than
semblance to the "accessibility" relation that figures in a they would be in what I had to say about free will. But I want
Kripke-style semantics for modal logic. Our "having access" to talk about free will.'
relation holds between persons or agents and worlds; the If "access" talk is artificial, it is not therefore unusable in
88 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 89

everyday life. The following bit of dialogue shows how our Let us call any location within the system of corridors an
moral discourse might sound if we gave up ordinary ability- event. Then we may say that the man can bring about a certain
talk and adopted in its place the language of access to possible event just in the case that there is some path through the cor-
worlds. ridors leading from where he is to that event (location) without
passing through any barred corridors.'
A. You ought not to have cut my lecture on Friday. Let us call a possible world any infinitely long path through
B. But I did not have access to a possible world in which I the system of corridors that does not cross itself. The actual
attended your lecture on Friday, since I suffered an un- world is that one path through the corridors along which the
foreseen paralysis of my legs on Thursday that mysteri- man always has walked, is walking, and always will walk. Those
ously vanished on Saturday. In every possible world to worlds to which the man has access at any given moment are
which I had access, I spent Friday in bed. just those infinite paths that do not pas's through any barred
corridors, and which are continuations of the path-segment
A. Have you access to a possible world in which a doctor
along which he already walked.
writes me a note verifying your story? This picture has its limitations as a model for talk of access
B. Unfortunately not: no possible world that I had access to possible worlds: it is no longer applicable if we assume (as
to on Friday contained a doctor in this city who makes is the case) that which possible world the actual world is de-
house calls. pends on the choices of more than one person. We might, of
And so on. Perhaps the relationship between ordinary ability- course, elaborate our imagery by assuming that there are n
talk and access-talk might best be explicated by showing the persons walking through the system of corridors, and call a
relationship between access-talk and a rather artificial near- possible world any n-membered set of infinite paths. The
relation of ordinary ability-talk, namely, talk of one's abilities actual world, then would be the set of paths that are taken,
with respect to bringing about events of some specified sort: and a person P would have access at any given moment to
to say that a person can bring about an event satisfying a cer- those possible worlds that are such that (i) they differ from
tain description is to say that he has access to at least one the actual world by at most one member, (ii) this member is
possible world in which an event satisfying that description the path that P is in fact going to take, and (iii) each of them
happens; and to say that a person has access to a possible world that does not contain the path that P is in fact going to take,
satisfying a certain description is to say that he can bring about contains instead a continuation of the path-segment that P
events of a sort that happen only in worlds satisfying that has already walked that does not pass through any barred
description. corridors.
In order to make this relationship intuitively more clear, I But this more elaborate picture breaks down in its turn if
shall devise a sort of metaphor or picture that might be used we assume (as is the case) that persons come into and go out
as an informal model both for talk of being able to bring about of existence, and that the choices they make partly determine
events and talk of access to possible worlds. Consider a man what choices it is possible for their fellows to make. I do not
who is walking through an infinite system of branching corri- think, however, that there is anything to be gained from con-
dors. He has always been walking and must always keep walk- structing a yet more elaborate picture in order to accommodate
ing, never stopping and never retracing his steps. He finds these facts. We should note that the relation expressed by 'H'
that some branches are sealed off by bars and some are not. is, strictly speaking, a non-temporal relation between persons
Frequently he comes to a branching of the corridor from and possible worlds: it is not a triadic relation satisfied by
which at least two unbarred branches lead away, and he must ordered triples of the form (person, world, instant), but a
make a choice about which to take. dyadic relation satisfied by ordered pairs of the form (person,
90 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 91

world). For example, if Tom, a doctor, once had access to a that a person's unexercised abilities may be described in
possible world w in which his profession is law, then, even if terms of his having access to particular non-actual worlds.
he no longer has access to w, it is true that Tom bears the re- This will result in a very neat formal argument. It would
lation expressed by to w. Thus, a better English reading of be a simple, almost a mechanical, task to transform this argu-
`I-1.xy' might be 'x had, has, or will have access, at some point ment into an argument involving access to sets of worlds,
in his life, to y
'. and the only expense would be some clutter. But part of the
It is also perhaps worth noting a respect in which the point of the Second Argument is the elimination of clutter.
"access" idiom is unrealistic. A possible world is a very detailed Those who like clutter will find all they could desire in the
thing. To see what I mean by this, try to name any possible First Argument. My hope is that the Second Argument is
world other than the actual world. Probably this cannoqe unrealistic only as the Kinetic Theory of Gases (which ignores
done: for any predicate we might devise, this predicate will certain features of the real world, such as inter-molecular
be satisfied by many non-actual worlds if it is satisfied by forces) is unrealistic: it makes certain simplifying assump-
any. That is, while we can pick out various sets of non-actual tions, assumptions that are literally false, in order to lay bare
worlds—for example, the set of worlds in which Napoleon won the most important features of an interrelated family of
at Waterloo—we cannot pick out any single merely possible phenomena.
world." Thus, talk of access to particular merely possible
worlds is very likely too "fine-grained" to be a literally correct
way of describing our unexercised abilities: none of my un- 3.9 Let us now return to the Second Argument. We shall call
the thesis
exercised abilities is so "definite" an ability as to be correctly
describable as an ability to get some particular possible world (3 x) (3y) (Hxy & y A)
to be actual. No act I might have performed is such that my
having performed it entails the actuality of any particular the minimal free-will thesis (MFT). MFT is a very weak thesis.
world. No act I might have performed is such that, for some It tells us only that some person, past, present, or future, had,
world, if I had performed that act, then that world would has, or will have access to some possible world besides the
have been actual. No world is the world that would have been actual world. MFT is true, for example, if Julius Caesar had
actual if I had raised my hand one minute ago. Thus it would access to some world w in which he did not cross the Rubicon,
probably be more realistic to describe my unexercised abilities even if no other person, past, present, or future, has access to
in terms of access to sets of possible worlds. For example, the any world besides A, and Caesar himself had access only to w
state of affairs that consists in my having an unexercised ability and A. But if MFT is false, then any more interesting free-will
to break this cup could be described as being identical with thesis is false. And, therefore, if determinism is incompatible
the state of affairs that consists in my having access to the set with MFT, it is incompatible with any more interesting free-
of possible worlds in which I break this cup. Moreover, the will thesis.
concept of access to a set of possible worlds can easily be de- Determinism may be represented in our present vocabulary
fined in terms of the concept of the ability to render a propo- by the simple formula `DA'. Clearly the denial of MFT can-
sition false: not be formally deduced from 'DA'. Nevertheless, there is
an important sense in which determinism is incompatible
s has access to the set of worlds x=df (3 y) (y is true in
just exactly those worlds that belong to x and s can render with MFT. There are two theses, which I shall call 'meta-
the denial of y false)." physical assumptions', each of which seems more likely to be
true than either determinism or the minimal free-will thesis,
Despite these difficulties, I shall retain the useful fiction such that the denial of MFT may be formally deduced from
92 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 93

determinism and these two theses. The two metaphysical The denial of MFT can easily be seen to be formally deriv-
assumptions are: able from `DA', MAA, and MAB. (Here is a sketch of a proof
MAA (x) (y) (Hxy D Sy A) of this that employs techniques common to most logic texts.
Assume Universally instantiate MAA and MAB. Modus
MAB (x) (y) (Hxy D NyA). Ponens yields `SyA' and `NyA'. From these two formulae and
MAA asserts that every world to which any person has the universal instantiation of the second conjunct of the for-
access must be indistinguishable from the actual world at some mula that `DA' abbreviates, we get, by Modus Ponens , 'y = A'.
instant. Or, alternatively, every world to which any person has Thus, by Conditional Proof and Universal Generalization, we
access must share a slice with the actual world. For example, have, '(x) (y) (Hxy D y = A)', which is logically equivalent to
however many possible worlds I have access to, surely they the denial of MFT.) The argument having 'DA', MAA, and
must all be indistinguishable from the actual world at some MAB as its premisses, and the denial of MFT as its conclusion,
time in the remote past—say, 10,000 BC, or indeed, any time is the Second Formal Argument. We have already commented
before I was born. In terms of the "infinite-system-of-corri- upon the premisses of this argument. We may put the con-
dors" metaphor: all the possible worlds (paths) that I have clusion of the Second Argument thus: if determinism is true,
access to are continuations of the path-segment I have already then no one has access to any non-actual world. That is, no
travelled. MAA may be regarded as a statement of the familiar one has an ability that may correctly be described as an ability
principle that no one can change the past. to realize some in fact unrealized possibility.
MAB asserts that no person has access to any world in which
the laws of nature are different from what they are in the 3.10 The Third Argument is a modal. argument; that is, it
actual world. This seems undeniable, for, as we have seen, no involves a modal operator, an operator that attaches to sen-
one can render a law of nature false. Moreover, it is clear that tences that have (or that express propositions that have) truth-
it is not up to anyone whether any given true proposition is a values, to form sentences that have truth-values; and the
law of nature. Such things are not matters of human choice, truth-value of a sentence formed in this way is not in every
any more than the theoremhood of a given mathematical case a function of the truth-value of the sentence to which
proposition is a matter of human choice. It is, in my opinion, the operator attaches. The operator is 'N'. For any sentence
possible "in the broadly logical sense" for the laws of nature p, the result of prefixing p with 'N' may be regarded as an
to have been different. The speed of light might have been dif- abbreviation for the result of flanking 'and no one has, or
ferent; Planck's constant might have been different; the charge ever had, any choice about whether' with occurrences of
on the electron might have been different. But if such possi- p. Thus
bilities exist, and if they are really possibilities about alter-
N All men are mortal
native sets of laws of nature, then it is beyond our power to
realize them. And this is just another way of saying that no is an abbreviation for
one has access to possible worlds in which the laws of nature
are different. It may of course be true that some people have All men are mortal and no one has, or ever had, any
access to possible worlds in which our beliefs about the laws choice about whether all men are mortal.
of nature are different. Perhaps if Newton or Einstein had `N' is a very interesting operator, and the task of constructing
acted differently, had done certain things they could have done a plausible and complete logic for 'N' would be an interesting
but did not do, then we should now hold other beliefs about one. It would be even more interesting if the vocabulary of
the laws of nature. But this, of course, is irrelevant to our this logic included the usual "alethic" modal operators. But it
present concerns. will not be necessary for our purposes actually to construct
94 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 95

such a logic.' I think the following two inference rules ought We now introduce a premiss:
to be valid rules of this logic: (4) N130 .
(a) op Np From (3) and (4) we have by Rule (/3):
(3) N(p D q), Np Nq. (5) N(L D P).
I shall require no formulae that contain 'N' to be logically true We introduce a second premiss:
other than those that these two rules allow us to deduce from (6) NL.
the empty set of premisses. The operator '0' in Rule (a) rep-
Then, from (5) and (6) by (0):
resents "broadly logical necessity".
The Third Argument will have the following logical struc- (7) NP.
ture: from a logical consequence of determinism (determinism This deduction shows that if determinism is true, then no one
being conceived as in the First Argument), we shall deduce ever has any choice about anything, since any sentence that
that no one has any choice about anything whatever. expresses a truth may replace 'P' in it. Consider, for example,
In the Third Argument we shall once more use the symbols the question whether anyone had a choice about whether
`130 ' and `L', which appeared in our statement of the First Richard Nixon would receive a pardon for any offences he
Argument. But these symbols will have a slightly different might have committed while in office. Nixon did receive such
sense. In the First Argument, 'P.' and 'I] were used as names a pardon. Therefore, if determinism is true,
for certain propositions. In the Third Argument, we shall use o(Po & L. D Richard Nixon received a pardon for any
them as abbreviations for sentences expressing these propo-
sitions. Thus, in our present usage, 'P.' and 'L' have the gram- offences he might have committed while in office).
matical features of sentences and not of terms: in the usage of Therefore, if (a) and (13) are valid rules, and if NL and NP„,
the First Argument, 'Po is true' was grammatical and 'P. D L' then we have:
was ungrammatical (for the same reason that `Quine D David- No one had any choice about whether Richard Nixon
son' is ungrammatical); in our present usage, 'P. D L' is gram- received a pardon for any offences he might have com-
matical, and 'Po is true' is ungrammatical (for the same reason mitted while in office.
that 'All men are mortal is true' is ungrammatical). Most of us, I presume, think this conclusion is false. Most of
We shall once more use the letter `P'. This letter will serve us think Gerald Ford had a choice about whether Nixon would
as a dummy for which one can substitute any sentence one receive a pardon. (Even if someone thinks that Ford's actions
likes, provided it expresses a true proposition—`Quine finds were wholly controlled by some cabal, he none the less thinks
quantified modal logic puzzling', say, or 'Paris is the capital someone had a choice about whether Nixon would receive a
of France'. pardon.) But the above deduction (The Third Formal Argu-
If determinism is true, then it follows that ment) shows that if we wish to accept this conclusion we must
(1) ❑ (P. & L. D P) reject one of the following five propositions:
is true. From (1) we may deduce Determinism is true;
(2) El(Po D (L D P))
NP.;
by elementary modal and sentential logic. Applying rule (a) NL;
to (2), we have: Rule (a) is valid;
(3) N(13. D (L D P)). Rule (fl) is valid.
96 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 97

My choice, of course, is to reject determinism. But let us exam- the laws of nature hold, or when we say that a proposition is
ine the alternatives. morally required if it is true in all morally permissible worlds.
I do not see how anyone can reject '1\11 30 ' or `NI.'. My reasons For example, we might say that Np is true if p is true in both
are essentially those that I gave in support of MAA and MAB the actual world and in all non-actual worlds about whose
in the Second Argument and in support of premisses (5) and actuality human beings have, or once had, a choice. Interest-
(6) of the First Formal Argument. The proposition that P. is ingly enough, the question of the membership of W is of no
a proposition about the remote past. We could, if we like, formal significance. If we accept any definition of Np of the
stipulate that it is a proposition about the distribution and following form: `Np is true if and only if p is true in all worlds
momenta of atoms and other particles in the inchoate, pre- such that . . .', where the condition that fills that ellipsis makes
siderial nebulae. Therefore, surely, no one has any choice about no mention of p, then (a) will "come out" valid. (Obviously,
whether Po . The proposition that L is a proposition that if p is true in every member of W, and if p D q is true in every
"records" the laws of nature. If it is a law of nature that angu- member of W, then q is true in every member of W.) Despite
lar momentum is conserved, then no one has any choice about its utter triviality, this formal result is not without persuasive
whether angular momentum is conserved, and, more generally, force. Nevertheless, it is far from decisive. It depends on the
since it is a law of nature that L, no one has any choice about assumption that there is some set of worlds such that Np can
whether L. plausibly be thought of as making the assertion that p is true
I do not see how anyone could reject Rule (a). If (a) is in every member of that set. While this assumption seems
invalid, then it could be that someone has a choice about what right to me, I have no argument for it, and a person who was
is necessarily true. Hardly anyone besides Descartes has been determined to reject (0) might very well reject the assumption.
willing to concede such a capability even to God. No one, so Secondly, one might attempt to show that an inference rule
far as I know, has ever suggested that human beings could is valid by "reducing" it to certain generally accepted inference
have a choice about what is necessarily true. (I take it that rules, that is, by deducing its validity from the assumption
"conventionalist" theories of modality don't really imply the that these generally accepted rules are valid. But it seems un-
thesis that human beings have a choice about what is neces- likely that this method can be used to show that ((3) is valid:
sarily true; or, at least, not in any sense relevant to our present I can think of no generally accepted inference rules that seem
concerns.) relevant to (g), much less of any that it can be deduced from.
Only Rule (g) remains to be considered. The validity of (P) The prospect of showing (a) to be valid, therefore, appears
is, I think, the most difficult of the premisses of the Third to be bleak, though perhaps no bleaker than the prospect of
Argument to defend. How might one go about defending it? showing anything of philosophical interest. I must confess
How, in general, does one go about showing that an inference that my belief in the validity of ((3) has only two sources, one
rule is valid? There would seem to be two ways. incommunicable and other inconclusive. The former source is
First, one might employ the methods of formal semantics. what philosophers are pleased to call "intuition": when I care-
In the present case, since 'N' is a modal operator, the methods fully consider (P), it seems to be valid. But I can't expect any-
of possible-world semantics might seem promising. Here is a one to be very impressed by this fact. People's intuitions, after
sketch of how we might apply these methods to (0)." We first all, have led them to accept all sorts of crazy propositions and
delimit a certain set W of worlds and say that Np is true just many sane but false propositions. (The Unrestricted Compre-
in the case that p is true in all these worlds. This would amount hension Principle in set theory and the Galilean Law of the
to a semantical definition of 'N'. It would be a definition of Addition of Velocities in physics are good examples of propo-
the same sort as that which we give when we say that a propo- sitions in the second category.) The latter source is the fact
sition is physically necessary if it is true in all worlds in which that I can think of no instances of (/3) that have, or could
 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 99
98
possibly have, true premisses and a false conclusion. Consider, he does not come to share, the intuition that I have expressed
for example, these two instances of (3): by saying, "when I carefully consider (13), it seems to be
valid". 31
Alice has asthma and no one has, or ever had, any It is interesting to note that rule (i3) seems to figure in recent
choice about whether she has asthma; discussions of the philosophical and social implications of
socio-biology and the question whether certain widespread
If Alice has asthma, she sometimes has difficulty features of human social behaviour are genetically determined.
breathing, and no one has, or ever had, any choice Consider, for example, the proposition (P) that there are cer-
about whether, if she has asthma, she sometimes has tain jobs (jobs that both sexes are physically capable of per-
difficulty breathing; forming) such that, in every society, these jobs devolve almost
hence, Alice sometimes has difficulty breathing, and no one entirely upon women. Suppose a socio-biologist alleges that
has, or ever had, any choice about whether Alice some- there is a certain fact or set of facts (F) about the evolutionary
times has difficulty breathing. history of our species that explains why P is true. Anyone who
says this is likely to be the target of some such criticism as this:
The sun will explode in 2000 AD, and no one has, or What you are saying is that "women's role" is genetically
ever had, any choice about whether the sun will ex- determined, and thus that all attempts at changing the
plode in 2000 AD; role of women in this or any other society are doomed by
If the sun explodes in 2000 AD, all life on earth will biology. This doctrine is pernicious. You are not a scien-
end in 2000 AD, and no one has, or ever had, any tist but an ideologue, and the ideology you are peddling
choice about whether, if the sun explodes in 2000 AD, makes you a most useful prop for the existing system.
all life on earth will end in 2000 AD; Anyone who is the target of such criticism is likely to defend
hence, All life on earth will end in 2000 AD, and no one has, himself in some way pretty much like this:
or ever had, any choice about whether all life on earth
will end in 2000 AD. Not so. While I believe that F explains why P is true, I
do not say that F makes P inevitably true. Given F, there
These arguments are clearly valid. There is simply no way is a tendency for P to be true, but tendencies can be re-
things could be arranged that would be sufficient for the truth sisted. I do not say that "biology is destiny". It may well
of their premisses and the falsity of their conclusions. Take be that we have a choice about whether we shall behave
the second. Conceivably we could do something to prevent in accordance with this tendency that our evolutionary
the explosion of the sun. Then perhaps the conclusion of this heritage has presented us with.
argument would be false; but its first premiss would also be Now I am not so much interested in where the right lies in dis-
false. Perhaps we could erect an enormous shield that would putes that take this form—after all, where it lies may depend
protect the earth from the explosion of the sun; if we could on what is substituted for 'F' and for `P'—as I am in its under-
do this, the conclusion would be false. But the second premiss lying logic. I believe that the logical skeleton of this dispute
would also be false. Perhaps we could spread a poison that looks something like this:
would destroy all life on earth before 2000 AD; but in that
case too the second premiss would be false. I cannot help feel- CRITIC: It follows from your position that the
ing that the reader who makes a serious attempt at construct- premisses of the following valid argu-
ing a counter-example to (13) will begin to appreciate, even if ment are true:
100 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 101

N F obtains involving a Freudian or a Marxist and the point of my example

would have been the same. Anyone who denies the validity
N(F obtains J P is true) of (13), it seems to me, must react to such disputes in one of
hence, N P is true two ways. He must either call them pseudo-disputes that arise
because both parties to the dispute wrongly accept ((3), or he
SOCIO-BIOLOGIST: The first premiss is certainly true, but must contend that the dispute does not really involve ((3) after
the second does not follow from my all. In the former case, he should recommend that the socio-
position and may very well be false. biologist reply to the critic like this:
It would probably never occur to the socio-biologist to deny I admit that F is a fact about the history of our species
that the conclusion of the argument he has been charged with and that that's something no one has a choice about. I
endorsing actually does follow from its premisses. And if he admit if F is a fact about the history of a given species,
did deny this, then the critic would rightly charge him with then, in societies that comprise members of that species,
sophistry, for if it is granted that no one has any choice about certain jobs will devolve almost entirely upon women,
whether, given that our history is such-and-such, we do so- and that no one has any choice about that. Still, we do
and-so, and if it is granted that our history is such-and-such have a choice about whether, in our society, these jobs
and that we have no choice about this, then it just obviously will devolve almost entirely upon women.
does follow that no one has any choice about whether we do
so-and-so. This does not, of course, entail that ((3) is valid, for I cannot imagine anyone saying this with a straight face. In the
it may be that while this instance of ((3) is valid, other instances latter case, he owes us an account of these disputes that shows
of (() are invalid. But I think that anyone who said that, while how the apparent acceptance of ((3) by both parties is merely
the argument the critic has formulated is valid, ((3) is not valid, apparent.
would be saying something that is not on the face of it very The point of this discussion may be summed up in a ques-
plausible. The validity of (13), in its full generality, certainly tion: Why is none of the participants in the debates about
does seem to be part of the "common ground" in the socio- biological determinism a compatibilist? Perhaps the answer is
biological dispute I have imagined. (Despite the fact that I that the participants in these debates take the idea of biologi-
have imagined it, its logical structure is typical of disputes cal determinism much more seriously than philosophers are
about biological determinism.) That is, (13) seems to be ac- accustomed to take the idea of "universal" or "Laplacian"
cepted, and properly so, by both sides in the dispute; the dis- determinism, and that compatibilism with respect to a given
pute seems to turn simply on what—according to socio-biology type of determinism is possible only for people who do not
—we have a choice about, and not on any questions about the take that type of determinism very seriously. 32
validity of inferences involving 'having a choice about'. People I said above that I could think of no instances of ((3) that
who accept, or are accused of accepting, "special determin- had, or could possibly have had, true premisses and a false
isms"—that is, theories that say, or are sometimes interpreted conclusion. I meant, of course, that I could think of no in-
stances of (0) that could be seen to have true premisses and
as saying, that some important aspect of human behaviour is
determined by this or that factor outside our control—tend a false conclusion independently of the question whether free
to find themselves embroiled in disputes about the freedom will is compatible with determinism. If free will is compatible
of the will. I have used a socio-biologist as an example of such with determinism, and if determinism is true, then, presum-
a person because I am writing now, but the "biology is des- ably, at least one of the following two instances of ((3) has true
tiny" debate is not the only one of its type. If I had been premisses and a false conclusion:
writing a few years ago, I should have constructed an example N (P. D (L D Nixon received a pardon))
102 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 103

N Po You yourself admit that the conclusion of this argument

follows from its premiss." (I do.) "You may not accept its
hence, N (L D Nixon received a pardon). premiss, but that's your problem, for that premiss is true.
Moreover, you can hardly object to this little argument of
N (L 3 Nixon received a pardon) mine on the ground that it begs the question. It's no worse
NL in that respect than your argument, which is essentially this:
hence, N Nixon received a pardon. ((3) is valid
But it would be nice to see a counter-example to 43) that did hence, Compatibilism is false."
not presuppose the compatibility of free will and determinism.
After all, the examples I gave in support of 03) did not pre- What am I to say to this? I suppose I can do no more than
suppose the incompatibility of free will and determinism. I appeal to the intuitions of my readers. Here's how it looks to
should think that if there are any counter-examples to (J3), me (and doesn't it look this way to you?); Rule ((3) seems
then some of them, at least, could be shown to be such in- obviously right and compatibilism does not seem obviously
dependently of the question whether free will and determin- right. If two principles are in conflict and one of them seems
ism are compatible. obviously right and the other does not seem obviously right,
It may be hard to credit, but there are almost certainly then, if one must choose, one should accept the one that seems
philosophers who would say that the fact that determinism obviously right.
and compatibilism together entail that there are counter- Well, suppose the compatibilist says he finds compatibilism
examples to ((3) shows that my use of ((3) in an argument for obviously right. Presumably this assertion is either grounded
incompatibilism is question-begging. But if this accusation in an immediate and intuitive relationship to compatibilism-
were right, it's hard to see how any argument could avoid he claims to see that it's true, just as I claim to see that 03) is
begging the question. If one presents an argument for a cer- valid—or it is grounded in some argument for compatibilism.
tain proposition, then, if that proposition is false, some step Let us first look at the case of the philosopher who claims to
in the argument is wrong (I am counting the use of a false see the truth of compatibilism intuitively. Arguments, like ex-
premiss as a "wrong step"); and one may believe of a parti- planations, must come to an end somewhere. Perhaps, if there
cular step in the argument that if any step is wrong, that one is such a philosopher, he and I constitute a genuine case of a
is. But it hardly follows that one is "begging the question" conflict of rock-bottom intuitions. But I must say I should
by taking that step. One may be begging the question—what- find any such claim as the one I have imagined incredible.
ever, precisely, begging the question is—but that one is begging Compatibilism looks to me like the kind of thing one could
the question is not a consequence of the mere existence of a believe only because one had an argument for it. I simply
"weakest link" in one's chain of reasoning. cannot see what could be going on in the mind of someone
Now these questions about question-begging and where the who claimed to know it intuitively. I don't know what that
burden of proof lies, and so on, are very tricky. Let's look at would feel like.
them from a different angle. Suppose a compatibilist were to The philosopher who accepts compatibilism on the basis of
say something like this: "The Third Argument depends on the an argument is not likewise mysterious to me. But I shall want
validity of ((3). But this rule is invalid. I prove this as follows: to know what the premisses of his argument are. And I shall
raise the following question about his premisses: Are they
Compatibilism is true really intuitively more plausible than ((3)? I find it hard to be-
hence, 03) is invalid. lieve that there are any propositions that entail compatibilism
104 THREE ARGUMENTS FOR INCOMPATIBILISM THREE ARGUMENTS FOR INCOMPATIBILISM 105
that are more plausible than (13). Still, who knows? Perhaps If there is anything to the objection we are considering, then
there are. What can we do but proceed by cases? In Section at least one of the four propositions,
4.3, we shall compare (13) with the premisses of a popular
argument for compatibilism, the argument I called the "Con-
ditional Analysis Argument" in Chapter I." NL,
This completes my presentation of the Third Argument and Rule (a) is valid,
my defence of its premisses. I have now given three arguments
Rule (13) is valid,
for the incompatibility of free will and determinism. If the
compatibilist wishes to refute these arguments—and, of course, is false, given that 'N'—which occurs in (a) and (0—is inter-
nothing obliges him to do this—here is what he will have to do: preted as above. But this does not seem to be the case. If one
he will have to produce some set of propositions intuitively carefully retraces the steps of the Third Argument, one will
more plausible than the premisses of these arguments and show find, I think, that no step becomes doubtful under our new
that these propositions entail compatibilism, or else he will interpretation of 'N'. The conclusion of the present chapter is,
have to devise arguments for the falsity of some of the pre- therefore, that if determinism is true, then no one has any
misses employed in the present chapter, arguments that can be choice about anything, in just that sense of having a choice
evaluated and seen to be sound independently of the question that is relevant in debates about moral responsibility.
whether free will and determinism are compatible.

3.11 Some compatibilists, when they are confronted with

arguments for the incompatibility of free will and determinism,
say something like this: "Your argument simply demonstrates
that when you use phrases like 'could have done otherwise'
or 'has a choice about', you are giving them some meaning
other than the meaning they have in our actual debates about
moral responsibility". This criticism is equally applicable,
mutatis mutandis, to all three of our arguments for incom-
patibilism. And my answer to it is essentially the same in each
case. But this answer can be presented very compactly and
efficiently in terms of the vocabulary employed in the Third
Argument. Therefore, I shall answer only this charge: "When
you use the phrase 'has a choice' you are giving it a meaning
different from the meaning it has in our actual debates about
moral responsibility", and I will leave to the reader the mech-
anical task of adapting this answer to the requirements of the
First and Second Arguments. 34 My answer consists simply in
a reinterpretation of 'N':
Np= df p and, in just the sense of having a choice that is
relevant in debates about moral responsibility, no one has,
or ever had, any choice about whether p.
 THREE ARGUMENTS FOR COMPATIBILISM 107

considerations, we may represent part of his argument as

follows:
Chapter IV There are various words and phrases we use in ascribing
free action to people: besides the obvious 'acted freely'
Three Arguments for Compatibilism and 'did it of his own free will', there are such phrases as
`could have done otherwise', 'had a choice about what
she did', 'had alternatives', and 'could have helped doing
4.1 If determinism is true, no one can act otherwise than he what he did'. We learn these phrases by watching people
does. If determinism is true, no one has it within his power to apply them in concrete situations in everyday life, just
realize any possibility that is in fact unrealized. If determin- as we learn, for example, colour words. These concrete
ism is true, no one has any choice about anything. In short, if situations serve as paradigms for the application of these
determinism is true, there is no free will. This was the con- words: the words mean things of that sort. Therefore
clusion of Chapter III. they must apply to something; they must apply at least
In the present chapter, I shall examine arguments for the to the paradigmatic objects or situations. Careful investi-
conclusion that free will and determinism are compatible. I gation, philosophical or scientific, of these situations may
shall not examine arguments that turn on the contention that indeed yield information about what freedom of choice
a belief in the incompatibility of free will and determinism really consists in, but it cannot show us that there is no
rests on some confusion or other—say, a confusion of "des- such thing as freedom of choice. This is strictly parallel
criptive" with "prescriptive" laws, or a confusion of caus- to the following proposition: careful investigation, philo-
ation with compulsion. My belief in the incompatibility of sophical or scientific, may show us what colour really
free will and determinism rests on the arguments of Chapter consists in, but it cannot show us that there is no such
III, and these arguments involve no such confusion. I shall thing as colour.
examine only positive arguments for compatibilism, arguments
Well, so far so good. But this argument is not an argument for
that purport to establish compatibilism and not merely to
expose an error that might lead one to believe in incompati- the compatibility of free will and determinism, though it may
be part of such an argument; rather, it is an argument for the
bilism. existence of free will. There seem to be two arguments with
I know of only three positive arguments for compatibilism
that have any currency. I shall call them 'the Paradigm Case which Flew supplements the above argument in order to show
Argument', 'the Conditional Analysis Argument', and 'the that free will not only exists, which many incompatibilists
will grant, but is also compatible with determinism.
Mind Argument'. I have given brief descriptions of these argu-
ments in Section 1.6. (a) When we do carefully investigate the paradigm cases of
free action, we find that their common feature is just this: we
4.2 The classical statement of the Paradigm Case Argument apply the word 'free' to a person's act just in the case that, "if
is due to Antony Flew.' Flew is primarily concerned to show he had chosen to do otherwise, he would have been able to
that the proposition that we always freely do what is right is do so; that there were alternatives, within the capacity of one
compatible with the proposition that God so arranged things of his physical strength, of his IQ, with his knowledge, and
at the first moment of creation that our always doing what is open to a person in his situation". 2 And, of course, a person's
right is the inevitable consequence of the initial arrangement. act may have this feature whether or not determinism is true.
But Flew explicitly says that his argument shows that freedom (b) We have seen that scientific investigation could not show
and determinism are compatible. If we leave aside theological that none of us has free will. But scientific investigation could
108 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 109

show that determinism is true. Hence, if free will were incom- be the argument that (i) purports to establish the existence of
patible with determinism, scientific investigation could show actual occasions of free action by an appeal to the necessary
that we have no free will, which is impossible. correctness of applying the term 'a free act' to those acts that
In order the better to see the logical relationship between are used to teach the meaning of this term, and (ii) proceeds,
these two supplementary arguments, let us look at two parallel by means of argument (b), to derive from the possibility of
and familiar arguments about colours. Let us give the label so establishing the existence of free action the conclusion that
`hypochromatism' to the thesis that visible objects are com- free will and determinism are compatible.
posed entirely of tiny objects that are themselves colourless. In calling this argument (part of) Flew's argument, I am
Some philosophers have thought that hypochromatism is in- interpreting Flew. There is a great deal in his paper that I do
compatible with the thesis that ordinary visible objects are not understand, and I may simply have got him wrong. In any
coloured. But we may refute them as follows: case, this argument is the clearest and most cogent argument I
can find, or think I can find, in Flew's paper. If it is not Flew's
(i) When we carefully investigate the paradigm cases of argument, perhaps it is sufficiently similar to his argument
coloured objects, we find that their common feature is just that what I say in the sequel will be relevant to his argument.
this: we apply the word 'coloured' to an object just in the case The argument I have attributed to Flew is invalid. It is ob-
that, if that object were presented to a normal observer under viously possible to think of propositions that are consistent
standard conditions of observation, it would cause him to ex- with all our observations and which have the consequence that
perience colour-sensations. And, of course, an object may have no one can do otherwise than he does. Here is a fanciful but
this feature whether or not hypochromatism is true. logically adequate example.
(ii) We have seen that scientific investigation could not
show that no objects have colours. But scientific investigation (M )
When any human being is born, the Martians implant in
could show that hypochromatism is true. Hence if the exist- his brain a tiny device—one that is undetectable by any
ence of coloured objects were incompatible with hypochro- observational technique we have at our disposal, though
matism, scientific investigation could show that no objects it is not in principle undetectable—which contains a
have colours, which is impossible. "program" for that person's entire life: whenever that
person must make a decision, the device causes him to
Obviously, arguments (a) and (b) have mutually consistent decide one way or the other according to the require-
sets of premisses, and there is no reason why Flew should not ments of a table of instructions that were incorporated
employ both. Argument (a), however, is essentially what I into the structure of the device before that person was
have called the "Conditional Analysis Argument". The thesis conceived.'
that 'he could have done otherwise' means something like 'if
he had chosen to do otherwise, he would have been able to' is Now someone might object that (M) is not in fact consistent
frequently employed as a premiss by compatibilists who would with our observations, since we can normally "feel" our
reject any appeal to paradigm cases in philosophical argument. decisions "flowing" naturally from our desires and our beliefs;
Moreover, if this thesis is correct, then it can be used to show but if (M) were true (so the objection runs), we should "feel"
"directly" that free will is compatible with determinism, that ourselves being interfered with. But to meet this objection we
is, it can be used to demonstrate this compatibility without need only suppose that the Martian device causes us to have
the use of paradigm cases—or any other device—first to estab- desires and beliefs appropriate to the decisions it will cause us
lish that there actually are occasions on which people act freely. to make. (If Davidson is right, the device could cause our
I shall discuss the Conditional Analysis Argument in the next decisions simply by causing us to have certain desires and
section. Let us understand "the Paradigm Case Argument" to beliefs. It makes no difference for our purposes whether this
110 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 111

is so or whether, having caused our desires and beliefs, the the following features: (i) determinism belongs to C and (M)
device must then do something further to cause our decisions: does not, and (ii) propositions belonging to C can validly be
it does whatever is necessary to cause our decisions.) But if shown to be compatible with free will by some style of argu-
the Paradigm Case Argument were valid, then it would follow ment that is recognizable as an appeal to paradigm cases, and
that (M) was compatible with free will. If (M) were true, the propositions not belonging to C cannot be shown to be com-
world would look the way it in fact looks, and our linguistic patible with free will by this style of argument.
practices would be the same—or, at least, we should emit the I do not think this can be done. There are, of course, all
same sounds in the same situations, and our production of sorts of important differences between determinism and (M).
these sounds and their reception by our audience would be (M) entails that each human being's dispositions to act are
accompanied by the same internal sensations. chosen for him by a non-human intelligence, and determinism
If the Paradigm Case Argument is valid, only such super- does not. (M) entails that each human being's acts have among
ficial features of people and their acts as these can be relevant their immediate causes an object which, though it is inside
to determining whether to apply the term 'free': one must be that person's skull, is not a natural part of him, and determin-
able effectively to compare some present act in which one is ism does not entail this. These are indeed differences between
interested with the paradigms of free action that one remem- (M) and determinism that are of the highest importance: they
bers in order to see whether the present act is sufficiently are what makes it evident that (M) is incompatible with free
similar to the remembered paradigms to be correctly called will, while the incompatibility of determinism with free will
`free'. And, obviously, one could never make an effective com- is a matter of debate. What is hard to see is the relevance of
parison of the present act with the paradigm if the relevant these incontestably important facts to the Paradigm Case
features of the paradigm were unobservable. Thus, if the Para- Argument. Indeed, it is hard to see how, if the Paradigm Case
digm Case Argument were correct, the extension of the term Argument is correct, any features of a given proposition could
`free' could be just what it in fact is even if (M) were true, be relevant to the question whether that proposition is com-
since (M) does not require any observable features of the world patible with free will, other than those of its features that
to be different from what they are in actuality. But (M) ob- would make a difference in how things appear to us. But there
viously does entail that no one can act otherwise than he just obviously are other features of a proposition than these
does. If we should discover that some particular person- that are relevant to whether that proposition is compatible
Himmler, say—acted as he did because a Martian device, im- with free will, or, indeed, compatible with any given propo-
planted in his brain at the moment of his birth, had caused all sition. It is possible to exploit this fact to construct propo-
his decisions, then we should hardly want to say that Himmler sitions, which, while they are consistent with all of our
had free will, that he could have helped what he did, that he observations, are inconsistent with the thesis that we have free
had any choice about the way he acted, or that he ever could will. This is how we constructed (M). The fact that we can
have done otherwise. And I don't see why matters should be construct such propositions shows that it is at least possible
different if we discovered that everyone was "directed" by a that certain propositions that we know of but have not our-
Martian device: then we should have to make these judgements selves constructed are also consistent with our observations
about everyone.' but inconsistent with the free-will thesis. Perhaps determinism
If the proponent of the Paradigm Case Argument is to meet is such a proposition.
this objection, then, it seems to me, he must show that there There seems to be no reason to think this implausible. Deter-
is some difference between (M) and determinism that he can minism is a very general thesis, and it essentially involves con-
exploit in constructing an improved version of his argument. cepts (like 'law of nature') that are intimately connected with
That is, he must delimit some class of propositions C having the concept of ability (since no one is able to change the laws
112 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 113

of nature). Mightn't it simply turn out that determinism and I would point out that some philosophers have believed that
free will are incompatible? How can we be sure that this isn't determinism is a necessary truth) or by restricting it in some
what careful investigation will show? I believe that this is what other way. It would be less easy to restrict the principle so that
careful investigation has shown. (M) is not a counter-example to it—which it is—but no doubt
There are certain arguments for the compatibility of free this could be done. I feel reasonably confident, however, that
will and determinism that are sufficiently similar to the Para- for any revision of the principle, I can find a counter-example
digm Case Argument that they too can be seen to be invalid to the revised principle that can be seen to be a counter-exam-
by reflection on their implications for the relation between ple independently of the question whether 'he could have done
free will and propositions like (M). That is, reasoning very otherwise' is compatible with determinism. Or, at least, I feel
much like the above reasoning can be used to show that if confident of my ability to do this up to the point at which
these arguments succeeded in establishing that free will was the class of general theses to which the revised principle is
compatible with determinism, then similar arguments would supposed to apply has been so restricted that the principle
succeed in establishing the absurd conclusion that free will has become no more than a shamelessly transparent statement
was compatible with (M). of compatibilism. However this may be, the argument in its
Consider, for example, the following argument: present form is obviously invalid, since, if it were valid, an
almost exactly similar argument could be used to show some-
Our everyday ascriptions of the ability to act otherwise thing false: that (M) is compatible with free will.
make no reference to determinism. That is, we do not A second argument that can be seen to be invalid owing to
find out whether an agent's act was undetermined by the fact that its validity would entail the compatibility of (M)
past events in order to find out whether he could have and free will is due to Keith Lehrer.' Lehrer points out that
acted otherwise. Therefore, the thesis normally expressed it would be easy to collect evidence on which the hypothesis
by 'he could have acted otherwise' does not entail the that a certain agent could have acted otherwise is highly prob-
falsity of determinism. able. This evidence might, for example, consist of a large body
I am chagrined to report that I am unable to find a clear exam- of propositions each asserting that in very similar circum-
ple of this argument in print. But I have heard it many times stances in the past, the agent has acted otherwise than he did
in conversation and read it many times in correspondence. I upon this particular occasion. It is this sort of evidence, after
believe, therefore, that I am justified in devoting a bit of space all, upon which everyone, compatibilist or incompatibilist,
to it. It is not absolutely clear to me how this argument is does base his judgements about people's abilities to act other-
supposed to work, but the most obvious way to read it is as wise. But such evidence as this clearly does not render indeter-
an application of the rule of Universal Instantiation to the minism highly probable. It would be absurd to argue like this:
following premiss: "In similar circumstances, Tom has often acted otherwise
than he did upon this particular occasion, and it is therefore
The thesis normally expressed by 'he could have acted highly probable that determinism is false". Now consider the
otherwise' does not entail the falsity of any general thesis following theorem of the calculus of probabilities:
to which our everyday ascriptions of the ability to act
otherwise make no reference. If H, is highly probable on e and H2 is not highly prob-
able on e, then H 1 does not entail H2.
This principle is false. It suffices to point out that either
Goldbach's\ Conjecture or its denial is a counter-example to it, It follows from this theorem and what we have established
but such counter-examples could be evaded by replacing this above that free will does not entail indeterminism; that is, that
principle with oni-that refers only to contingent truths (though free will and determinism are compatible.
114 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 115

However plausible or implausible this argument may seem One point on which conditionalists differ is that of the
at first, it is exactly as plausible as the argument that can be proper content for the antecedents of their conditionals. Where
got by substituting `(M)' for 'determinism' in it. And that one conditionalist will say, "If Smith had chosen . .", others
argument has an obviously false conclusion.' will say, ". . . had willed .. .", ". . . had decided . ..", "... had
The general lesson of the present section is this. There are set himself . . .", or ". . . had tried ...". I shall discuss only cer-
theses that are incompatible with free will. The philosopher tain "had chosen" and "had wanted" versions of conditional-
who attempts to show that free will and determinism are com- ism explicitly. But what I say will apply mutatis mutandis to
patible should make sure that his arguments cannot be used versions involving other verbs.
to "show" that some thesis that is just obviously incompatible If an adequate conditional analysis of ascriptions of ability
with free will is compatible with it. is possible, it will assign at least some of them conditional
paraphrases of rather more complicated forms than the form
4.3 In this section, I shall examine an argument for the com- exemplified by the above paraphrase of the "drowning child"
patibility of free will and determinism that is accepted, in proposition. For example, it could hardly be true that
one form or another, by the great majority of the present-day Napoleon could have won at Waterloo
defenders of compatibilism. This argument rests on a theory
about the meaning of ascriptions of ability. According to pro- really means
ponents of this theory, which I shall call conditionalism, ascrip-
tions of ability are really disguised conditionals. For example, If Napoleon had chosen to win at Waterloo, Napoleon
according to one version of conditionalism, what the propo- would have won at Waterloo!
sition I shall, however, confine my discussion to questions about our
Smith could have saved the drowning child abilities to perform acts that do not involve the execution of
"really means" is elaborate plans or demand special knowledge or skill. Thus I
can concentrate on situations in which relatively simple con-
If Smith had chosen to save the drowning child, Smith ditional analyses are adequate if any conditional analyses are.
would have saved the drowning child. Even in these simple cases, however, it is doubtful whether
(If this contention is correct, then the proposition that Smith any very simple conditional analysis of ability is correct. R. M.
could have saved the drowning child but did not is obviously Chisholm has called our attention to a fundamental difficulty
no more incompatible with determinism than is the propo- facing conditionalism." An example of Keith Lehrer's nicely
sition that this sugar cube would dissolve if, contrary to fact, illustrates this difficulty.' Consider the proposition:
it were placed in water.)
Conditionalism comes in a great many varieties. In one Smith could have eaten one of the red candies.
rather primitive variety, it entails that 'Smith could have saved This proposition is not equivalent to
the drowning child' means
If Smith had chosen to save the drowning child, Smith If Smith had chosen to eat one of the red candies, then
could have saved the drowning child. Smith would have eaten one of the red candies.
This version of conditionalism, however, faces a great many For suppose that Smith is pathologically afraid of the sight of
difficulties that I do not propose to enumerate.' The only blood, and that the candies are the colour of blood. Then it
interesting varieties of conditionalism, I think, are the "would may well be that Smith was unable to choose to eat one of
have" varieties. the red candies. And, in that case, he could not have eaten one
116 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 117

of the red candies. Nevertheless, we may suppose, if he had to denote the required "new" sense of 'could have', and use
chosen to eat one of the candies, he would have. `could have' only in its power-or-ability sense. The conditional
A natural reaction to this example would be to add the analysis may now be written:
clause x could have done y = d f if x had chosen to do y, x would
. . . and Smith could have chosen to eat one of the red have done y, and x COULD HAVE chosen to do y.' °
candies Whatever 'COULD HAVE' may mean exactly, there would
to the conditional proposition. But what does this second seem to be a condition that must be met by any adequate
`could have' mean? Either it means the same as the 'could account of its meaning: 'x COULD HAVE chosen to do y'
have' in 'Smith could have eaten one of the red candies' or must entail 'x could have chosen to do y'. If this condition
it means something else. If it means the same, the emended is not met, then it will be easy enough to produce a Chisholm-
definition will be of no help to us, since the question we are style counter-example to this definition. Here is the recipe.
considering is a question about the relation between 'could Simply pick one of those logically possible cases in which
have' statements and determinism. Moreover, if the second someone COULD HAVE chosen to do something, but could
`could have' means the same as the first, and if the expanded not have chosen to do it (and, of course, construct the example
conditional proposition really is equivalent to the proposition in such a way that his choice would have been effective). This
that Smith could have eaten one of the red candies, then the will be a case in which our imaginary person could not have
latter is equivalent to the proposition performed a certain act, though, according to the proposed
definition, he could have. I do not myself see any way to de-
If Smith had chosen to eat one of the red candies, then fine 'COULD HAVE' that will meet this condition. Moreover
Smith would have eaten one of the red candies, and if if I knew how to do this—if I knew how to give 'COULD
Smith had chosen to choose to eat one of the red candies, HAVE' a sense such that 'COULD HAVE chosen' entails 'could
then Smith would have chosen to eat one of the red can- have chosen'—it seems likely that I should be able to give
dies, and Smith could have chosen to choose to eat one `COULD HAVE' a sense such that 'COULD HAVE' tout court
of the red candies. entails 'could have' tout court. And if I could do that, then
Not only does the monstrous sentence I have just written out why should I not be able to apply this insight "directly", that
contain an unreduced 'could have' like its "predecessor" and is, without making any use of the conditional that is the first
like all its ever more monstrous successors, but it contains the conjunct of the definiens of the conditional analysis?
mysterious clause 'if Smith had chosen to choose to eat one There is a way for the conditionalist to avoid all these dif-
of the red candies'. But what is it to choose to choose some- ficult problems. An act of choice is, of course, an act," and
thing? One can, of course, choose to choose between two or we may therefore raise the question whether its agent could
more things: in choosing to drink wine, I may in effect be have refrained from performing it, thus precipitating the un-
choosing to choose between drinking claret and drinking bur- happy infinite regress whose consequences we have been
gundy. But that is not to choose to choose something; what exploring in the preceding paragraphs. But suppose a con-
would count as a case of choosing to choose claret? Nothing ditionalist were to offer an analysis of 'could have' of this
that I can see. I think we must abandon the hypothesis that form:
the second 'could have' means the same as the original, for x could have done y =df if Rxy, then x would have done
this hypothesis has led us into incoherency. y, and A,
But if the second 'could have' is not the 'could have' of
power or ability, just what is it? Let us use 'COULD HAVE' where `Rxy' represents some condition on x and y such that
118 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 119

it is possible that this condition hold without x's doing any- temptation to cover the dish of candies with a napkin. Now
thing antecedent to his doing y. The second conjunct of the suppose he did want to eat one of the candies. Presumably he
definiens—represented in this schema by `A'—may be any would. But also, presumably, he would not be afflicted with
sort of qualification. If the antecedent of the first conjunct his curious neurosis, a neurosis which, in actuality, renders
of our definiens can be true even if x does nothing, then we him unable to eat one of the candies. So it would seem that
shall not be able to raise Chisholm-style objections to this both the propositions
definition. One very simple definiens of this type is If Smith had wanted to eat a red candy, he would have
if x had wanted to do y (more than anything else), then and
x would have done y.
-
Smith could not have eaten a red candy
Wanting to perform an act (unlike choosing to perform an are true. Therefore, the analysis we are considering fails. The
act) is not itself an act. Therefore, if someone says, for exam- general lesson is that if a certain agent had had desires he did
ple, "If he had wanted to eat a red candy, then he would have not in fact have, then his powers to act might also have been
eaten one", we may not say in response, "Yes, but could he different, a possibility that vitiates the analysis we are con-
have wanted a red candy?" Or, more precisely, we could so sidering. Here is a simpler case. Smith, considerably shaken
respond, but our words would have to mean something like by his experience with the candies, wanders out into traffic, is
"was it possible—in some unspecified sense of `possible'—in struck by a bus, and is taken to a hospital in a deep coma from
that situation that he should have wanted a red candy?" If it which he never emerges. Consider poor Smith, comatose in
was not possible that he should have wanted to eat a red candy, his hospital bed. The two propositions
this does not seem to imply what 'it was not within his power
to choose to eat a red candy' does seem to imply: that it was Smith cannot get out of bed
not within his power to eat a red candy. Suppose, for example, If Smith wanted to get out of bed, he would
that Jones has been given a drug that makes one indifferent
to food of any sort—not revolted by the thought of food, but would seem both to be true, the former because he is in a
simply wholly indifferent to eating. It seems to be correct to coma, and the latter because, if he did want to get out of bed
say (in some sense of 'possible') that it was not possible that, he wouldn't be in a coma." (We must also suppose, of course,
in that situation, Jones should have wanted to eat a red candy. that nothing keeps him from getting out of bed but uncon-
Yet it does not seem to follow that Jones could not have sciousness, a state which his wanting anything would suffice
eaten one, that it was not within his power to eat one. After for the absence of.)
all, people sometimes do what they have no particular desire This sort of example does not show that the sort of analysis
to do. we are considering is hopeless, but only that its definiens must
But if the definiens we are considering is immune to contain some qualification, some further condition to be in-
Chisholm-style counter-examples, it nevertheless faces prob- serted in the space occupied by 'A' in the abstract schema
lems that are not entirely dissimilar to the problems that such above.
counter-examples created for the "choice" version of con- What should this qualification be? The best answer to this
ditionalism. Let us look once more at the "red candy" case. question I can think of is suggested by an analysis of 'could
Note that Smith—the man who has a pathological aversion to have' devised by Keith Lehrer." If we incorporate Lehrer's
the colour of blood, and, per accidens, to the colour of the idea into the analysis we are considering, we obtain:
candies—doesn't want to eat a red candy. The very sight of x could have done y =df if x had wanted to do y, x would
them, we may suppose, makes him uneasy: he feels a strong have done y and it is false that if x had wanted to do y,
120 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 121

x would have possessed some advantage with respect to ability, or one of them if there is a tie: let us suppose we have
doing y that x did not actually possess." found an extensionally adequate analysis if any such exists,
This analysis does not have the consequence that Smith (our or, if none exists, have found one that can't be improved by
neurotic) could have eaten a red candy. For, though it is true tinkering or by the addition of further clauses and is not mark-
that Smith would have eaten a red candy if he had wanted to, edly farther from extensional adequacy than any fundamen-
it is also true that if he had wanted to eat a red candy, he tally different analysis. Call it the Analysis. The Analysis is
would have possessed an advantage with respect to eating a right if any conditional analysis is right, and wrong only if
red candy that he did not in fact possess: freedom from his no conditional analysis is right.
peculiar neurosis. Similarly, if Smith had wanted to get out What does the Analysis do for us? How does it affect our
of his hospital bed, he would have possessed an advantage understanding of the Compatibility Problem? It does very little
with respect to getting out of bed that he did not in fact pos- for us, so far as I can see, unless we have some reason to think
it is correct. Many compatibilists seem to think that they
sess: consciousness.
But this analysis rules out obvious cases of ability. If I need only present a conditional analysis of ability, defend it
wanted to fly to Washington, I should (come to) possess a against, or modify it in the face of, such counter-examples as
certain advantage with respect to flying to Washington: a may arise, and that they have thereby done what is necessary
reserved seat on a Washington-bound flight; and this advan- to defend compatibilism.
tage is one I do not actually possess. But this fact obviously That is not how I see it. The particular analysis of ability that
does not entail that I could not fly to Washington. I am not a compatibilist presents is, as I see it, simply one of his prem-
sure how the above definiens ought to be rewritten to avoid isses; his central premiss, in fact." And premisses need to be
this difficulty. I tentatively suggest this: defended. I see no particular reason to think that statements
of ability can be analysed as conditionals, much less that any
if x had wanted to do y, x would have done y and x's particular conditional analysis of such statements is correct.
wanting to do y would not have been sufficient (in the Moreover, I see what seems to me to be a good reason to think
"broadly logical" sense) for x's possessing an advantage that no conditional analysis is correct—or, at least, that no con-
with respect to doing y that x did not actually possess. ditional analysis that supports compatibilism is correct. If
If I wanted to fly to Washington this would perhaps be causally any conditional analysis is correct then ex hypothesi the Analy-
sufficient for my acquiring a reserved seat on a flight to sis is. Let us consider the Analysis in relation to any of the argu-
Washington, but it would not have been logically sufficient. ments for incompatibilism that were presented in Chapter III.
But if Smith had wanted to get out of bed, this would have For the sake of concreteness, let us consider the First Argu-
been logically sufficient, in the broad sense, for his being con- ment. The First Formal Argument is valid. Therefore, if the
scious. It is not so clear whether Smith's wanting to eat a red Analysis is correct, at least one of the premisses of the First
candy would be logically sufficient for the removal or non- Formal Argument is false. (I assume this conditional is true. If
existence of his pathological aversion to red things, but let us it isn't, of course, then free will and determinism are incompat-
suppose this is so. Better, let us suppose that we have found ible even if the Analysis is correct.") Conversely, if all the prem-
some sense of sufficiency—certainly stronger than causal suf- isses of the First Formal Argument are true, the Analysis is
ficiency, possibly weaker than broadly logical sufficiency— incorrect. Therefore, since I have presented arguments for each
that is "best" for our present purposes, that either makes our of the premisses of the First Formal Argument—arguments that
definition extensionally adequate or at least comes as close to do not in any clear sense presuppose that statements of ability
satisfying this requirement as possible. Better still, let us sup- are not disguised conditionals—I have in effect presented an
pose we have found the best possible conditional analysis of argument for the conclusion that the Analysis fails."
122 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 123

Now if we had at our disposal some argument for the thesis implausible to suppose that, whereas simpler conditional analy-
that the Analysis was correct, we could compare this argu- ses of ability are incompatible with the validity of (3) and
ment with my arguments for the premisses of the First Formal apparently compatible with the other premisses of the Third
Argument and try to determine which side had the stronger Argument, the Analysis is compatible with (3) but incom-
case. But, unless there is something I'm missing in the many patible with some other premiss.
"conditional" defences of compatibilism to be found in the I will content myself with showing that the simplest analysis
philosophical literature, no argument has ever been given in we have considered entails the invalidity of (j3). I leave it as
defence of the thesis that statements ascribing to an agent an exercise for the reader to show that the several more com-
the power to act can be correctly analysed as conditionals. plex revisions of this simple analysis that we have considered
It's not that this thesis seems inherently implausible to me. If also have this consequence.
I consider it in isolation from the Compatibility Problem, I Rule (i3), it will be remembered, is this: Np,N(pD q) Nq,
feel no inclination either to accept or to reject it. But when I where ‘Np' abbreviates 'p and no one has, or ever had, any
consider the fact that it is incompatible with certain propo- choice about whether p'. If our first conditional analysis of
sitions—the premisses of the First Formal Argument, for exam- ability is correct, then `Np' is equivalent to something like the
ple—that seem to me to be highly plausible right at the outset, following:
then, of course, I feel very strongly inclined to reject it indeed. p and, for every agent x, it is not the case that if x had
These considerations are very abstract. They can be pre- ever chosen (or were ever to choose) to make it false that
sented in a form that is more concrete. It is possible to show p, then x would have made (or would make) it false that p.
exactly where the Analysis comes into conflict with our in-
compatibilist arguments, to show just which premiss of each Let us suppose that this equivalence holds. Now consider the
of the arguments must be false if the Analysis is true. And if following instance of (3):
we have located the premiss of any of these arguments that, N It is psychologically impossible for Smith to choose
from the compatibilist's point of view, is the crucial premiss, to eat a red candy;
then we may compare that premiss and the Analysis and see
which seems to be the more plausible. (Of course we don't N (It is psychologically impossible for Smith to choose
really know what the Analysis is. Nevertheless, we know vari- to eat a red candy D Smith does not eat a red candy);
ous things about it. I shall try to show that we know enough.) hence, N Smith does not eat a red candy.
It will be most convenient to apply this strategy to the
Third Argument. I said in Chapter III that the crucial premiss, We may easily imagine the premisses of this argument to be
the premiss most in need of support, of the Third Argument true and its conclusion false. Let us take our story of the neur-
was the validity of Rule (13). It therefore seems likely that if otic Smith and add to it the following conditions. Suppose,
the Analysis is correct, Rule (/3) is invalid. This speculation just to simplify the example, that Smith is the only person.
can be shown to be correct.' 8 Or, at any rate, it can be shown Suppose that if Smith were to choose to make it false that it
that Rule (3) is incompatible with the conditional analyses was psychologically impossible for him to choose to eat a red
we have actually stated, and it therefore seems likely that if candy, he would fail. Then the first premiss of our argument
the Analysis—whatever precisely it may be—is incompatible is true. Moreover, let us suppose that something renders it
with any of the premisses of the Third Argument, it is incom- impossible for Smith to eat a red candy without choosing to
patible with the validity of (13). If the Analysis is incompatible —perhaps a mechanism outside his control snatches the can-
with no premisses of the Third Argument, then incompati- dies away if he inadvertently reaches for one, or something
bilism is true even if the Analysis is correct. And it is highly of that sort. Then, since no one does by choice what it is
124 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 125

psychologically impossible for him to choose to do, the second it is capable of grasping the average candidate for the "fourth
premiss is true. Suppose that if—per impossibile psychologicale condition" of knowledge. One cannot—I cannot, at any rate—
—Smith were to choose to eat a red candy, then he would eat see them steadily and see them whole. Therefore, in making
a red candy. Then the conclusion is false. Since these suppo- a choice between Rule (p) and the Analysis, I feel no qualms
sitions seem consistent, we have devised a counter-example to of conscience about applying the epistemic rule-of-thumb that
Rule (13) that holds provided that ‘Np' is interpreted according was introduced in Section 3.10: Rule (13) seems obviously
to the terms of the first conditional analysis. So much the correct; the Analysis, owing to its extreme complexity if to
worse for the first conditional analysis, I say. The thesis that nothing else, does not seem obviously correct; therefore, I
Rule (3) is valid seems to me to be obviously true and the first accept Rule (/) and reject the Analysis. And for this reason
conditional analysis does not seem to me to be obviously true. I reject the Conditional Analysis Argument.
If two propositions are incompatible and one seems obviously If someone produced a convincing counter-example to (/),
true, then, as I said in Chapter III, if one must choose one this would vitiate my argument. If someone produced a plaus-
should accept the proposition that seems obviously true. ible argument for the conclusion that (13) is invalid, this would
We have been talking about the first conditional analysis, vitiate my argument. If someone produced a simple, easily
which is known to be defective on other grounds than its in- graspable conditional analysis of ability, and if this analysis
compatibility with Rule (13). What about the Analysis itself? seemed right and if there were no known counter-examples
Well, as I have argued, it is certainly reasonable to believe that to it, this would vitiate my argument. If someone produced a
the Analysis is incompatible with the validity of (/). (And if plausible argument for the correctness of some conditional
it's not, the compatibilist is in deep trouble: if it's not, then analysis—however complex that analysis might be—this would
either the Analysis is incompatible with some premiss of the vitiate my argument. But none of these things has ever been
Third Argument that is even more evidently true than the produced and I do not think that any of them is ever going to
proposition that (0) is valid, or else the Third Argument is be produced. I conclude that since Rule (i3) is more plausible
sound even if the Analysis is correct.) Which is it more reason- than the Analysis and, a fortiori, more plausible than any
able to believe: that Rule (0) is valid or that the Analysis is other conditional analysis of ability, the Conditional Analysis
correct? I believe I know enough about the Analysis to answer Argument fails.
this question with some confidence. The Analysis, whatever I gather that most compatibilists think that compatibilism
its exact content may be, will almost certainly look something is an inherently plausible thesis and that incompatibilism is
like the analysis in terms of "advantage" that we examined inherently implausible. If someone thinks this, perhaps he will
earlier in the present section. Even if it does not make use of regard the argument
the same "idea", it will at least be of a comparable level of Compatibilism is true
complexity. When I look at philosophical analyses that are
that complex, then, unless I have been shown some convincing hence, (0) is invalid
argument for their correctness, I simply have no idea whether as a "plausible argument for the conclusion that (i3) is invalid".
they are correct. My reaction to them could be expressed in As I said in Section 3.10, I am willing to grant that this argu-
these words: "Well, that might be right. I don't know. Why ment is valid. But why should anyone find compatibilism in-
do you think it's right?" Rule 0(3), on the other hand, appeals herently plausible? I think there is only one possible answer
immediately to the reflective intellect; like Modus Tollens, it to this question: because (a) the thesis that we have free will
can be easily grasped by the mind and seen to be true. But the is inherently plausible, and (b) because, many philosophers
mind is no more capable of grasping analyses of can that are think, it is inherently implausible to suppose that a free act
of the order of complexity of the "advantage" analysis than involves some causally undetermined event. Why do they think
126 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 127

this? What is implausible about it? These questions are most between the ages of, say, one and twelve, events that were
conveniently discussed in connection with the Mind Argument. central to my moral education and to the formation of my
character. Suppose that these events were somehow undeter-
mined, but that all subsequent events that made up my history
4.4 The Mind Argument occurs in three forms. Or, at any rate, were determined—or at least were determined given these
there are three closely related strands of argument that are earlier undetermined events. This state of affairs, too, is in-
often twisted together and which it is important to separate. compatible with my having free will, provided determinism
These three "forms" or "strands" have a common beginning simpliciter is.
in a certain set of reflections on what the nature of free action It should be evident by now that there is only one point
must be if the incompatibilist is right. I shall now set forth in the history of each of my acts at which the positing of an
this common beginning. Until further notice, I shall speak in undetermined event could be conceptually relevant to the
persona compatibilistae. question whether that act was a free act: the point at which
the act itself (or the deliberation that immediately preceded
Incompatibilists maintain that free will requires indetermin- it) occurred. For if my acts and decisions and deliberations
ism. But it should be clear even to them that not just any sort are themselves determined by earlier events or states of affairs,
of indeterminism will do. Suppose, for example, that there is then these earlier states of affairs are outside my control at
exactly one undetermined particle of matter somewhere in the time at which I must act or decide or deliberate. They are
the universe, and that it is far from any rational agent, the rest outside my control because they are in the past.
of the universe being governed entirely by strict, deterministic The arguments of Chapter III exploit the fact that if deter-
laws. In that case, determinism is, strictly speaking, false. But, minism is true, then one's acts are determined by states of
clearly, if determinism is incompatible with free will, so is the affairs that obtained before one was born. Whatever dialectical
thesis that everything except one distant particle of matter is advantage an appeal to this fact may win the incompatibilist,
determined. Suppose now we assume that there are many un- however, it has no more logical or philosophical significance
determined particles of matter, some of them very near us. than has an appeal to the fact that if determinism is true,
In fact, let us suppose that at every moment many undeter- then one's acts are determined simply by past events. As
mined events take place inside each human body. But let us Agathon observed, not even the gods can change the past,
also suppose that these undetermined events play no role in and this is no more true of the events of a thousand years ago
shaping or influencing anyone's acts. Just as clearly, if deter- than it is of the events of one second ago. Therefore, if a man's
minism is incompatible with free will, then so is the state of acts were determined by events that occurred earlier than
affairs we are imagining. If the question whether there are those acts—by however short an interval—then one could show
any undetermined events is relevant to the question whether that his acts were unfree using arguments essentially similar
we have free will, this can only be because the question to the arguments of Chapter III. Therefore, if the incompati-
whether there are undetermined events that shape or in- bilist is right, a free act must be an undetermined act, or, at
fluence our acts is relevant to the question whether we have least, the immediate result of undetermined deliberations.
free will. Therefore, the incompatibilist must believe not only But now let us see what an undetermined act would really
that free will entails that there are undetermined events; he be like. Let us consider the case of a hardened thief who, as
must believe that free will entails that there are undetermined our story begins, is in the act of lifting the lid of the poor-box
events that shape our behaviour. And not just any such event in a little country church.' 9 He sneers and curses when he sees
will do for his purposes. Among the events that have shaped what a pathetically small sum it contains. Still, business is
my behaviour are certain events that I witnessed or experienced business: he reaches for the money. Suddenly there flashes
128 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 129

before his mind's eye a picture of the face of his dying mother reason whatever to believe this. There are computers that
and he remembers the promise he made to her by her death- sometimes change state in ways that are not determined by
bed always to be honest and upright. This is not the first oc- their earlier states and their input, but their output is not
casion on which he has had such a vision while performing random in the way in which a table of random numbers is
some mean act of theft, but he has always disregarded it. random.) It is not clear what 'random' and 'chance' mean
This time, however, he does not disregard it. Instead, he thinks when they are applied to single events. These words might
the matter over carefully and decides not to take the money. simply mean 'undetermined% but in that case we should have
Acting on this decision, he leaves the church empty-handed. no argument but only an assertion that undetermined events
We may suppose that this decision was undetermined. That are not the sort of thing that can be called free acts. A more
is, we may assume that there are possible worlds in which interesting possibility is that, in this strand of the Mind Argu-
things were absolutely identical in every respect with the way ment, 'chance' and 'random' are being used, properly or not, to
they were in the actual world up to the moment at which our mean 'uncaused'. But, as we shall discover when we discuss the
repentant thief made his decision—worlds in which, moreover, second strand, an undetermined act need not be an uncaused
the laws of nature are just what they are in the actual world— act. Therefore, if 'random' and 'chance' mean 'uncaused' the
and in which he takes the money. first strand is invalid. I know of no other things that 'random'
According to the incompatibilist, this indetermination is or 'chance' might mean,' and I therefore conclude that the
necessary for the thief's act of repentance to have been a free first strand fails to disprove incompatibilism. Let us turn to
act. But if we look carefully at the idea of an undetermined the second strand.
act, we shall see that such an act could not be a free act.
(ii) If an act, or what looks superficially like an act, is not
(I now revert briefly to speaking in propria persona. It is this determined to occur by prior states of affairs, then it is not
argument about what the incompatibilist's account of free really an act at all. (It should be clear that something can look
action is committed to that the three strands of theMind Argu- like an act and yet not be an act. If a clever brain-physiologist
ment have as their common beginning. I have no wish to dis- stimulated the motor centres of your brain in the right way,
pute the premisses of this common beginning or to question then he might make it look as if you had raised your hand
its validity. I shall now let the compatibilist set forth the when you had not.) This is because, whatever else an act may
three strands separately. I shall reply to them in turn.) be, it is a production of its agent. But if an "act" is undeter-
(i) If the incompatibilist's account of free action is correct, mined, it is not a production of its putative agent and hence
then a free act is an act that is undetermined by prior states not really his act at all. Let us consider our thief who refrained
of affairs. But an act that is undetermined is a mere random —or so we should say if we went by outward appearances—
or chance occurrence, and a random or chance occurrence is from robbing the poor-box. His refraining, or the event that
hardly the kind of thing that could be called a free act. we should initially be inclined to call his refraining, was ex
Reply: The doubtful premiss in this argument is the assertion hypothesi undetermined. Let us examine this event carefully.
that if our acts are undetermined they are mere "random" or We shall assume that some version or other of the mind-body
"chance" events. What does this mean? The words 'random' identity theory is true, but this assumption will be only an
and 'chance' most naturally apply to patterns or sequences of aid to our intuition, and not a premiss of our argument: it is
events. (If the argument is that if our acts were undetermined easier to visualize changes in a material substance than changes
they would issue from us in a meaningless and incoherent in a spiritual substance, but the point of our argument would
jumble, as if we were perpetually deciding what to do by remain if it were restated on the assumption that, for example,
consulting a table of random numbers, there seems to be no Cartesian Dualism provided the correct description of our
130 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 131

nature. On the assumption of psycho-physical identity, the something that rendered it causally impossible for this in fact
thief's act of refraining was identical with a certain event in undetermined event not to occur.
his brain. Now suppose that some event in your brain is ident- Suppose the demon had done this. Then the thief would
ical with one of your acts. And suppose that this event was not really have acted at all, because he would not have been
undetermined by earlier events in your brain. Let us call this the producer of the event in his brain that initiated the bodily
undetermined event E. There would seem to be no reason to motions characteristic of a man leaving a poor-box empty-
suppose that E or any other undetermined event is essentially handed. The case is really no different from the case of the
undetermined. Suppose, for example, that a cup sitting un- man whose brain has been "wired" by a brain-physiologist and
touched on a table suddenly breaks at t and that this event— whose arm rises whenever the physiologist presses a certain
the cup's breaking at t—is undetermined by earlier states of button. That man does not raise his arm when the physiologist
affairs. It would seem that it might have been determined by presses the button; rather, his arm rises as a consequence of
earlier states of affairs: a hammer might have struck the cup what the physiologist does.
at just the proper moment that its breaking at t should have The thief, therefore, if his (supposed) act is caused by the
been causally determined by this blow. (Some philosophers freakish demon, does not act. Now let us restore the "actual"
might want to raise the question whether the event that would situation: we remove from our imagined case the demon and
have occurred in such a case was numerically the same as the his works, leaving an uncaused event and its consequences.
event that in fact happened," but we need not consider this Does this somehow change the "demon" story to a story in
question. It will do for our purposes to note that if an undeter- which the thief does act? How could that be? If a certain
mined event happens, it is logically possible for a descriptively change in one's body is determined to occur by events that
identical determined event to have happened at the same were produced at the pleasure of some rational being other
moment.) Let us suppose that this had happened in the case than oneself, this shows that neither this change nor its con-
of E: suppose that an event just like E and having the same sequences could be, or be a part or component of, one's act.
consequences had happened—it may have been E itself or some But it shows this only by bringing into prominence the fact
other event—and that it was determined. Let us suppose that that the determinants of the change are not to be found within
it was determined by the action of some supernatural power: oneself. Now if an event is undetermined, then it is just as true
a freakish demon caused it.' The sequence of events is this: that the determinants of that event are not to be found within
the freakish demon performs some supernatural act such that oneself as it would be if these determinants were to be found
it is causally impossible for this act to be performed and E— in the capricious acts of a demon, for if the determinants of
or an event just like E—not to occur; E is identical with the one's act are not to be found, then, a fortiori, they are not to
event we also call 'the thief's deliberations', under which des- be found within one. If we are asking whether a certain event
cription we include the outcome of these deliberations; this is a production of mine, then we must admit that there is no
outcome then determines that the thief shall refrain from rob- philosophical difference between this event's being undeter-
bing the poor-box and shall depart. mined and its being determined by an external force, for each
Or we may describe the case as follows. In the "actual" case of these cases is quite simply a case of interference with my
the thief's deliberations take a certain course that results in his plans, desires, deliberations, and hopes. In the one case we
refraining from stealing. These deliberations are identical with have a causally explicable sort of interference. In both cases
a certain rather protracted event in the thief's brain. If his act we have some sort of interference.
is free, and if incompatibilism is true, then this protracted If this is not sufficiently obvious, perhaps we can make it
brain-event is undetermined. We then imagine something that so by elaborating the story of the freakish demon. Let us sup-
might have happened: a freakish demon might have done pose he exercises his influence over the thief in this way: there
132 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 133

is an invisible wire that passes through the thief's skull and putative acts are not real acts, are not among the things he
into his brain (no hole is required, for the wire is made of the does or produces. Now let us suppose that the subtle wire is
subtle matter that the Stoics postulated); at the other end of replaced by one of too, too solid flesh, by a "wire", or wire-
the wire is a sort of piano that is also made of subtle matter and shaped thing, made of brain cells. This change does not weaken
upon the keyboard of which the demon plays; by what he our argument. What difference does it make what the wire
plays, he can "direct", via the subtle wire, the motions of the that delivers undetermined impulses to the thief's brain is
atoms in the thief's brain and can thereby direct the thief's made of? The important thing is that the impulses of which
inner life, including his deliberations. And that is just what he it is the carrier should be undetermined by past states of affairs.
has done in the case we have imagined: by striking the keys Let us now make one final change: let us suppose that the
in a certain order, he has guided the thief's brain through just wire-shaped thing made of brain cells is a natural part of the
that sequence of states that correspond to the deliberations of thief's brain.
a man who refrains from stealing. It is clear, in this case, that Does this make a difference? It's hard to see why it should.
neither the thief's bodily motions nor the pseudo-deliberations Take any part of an agent's brain and replace it with an arti-
that preceded them were things he produced: he was merely ficially produced perfect duplicate; then, assuming the re-
the demon's instrument and acted only in the etiolated sense placed part of the brain was not something essential to the
in which a fiddle acts in the hands of a violinist or a scalpel agent's identity, this would certainly not affect the agent's
acts in the hands of a surgeon. free will. If this is true, such a replacement performed "in
Let us now gradually, very gradually, modify this case. Let reverse" should be equally irrelevant to the agent's free will.
us first suppose that the demon's actions at the keyboard are Our thief has a wire-shaped thing made of brain cells within
undetermined by the demon's own inner states and by any- his brain, in which there arise impulses, undetermined by past
thing else. This change in the case does not weaken the argu- states of affairs, which determine his every action. If we now
ment for our earlier conclusion (that the thief did not act). suppose that this thing is a natural part of the thief, we are
Now let us remove the demon and suppose that the "sort of making a supposition not about its momentary operations,
piano" we imagined the demon to be playing is a sort of but only about its origins. But only the present, momentary
player piano: let us imagine that the keyboard is worked by operations of the parts of one's brain should be relevant to
a mechanism internal to the piano; and let us suppose that this the question whether one is now acting freely. Facts about
internal mechanism is an indeterministic one. This further the origin of a part of one's brain can be relevant to questions
change in the case does not weaken our argument for the con- of free will only in so far as they are relevant to questions
clusion that the thief does not act. Now let us remove the about the momentary operations of the part. Thus, to replace
piano and suppose that impulses simply appear in the subtle a part of one's brain with a perfect duplicate of different ori-
wire (which now protrudes from the thief's skull, its far end gins can have no effect on one's free will—always assuming
unattached to anything) undetermined by prior states of af- that the replacement leaves one's identity intact.
fairs. Again, this change does not weaken our argument for the We have now got to the point of imagining our thief's "acts"
conclusion that the thief did not act. Now let us imagine the being the result of undetermined impulses originating in a
wire becoming shorter and shorter till only the part inside the certain section of his brain. This section of his brain plays the
thief's skull remains. This change, too, does not weaken our same part in our final story that a freakish demon played in
argument: if impulses undetermined by past states of affairs our original story. We moved from one point to the other by
appear in the wire and if these impulses determine the putative gradually modifying the freakish demon and his apparatus
acts of the thief—determine what movements are made by until we had turned them into a natural part of the thief's
his limbs and what thoughts pass through his mind—then these brain. Now in our original story, it was clear that the thief did
134 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 135

not act. We have seen that none of our series of modifications course, say that an undetermined change in a natural part of
introduced any "opportunity" for him to act. We must there- a human being will necessarily have an action of that human
fore conclude that in our final story the thief did not act. But being's among its consequences; I do say that it's not clear that
our final story is just exactly the tale the incompatibilist tells it couldn't. If I am right in supposing this, then the Mind
when he is asked what goes on inside someone who acts freely. Argument—or the strand of it that we are currently investi-
Therefore, the incompatibilist's story is incoherent. gating—errs in supposing that the step from "wire-shaped
All this windoW-dressing, of course, is not strictly necessary. thing" to "natural wire-shaped thing" is an obviously harm-
Anyone whose "analytical imagination", as Hobart calls it, is less step. And therefore, if I am right, this strand of the Mind
in good working order can see right at the outset that a person Argument does not succeed in showing that an undetermined
whose "acts'? are the consequences of undetermined events act is a contradiction in terms.
—whether these undetermined events occur within a natural I shall now attempt to show that it is not clear that an
part of him or not—is not really an agent at all. Such a person apparent act of a human being that was the consequence of
is not acting, but is merely being pushed about or interfered an undetermined change in a natural part of a human being
with. People whose analytical powers are modest, however, could not be a real act. I shall attempt to show this by con-
may find our technique of telling a sequence of stories, each structing two "models" of human action according to which
of which is very slightly closer than its predecessor to the in- this is possible. Since neither of these models is known to be
compatibilist's story of the inner nature of a free act, a useful impossible or incoherent, it will follow that there may well
substitute for adequate analytical powers. be a gap in the apparently continuous progression of modi-
We have seen, therefore, that an act, and a fortiori a free fications of the "freakish demon" case that the Mind Argu-
act, cannot be an undetermined act, for an undetermined act ment presents us with. And if that is true, then the Mind
is a contradiction in terms. Argument fails to establish its conclusion.
I shall not contend that either of the models is right. I do
Reply. My analytical powers are apparently inadequate. The not know—except as we all know—what human action is. I
"longer story" helps me to see why someone would say that shall not contend that either of these models entails that an
an apparent act that is undetermined is not really an act at all. undetermined free act is possible. (I shall presently discuss
This story is, of course, a "slippery-slope" argument. Such the question whether an undetermined free act is possible.)
arguments are, to my mind, a useful but dangerous tool for I shall show only that if either of these models is correct, then
philosophers. They are dangerous because, when they are an undetermined act is possible.
moving briskly along, they acquire a kind of dialectical The first model I shall mention only briefly, since so
momentum that can carry the unwary audience smoothly many philosophers are convinced of its incoherence—on what
over gaps in the argument." grounds, I am not clear—that it would be a tactical mistake
I think there is a gap in the argument. That is, I think there for me to discuss it at length. The model I am alluding to is
is a point at which the allegedly philosophically unimportant that of immanent or agent causation.
alteration that the proponent of the Mind Argument makes The medieval philosophers, like all theists, believed that
in his story does make a philosophical difference. At a certain God was the cause of all things other than himself. Unlike
point in the story, the reader will remember, the freakish many theists, they—or most of them—believed that. God was
demon and his keyboard of subtle matter have been reduced changeless. This raised a difficult philosophical problem, since
to a wire-shaped thing of brain-stuff. The next step is to sup- the causes we observe in nature produce their effects only at
pose that this thing is a natural part of the thief. This next step the expense of some change in themselves. So far as I know, the
is, in my view, the one that makes the difference. I do not, of Schoolmen never solved this problem, though they postulated
136 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 137

a solution and gave this solution a label. They postulated a `act' inside scare-quotes. If the thief moves in a way typical
type of causation that does not involve a change in the cause of a man who very nearly robs a poor-box but draws back from
and called it immanent causation. (Garden-variety causation doing so at the last moment, and if the same thoughts pass
they called transeunt causation.) through his mind that would typically pass through the mind
In our own time, this medieval idea has been revived by of a man who considers but refrains from robbing a poor-box,
Roderick Chisholm and applied not to God but to the human and if these thoughts and movements are caused by the agent
agent.' More or less the same idea has been employed by himself, then it surely follows that if we say "He refrained
Richard Taylor, who prefers the term "agent causation"." from robbing the poor-box", we are describing his action and
(Chisholm has lately come to prefer this second term himself.) describing it in words that are literally correct.
According to Chisholm and Taylor, the agent himself is some- Therefore, if immanent causation is a coherent concept,
times (when he acts freely) the cause of his own acts, or, per- the second strand of the Mind Argument fails to show that
haps, the cause of the bodily or mental changes that manifest the notion of an undetermined act is a contradiction in terms.
them. This does not mean that a change in the agent causes The proponent of the Mind Argument is likely to concede
these acts or changes; rather the agent causes without himself this and to argue, or assert, that immanent causation is not a
changing in any way. (Save accidentally. The human agent, un- coherent concept. I will not dispute this. I have mentioned
like the God of the Philosophers, is not only mutable but is immanent causation because it is an idea that has won the
constantly undergoing change. But, according to the Chisholm- allegiance of highly respected philosophers and which is there-
Taylor theory, none of these changes he undergoes is the cause fore worthy of the attention of anyone who is troubled by
of his free acts.) the Mind Argument. I commend their writings to you. Let
Now let us ignore the contention that an immanently caused us proceed to the second "model" of human action I have
act would be a free act, which is not our present concern, and promised.
consider only the thesis that we are the immanent causes of Donald Davidson has proposed an account of human action
at least some of our acts. If this thesis is true—and it is not that is an elaboration—an elaboration constructed with great
my purpose here to argue that it is coherent, much less that sophistication and philosophical sensitivity—of the following
it is true—then it is easy to see that an act can be an undeter- very simple idea: an act is caused by the very factors its agent
mined act. Let us return to the case of our thief. To say that would adduce if asked to give his reason for performing it, to
it was not determined that he should refrain from stealing is wit, by the agent's desire that a certain state of affairs should
to say this: there is a possible world that (a) is exactly like the be realized and his belief that an act of that type was the best
actual world in every detail up to the moment at which the —or, at least, an unsurpassed—means to realizing that state of
thief refrained from stealing, and (b) is governed by the same affairs.' Suppose, for example, that a certain man has raised
laws of nature as the actual world, and (c) is such that, in it, his hand; and suppose that, when he is asked why he raised
the thief robbed the poor-box. his hand, he says that he did this because he wanted to vote
It is plain that immanent causation allows for the possibility for the measure before the meeting and believed that the way
of such a world: its divergence from the actual world is possi- to vote for this measure was to raise his hand. If what he says
ble because the immanent causation of an event does not im- about the reasons for his act is correct, then, according to
ply that that event is determined by earlier states of affairs— Davidson, the cause of his act is his desire to vote coupled
if anything, the immanent causation of an event implies that with his belief that he could realize his desire by raising his
that event is not determined by earlier states of affairs. More- hand.
over, there is no basis for the caution implied by the use of Now this is a very sketchy and simple-minded account of
phrases like 'what looks superficially like an act' or by writing Davidson's theory. Indeed, if it were my purpose to criticize
138 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 139

Davidson, and if I directed my criticisms against the theory effect is determined by the cause together with the laws of
presented in the preceding paragraph, I should be guilty of nature, however he may understand 'the laws of nature'.
criticizing a travesty. But I don't wish to criticize Davidson's Considering its vast army of adherents, the really striking
account of action, but rather to draw upon it in constructing thing about the standard theory is that there seems to be no
the "model" of human action I have promised. The simple- reason to think it is true. (In what follows, I am indebted to
minded outline of Davidson's theory that I have presented Professor Anscombe's inaugural lecture,' though I do not
will be useful as a first step in the construction of this model. claim to be saying just what she says.)
It makes no difference, really, whether I have accurately rep- Suppose someone throws a stone at a window and that the
resented Davidson, even within the broad limits of accuracy stone strikes the glass and the glass shatters in just the way
appropriate to a rough sketch. I have mentioned Davidson's we should expect glass to shatter when struck by a cast stone.
name only because the model I am going to construct will Suppose further that God reveals to us that the glass did not
immediately suggest his theory to anyone with the most super- have to shatter under these conditions, that there are possible
ficial knowledge of the theory of action, and because I wish worlds having exactly the same laws of nature as the actual
specifically to disclaim any intention of presenting an accurate world and having histories identical with that of the actual
introduction to his theory. Those who want accurate infor- world in every detail up to the instant at which the stone
mation about Davidson's views should seek it in his own writ- came into contact with the glass, but in which the stone re-
ings. (Something that is well worth doing, by the way.) bounded from the intact glass. It follows from what we imagine
We have before us now, whatever its origin may be, a theory God to have told us that determinism is false. But does it also
about the causes of human action. Let us supplement this follow that the stone did not break the glass, or that the glass
theory with a theory about causation itself. This theory is not did not break because it was struck by the stone? It is not easy
even remotely inspired by Davidson; it is in fact the very to see why we should say this follows; perhaps the only reason
antithesis of his theory of causation. Before presenting this we could have for saying this is that we accept a corollary of
theory, however, I wish to say a few words about the standard the standard theory of causation: that instances of causation
or Humean theory of causation, of which Davidson's theory simply are instances of universal, exceptionless laws, that the
is a version. concept of the instantiation of an exceptionless law and the
Many, perhaps most, philosophers believe that causes deter- concept of causation are one and the same concept. But this
mine their effects. They believe, that is, that given the cause— proposition is very doubtful.
we are talking once more about "normal" causation, a relation Let us suppose that we are watching a slow-motion film
that takes events or states of affairs or some such, and not taken by a camera trained on the point at which our stone
persons, as its terms—the effect must follow. (This 'must', of came into contact with our pane of glass. We observe the fol-
course, is the 'must' of physical, not logical, necessity.) They lowing: the stone moves through the air toward the point of
believe that if A is the cause of B, then, if A had happened impact; the stone touches the glass; the glass bends ever so
and had not been followed by B, this could only have been slightly; cracks appear in the glass, radiating outward from the
because the laws of nature were different. Some of the philo- point of impact; shorter cracks appear, joining these radial
sophers I am alluding to think that laws of nature are merely cracks, thus producing detached shards of glass; the stone
exceptionless regularities; others think that only some ex- moves through the space formerly occupied by the unbroken
ceptionless regularities are laws and differ as to what makes a pane, pushing, or apparently pushing, shards of glass out of its
regularity a law. These disputes are important, but they need way; the stone continues along its path, trailing a rough cone
not concern us at present. I shall count a philosopher a pro- of spinning shards. Watching this slow-motion film several
ponent of the standard or Humean view if he thinks that the times should make it very hard to believe that the stone did
140 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 141

not break the glass. We could perhaps construct some wild robbing the poor-box (DB). Let us suppose that the second
hypothesis, logically consistent with our visual evidence, model is correct: R was caused by DB and DB did not have
according to which the stone did not cause the glass to break to cause R; it just did. We may suppose that God has thou-
(for example, the stone didn't touch the glass; a demon caused sands of times caused the world to revert to precisely its
the pane to shatter a thousandth of a second before the stone state at the moment just before the thief decided not to steal,
would have touched it, carefully counterfeiting the appear- and has each time allowed things to proceed without interfer-
ance of what would have happened if things had been left to ence for a few minutes, and that DB caused R on about half
themselves). And if God revealed to us that something like these occasions. On the other occasions, we may suppose, DB
this had happened, then we should have to admit that, despite did not cause R: instead the thief's desire for money, coupled
appearances, the stone hadn't really caused the glass to break. with his belief that the best way for him to get money was to
But what about the revelation we earlier imagined God's rob the poor-box, caused him to rob the poor-box. (If we are
having delivered? Could this revelation really lead us to say right in supposing that causes may produce their effects with-
that, despite appearances, the stone didn't cause the glass to out determining them, then our supposition about the distri-
break? That this is a logical consequence of this revelation? bution of outcomes that results from God's thousands of
Wouldn't it be more reasonable to say this: that, while the "resettings" of the world is a perfectly possible one.)
stone did cause the window to break, it was not determined Does it follow that the thief didn't refrain from robbing the
that it should; that it in fact caused the window to break, poor-box? I don't see why it should be supposed to follow.
though, even if all conditions had been precisely the same, it The thief's act—or what looks very much like his act—was
might not have? caused by his desires and beliefs, after all. What more do we
This case convinces me that whatever the facts of the matter want?" (Remember, I am not saying that the thief acted freely
may be, it is at any rate not part of the concept of causation in refraining from taking the money but only that he acted.)
that a cause—or even a cause plus the totality of its accompany- I do not know what to say about this case to make it more
ing conditions—determines its effect. (A similar case: whatever persuasive. It seems to me to be at least possible that some
the facts of the matter may be, it is not part of the concept sort of Davidsonian theory of the causes of action is correct.
of a human being that a human being should have a liver. If a It seems to me to be at least possible that some causes do not
madman says he is a man who has no liver, we can't simply determine their effects. If these two things could be severally
tell him that it's part of the concept of a human being that a true, then I do not see anything to prevent their being jointly
human being has a liver. But just as all human beings do have true. If they could be jointly true, then the case we have pre-
livers, it may be that all causes do determine their effects.) sented seems to show that it is possible for an agent's act to
Suppose that some causes don't determine but merely, as be undetermined by antecedent states of affairs. We can now
we might say, produce their effects. Suppose that our simple- see why the compatibilist's move from 'wire-shaped thing
minded version of Davidson's theory of the causes of human made of brain cells' to 'natural part' was not so innocent as
action is correct. Suppose that among the causes that merely he alleged. A change in a natural part of one may well be ident-
produce their effects are (some of) the "Davidsonian" causes ical with one's coming to have a certain desire or with one's
of action. This is our "second model". Let us look once more acquiring a certain belief. But a change in something that is
at our thief. If the model of the causes of action we are con- not a part of one—even if it is inside one's head and made of
sidering is correct, then his refraining from robbing the poor- brain cells—could not possibly be identical with either of these
box (R) was caused but not necessitated by his desire to keep things."
the promise he made to his dying mother coupled with his We have examined two "models" of the causation of hu-
belief that the best way to do this would be to refrain from man action, neither of which, I maintain, is known to be
142 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 143

conceptually incoherent. Each has the feature of permitting is "extremely complex but determined in principle", another
the existence of acts that are truly the productions of their being might. Consider, for example, the Intelligence imagined
agents and which are nevertheless undetermined by antecedent by Laplace in his Essai philosophique sur les probabilitès. The
states of affairs. I do not know whether either of these models Intelligence might very well know that if he were to shake the
is correct. I do not know where the springs of action rise. But mechanism just so, then the red light would be the next to
I need not know what human action is in order to show, by flash, or something of that sort. And, in virtue of his having
means of these models, that the second strand of the Mind knowledge of this kind, he might well have a choice about
Argument does not, as it stands, prove that the idea of an un- what light would flash when he pressed the button. But even
determined act is self-contradictory. I have not proved that the Laplacian Intelligence could have no choice about the re-
this idea is an internally consistent idea, of course. But I do sponse of the truly undetermined device, since, in a situation
not think the burden of the proof lies with me. It is generally involving true indetermination, there is nothing that the Intel-
pretty difficult to show that a concept is internally consistent. ligence can apply his remarkable predictive talents to.
In fact, it is not clear what would count as showing such a We may now follow a path that was laid out for us in our
thing. But I will make a conditional promise to anyone who presentation of the second strand of the Mind Argument. Let
,

says I should show that the idea of an undetermined act is us imagine that our red-green device is "hooked up" to our
consistent: I will do it if he will show that the idea of a deter- thief's brain in such a way that, if it flashes green he will steal
mined act is consistent. the money and if it flashes red he will repent and depart; and
Let us now turn to the third strand of the Mind Argument. we may suppose that his coming to be in a state of uncertainty
This strand does not purport to show that the idea of an un- about whether to steal or to repent has the effect of pressing
determined act is inconsistent but rather that the idea of an the button. It should be clear that the thief has no choice
undetermined free act is inconsistent. In broad outline, the about whether to steal or to repent (even supposing that the
argument is this: a free act is an act one has a choice about; case we have imagined is consistent with the supposition that
but no one has any choice about that which is undetermined. he do either of these things). We may now, by a sequence of
minute alterations, turn this case into one in which the red-
(iii) Let us imagine a mechanism the salient features of which green device is replaced by a "functionally equivalent" natural
are a red light, a green light, and a button. If one presses the part of the thief's brain. The technique will be familiar from
button, we'll suppose, then exactly one of the two lights must our earlier discussion, and there is probably nothing to be
flash, but it is causally undetermined which. (If currently gained by actually rehearsing the sequence of alterations. We
accepted physical theory is correct, it would be easy to con- must, of course, have some reason to be sure that we could
struct such a mechanism.) Now suppose that you must press construct a sequence in which no step allowed the incompati-
the button on this mechanism. Have you any choice about bilist to contend that, at that point, the possibility of the thief's
which of the lights will flash? It seems obvious that you have having a choice was introduced. But it would seem that this
no choice about this. It is, of course, very likely that if the requirement will be satisfied provided only that our successive
mechanism were not truly indeterministic but merely ex- modifications of the original case preserve this feature of it:
tremely complex—if it contained something like a tiny roulette that there be a truly undetermined link in the chain that binds
wheel the motion of which was "in principle" determined— deliberation to action. We were able to achieve this in the
you would have no choice about whether the light flashed red "demon" case, and it is evident that the "red-green device"
or green. But this would not follow logically from the descrip- case is not essentially different.
tion of the mechanism. Though you or I might have no choice We shall therefore "end up with" a situation that would
about the result of pressing the button on the mechanism that seem to be just the situation that the incompatibilist sees as a
144 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 145

typical case of free action. And we have seen that in this situ- that the present argument for the incompatibility of free will
ation, the agent has no choice about the action that will follow and indeterminism is invalid. Let us briefly turn to the first
upon his deliberations. Therefore, free action is impossible if model, the model involving immanent causation. According
incompatibilism is true: the very essence of the incompati- to this model the thief himself is the cause of R. (Or, if this
bilist's theory of free action—indeterminism—ensures that the seems to imply that the thief's beliefs and desires are causally
agent who performs a "free" act has no choice about whether unrelated to R, we could employ the language of "causal con-
he performs it, and thus ensures that the act is not free after tribution". A flame, the presence of oxygen, and the dryness
all. Therefore, free action entails determinism and, since free of a certain stick may be said each to contribute causally to
action is possible, free action is a fortiori compatible with the stick's catching fire, though none of them is sufficient for
determinism. the stick's catching fire. If we accept the idea of immanent
This conclusion cannot be evaded by any argument that causation, there would seem to be no reason to reject the idea
exploits the two models of the causal antecedents of action of immanent causal contribution. If we have this idea, then
that were considered above. Let us look at the second of these we may say that DB contributed causally to R though it was
models, which most philosophers will probably find more not sufficient for R, and that the thief himself also contributed
plausible than the first. Suppose that one's acts are caused causally to R. 30 ) But it is far from clear that an appeal to im-
by a "mixture" of desire and belief, as the second model re- manent causation does any more than sweep the incom-
quires; and suppose that this causation is indeterministic, as patibilist's problem under the carpet. It is evident that the
the second model requires. We may concede that if the thief's thief must have become the cause of R at a certain moment.
apparent act of repentance is caused in the normal way by his Consider this event, the thief's acquisition of the property
desires and beliefs, then this apparent act was not merely ap- being the cause of R. This event is undetermined according
parent but real, even if the sequence of causation is indeter- to the theory of immanent causation, for if it were determined
ministic. But it does not by any means follow that it was a R would be determined. If it is undetermined, then the thief
free act, for it does not follow that the thief had any choice cannot have a choice about whether it obtains. Or if the thief
about whether to steal or to repent. Let us use the symbols can have a choice about whether it obtains, if, in general, an
`IV and `DB' once more. Since DB caused R, it may be correct agent can have a choice about an undetermined event, then
to say that the thief acted. But note that DB did not have to one might as well say right at the outset that the thief had a
cause R. Moreover, since DB did not have to cause R, and choice about R and not bring the mysterious notion of im-
since DB alone caused R, R did not have to occur. But then manent causation into the discussion at all.
did the thief have any choice about whether R would occur?
A principle we have already established entails that he did Reply: I begin with an artless question. Given that the thief
not: namely, the principle that no one has any choice about had no choice about whether R followed DB, why couldn't
the occurrence of an undetermined event. And this general someone who accepted the second model say that he never-
principle does not look any less plausible when we examine theless had a choice about whether R occurred? More gener-
the details of the present case. Once DB has occurred, then ally: suppose that s was the "inner state" of a certain agent
everything relevant to the question whether R is going to that was the cause of some act a of his (s might have been a
happen has Occurred. After that we can only wait and see. Davidsonian mixture of belief and desire, or anything you
In a perfectly good sense, it is going to be a matter of chance like); suppose that, while s caused a, and caused a in the "nor-
whether R occurs, whatever sophistical difficulties some phil- mal" or in a "non-deviant" way," s did not determine that a
osophers may raise about defining this notion. should occur; suppose the proponent of the Mind Argument
Thus the "second model" of action cannot be used to show is right in thinking that it follows from these two suppositions
146 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 147

that the agent had no choice about whether, ifs occurred, a an instance of one valid rule of inference is an instance of many.
followed; why couldn't one say that, despite all this, the agent But I should like to see a rule other than ((3) that licenses the
had—or at least might have had—a choice about whether a inference of 'N R occurs' from 'N DB occurs' and 'N(DB
occurred? occurs 3 R occurs)' that is anything other than an ad hoc re-
Here is the obvious reply to this artless question. striction of the general rule ((i) to some special subject matter,
"But that would be possible (in the case of the thief) only such as the causation of human action.
if the thief had a choice about whether DB occurred, and (in The third strand of the Mind Argument, therefore, depends
the general case) only if the agent had a choice about whether on rule 0(i) no less than do our arguments for incompatibilism.
s occurred. While we may sometimes have a choice about the (Of course, only the Third Argument of Chapter III explicitly
inner states that precede our acts, very often we don't. For involves ((3). But I am confident that since the "point" of the
example, it is unlikely that our thief had any choice about three arguments is essentially the same, the First and Second
whether DB occurred; and even if he did, this could only be arguments would be unsound if ((3) were invalid.) We might,
because he had a choice about some earlier states of affairs of in fact, think of the answer that the proponent of the third
which DB was a consequence; if so, the questions we are ask- strand of the Mind Argument has given to the "artless ques-
ing about DB could be asked about those earlier states of tion"—and this answer is an essential part of the third strand
affairs. If someone maintained that those states of affairs, too, of the Mind Argument—as being, like the arguments of Chap-
were states of affairs about which the thief had a choice, we ter III, a defence of an argument in the logic-text sense. More
could point out that they resulted from still earlier states of specifically, we might think of it as being a defence of the
affairs, and this process could, in principle, be carried on till premisses of this argument:
we reached the thief's 'initial state' about which he certainly (1) The thief's repentance was caused but not determined
had no choice. So let us assume at the outset that the thief
by DB, and nothing besides DB was causally relevant to
had no choice about whether DB occurred, for we should
the thief's repentance [assumption for conditional proof]
sooner or later have to make an assumption that would have
the same philosophical consequences. But if the thief had no (2) N DB occurred [premiss]
choice about whether DB occurred, and had no choice about ( 3 ) If (1) is true, then N(DB occurred 3 the thief repented)
whether, if DB occurred, then R followed, in that case, surely, [premiss]
the thief had no choice about whether R occurred." (4) No one (including, of course, the thief) had any choice
This reply to my artless question is, in my opinion, un- about whether the thief repented
answerable. I have asked the artless question and have imagined
(5) If the thief's repentance was caused but not determined
this obvious and unanswerable response, in order to make ex-
by DB, and nothing besides DB was causally relevant to
plicit a premiss of the third strand of the Mind Argument: the the thief's repentance, then the thief had no choice about
validity of the inference rule (3). For it is 0(3) that the pro- whether he repented.
ponent of the Mind Argument appeals to in the preceding
paragraph. If he refuses to countenance ((3), then his elabor- In this argument, (4) follows from (1), (2), and (3) by elemen-
ate examination of the internal "workings" of the thief per- tary logic and ((t); assuming the validity of ((3), (5) then follows
haps establish that N DB occurs and that N(DB occurs 3 R by the principle usually called Conditional Proof. It seems,
occurs). But these two propositions do not entail that N R therefore, that the third strand of the Mind Argument invokes
occurs unless ((3) is valid.' Or so it would seem. There are, of a rule of inference the validity of which is the only doubtful
course, other inference rules than ((3) that would support the premiss of one of our arguments for incompatibilism. It follows
contention that this entailment holds. Any argument that is that if the third strand of the Mind Argument is valid, then
148 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 149

free will is incompatible with determinism! Therefore, the how a compatibilist who accepted the second model of action
compatibilist had better not endorse the third strand of the might respond to this challenge. As a matter of literary con-
Mind Argument after all. And, therefore, the compatibilist is venience, I shall talk as if I accepted this model (which I cer-
left without any argument with which to support his position. tainly don't reject).
But am I any better off than the compatibilist? Suppose a Since I accept the arguments of Chapter III and believe in
compatibilist were to reply to the argument of the preceding free will, I must reject the contention of our imaginary com-
paragraph as follows: patibilist that if (P) were valid the third strand of the Mind
Argument would be sound. Since I accept the second model
I see now that the whole trouble was with ((3). I ought of action, I must therefore reject (3), the second premiss of
not to have accepted any argument that depended on it. the formal argument given above. More generally, I must reject
But neither should you! For if (13) were valid, then the the following proposition:
third strand of the Mind Argument would be sound! And
that's certainly a conclusion you don't want to accept. If an agent's act was caused but not determined by his
Therefore, you too should reject (/3) and with it your prior inner state, and if nothing besides that inner state
arguments for incompatibilism. was causally relevant to the agent's act, then that agent
had no choice about whether that inner state was fol-
This compatibilist is certainly right in his contention that I lowed by that act.
don't want to accept the third strand of the Mind Argument.
Though I have been treating this argument as an argument I must admit that I find it puzzling that this proposition should
for the compatibility of free will and determinism, it is, strictly be false. But if the second model of action is correct, then
speaking, an argument for the incompatibility of free will and the alternative to supposing that this proposition is false is to
indeterminism. Therefore, if it were sound, then either free suppose that one of the following propositions is false:
will would be compatible with determinism or else free will Free will is possible;
would be incompatible with both determinism and indeter-
(P) is valid;
minism and hence impossible. Moreover, as we have seen, if
it were sound free will would be incompatible with determin- Every true proposition about the way things were before
ism (since, if it were sound, (p) would be valid and the Third there ever had been any rational beings is such that no
Argument sound); hence if it were sound, free will would be one has, or ever had, any choice about whether it is true;
impossible. Therefore, anyone who believes in free will must Every law of nature is such that no one has, or ever had,
reject the third strand of the Mind Argument. And anyone any choice about whether it is true;
who believes in free will and who accepts the arguments of Every necessary truth is such that no one has, or ever
Chapter III must reject it on some ground other than the fact had, any choice about whether it is true.
that it depends upon (13). On what ground should the incom-
patibilist who believes in free will reject the third strand of And the falsity of any of these propositions, it seems to me,
the Mind Argument? I think there is no general answer to this is not simply puzzling but inconceivable.
question. Different incompatibilists who held different theo- Now I wish I knew how it could be that, for example, our
ries of action might answer it in different ways. And I, as thief had a choice about whether to repent, given that his re-
I have said, do not know what the right theory of action penting was caused, but not determined, by his prior inner
is. But I shall give an example of how an incompatibilist who states, and given that no other prior state "had anything to
accepted a particular theory of action might respond to the do with"—save negatively: in virtue of its non-interference
challenge offered by our imaginary compatibilist: I shall show with—his act. I have no theory of free action or choice" that
150 THREE ARGUMENTS FOR COMPATIBILISM THREE ARGUMENTS FOR COMPATIBILISM 151

would explain how this could be.' But then neither have I, 4.5 This completes my examination of those arguments for
and neither has the compatibilist, any theory of free action compatibilism that seem to me to be worth discussing at
or choice that would explain how any of the propositions in length. In this final section of Chapter IV, I wish to tie up a
the above list could be false. Moreover, it is certainly not loose end.
unheard of in philosophy for an incontrovertible argument I said in the preceding section that I had no theory of action
to force upon one a puzzling conclusion that one has no or choice that would explain how it could be that the crucial
theoretical account of. Consider, for example, the case of the premiss of the third strand of the Mind Argument was false.
man who has been elected President and who is just mount-
-
The friends of immanent causation may want to remind us
ing the platform to give what everyone except him believes that they do have a theory that purports to explain this.
will be his victory speech; but suppose that he, owing to some Consider our thief. His desire to obey his mother's final
operatic misapprehension, believes that his hated opponent injunction and his belief that he could best achieve this end
has been elected and that the audience expects to hear him by repentance cause or contribute causally to his repentance.
concede defeat. Bitter words burst from his lips: "Every- Moreover, no "deviant causal chains" are involved in this epi-
thing your new President will say to you tonight is false". sode, and his desire and his belief, we may suppose, though
Now I wish I knew how it could be that this man did not they cause or contribute causally to his repentance, do not
tell his audience that everything their new President would determine it. Given that he had no choice about the occurrence
say to them that night was false. I have no theory of truth of the inner state DB, he could have a choice about repenting
or assertion that would explain how this could be. But, as only if he has a choice about the truth of this conditional: if
a familiar sort of argument shows, to suppose that he did DB occurs then R occurs. But how can he have a choice about
tell them this commits one to accepting something incon- whether R follows DB if DB is insufficient for R, and nothing
ceivable. else is even causally relevant, save negatively, to R? This is our
I must choose between the puzzling and the inconceivable. problem. The solution according to the doctrine of immanent
I choose the puzzling. At the end of Section 4.3, I raised the causation is a simple one: "It is false that nothing besides DB
question why anyone should think that incompatibilism was is causally relevant to R. What is true is that no other prior
inherently implausible. Perhaps this puzzle that I am unable event or state of affairs is causally relevant to R. But there is
to resolve is what leads people to think that incompatibilism another sort of causation than "event causation": there is
is inherently implausible." If so, I continue to be convinced agent causation. We may therefore suppose that the thief him-
that incompatibilism, however implausible it may be, is on self, the man, the substance, the continuant, is "causally rel-
balance far more plausible than compatibilism. There are, after evant" to R. That is, we may suppose that the thief was the
all, arguments for incompatibilism that depend upon nothing immanent cause of—or, at any rate, contributed immanent-
more implausible than the validity of (13). And we have seen causally to—his repentance."
in the present chapter that the most promising arguments for This is a "simple" solution to our problem in the sense that
compatibilism turn out to be invalid (the Paradigm Case Argu- it is not very complex. But it is hardly unproblematic. I find
ment), or to depend upon premisses that lack the "luminous the concept of immanent or agent causation puzzling, as I
evidence" of 03) (the Conditional Analysis Argument), or to suspect most of my readers do (those who don't find it down-
be refutable on certain special grounds (the first and second right incoherent). In fact, I find it more puzzling than the
strands of the Mind Argument), or to be more damaging to problem it is supposed to be a solution to. Obscurum per
the position of the compatibilist than to that of the incom- obscurius!
patibilist (the third strand of the Mind Argument)." There is a way in which the idea of immanent causation can
be made no more puzzling than the problem it is supposed
152 THREE ARGUMENTS FOR COMPATIBILISM

to be a solution to. Roughly speaking, the thesis of immanent

causation may be regarded as the conjunction of the free-will
thesis and the second model. I offer a definition, rather in the Chapter V
style, and very much in the spirit, of Roderick Chisholm:
What Our Not Having Free Will Would Mean
s is the immanent cause of a = df
(i) a is an act that s performs;
(ii) a is caused ("transeuntly") by a prior inner state ofs; this 5.1 If the arguments of Chapters III and IV are correct, then
causation is not "deviant"; we cannot consistently believe both in free will and in deter-
minism. In Chapter VI, I shall argue that anyone who accepts
(iii) This prior inner state does not determine that a shall my arguments for incompatibilism should reject determinism
occur, and no other state, unless it is a cause or an effect and accept the free-will thesis. In the present chapter, I shall
of that inner state is causally relevant to a (except nega- set the stage for the argument of Chapter VI by examining
tively); the possibility of rejecting the free-will thesis. I wish to see
(iv) s has a choice about whether that inner state shall be what such a rejection would involve.
followed by a. I shall approach this problem from two angles. I shall first
This definition does not remove all obscurity from this idea ask what rejecting free will would mean for us who reject it,
of immanent causation. I have not said whether the acts and without attempting to answer the question, "What does the
events that the definiens talks of are universals or particulars, denial of free will logically entail?" That is, I shall not ask
and I have not said anything about distinguishing negative what we should be logically or rationally committed to believ-
from positive causal relevance. Still, it seems to me that this ing if we gave up a belief in free will, but, rather, what effects
definition is intelligible in a rough-and-ready sort of way. It at such a rejection would have on our lives whether or not we
any rate leaves the idea of immanent causation less mysterious accepted all those propositions to which the rejection of free
than it found it. Given this definition I have at least some sort will committed us. I shall then go on to ask what rejecting
of understanding of immanent causation and, what is more, I the free-will thesis would logically commit us to.
can see that it exists if and only if we have free will and some
indeterministic causal theory of action along the lines of the 5.2 Many people have said that there is no such thing as free
second model is correct. But, of course, if this is what imma- will. Baron Holbach, for example, writes:
nent causation is, then immanent causation cannot figure in a Man's life is a line that nature commands him to describe upon the
solution to the puzzle with which the third strand of the Mind surface of the earth, without his ever being able to swerve from it, even
Argument presents us. For, in the present sense of 'immanent for an instant . . . Nevertheless, in spite of the shackles by which he is
bound, it is pretended he is a free agent .
causation', that puzzle just is, "How can there be such a thing
as immanent causation?" 37 One might wonder how Holbach found this out, just as one
might wonder how Mark Twain, Clarence Darrow, and Freud
discovered the truth of the very similar propositions that are
to be found in their writings. But these thinkers, and many
others, certainly do reject the free-will thesis, at least verbally,
however shy they may be about giving arguments in support
of this rejection.'
Does it follow that they do not believe—that is, fail to have
154 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 155

the belief—that there is such a thing as free will? This follows to be followed by private detectives, this act manifests a belief
only on two assumptions: (i) that they mean what they say, that she is untrustworthy; if someone edges carefully away
and (ii) that their beliefs are consistent. These are assumptions from every cat he sees, this habit manifests a belief that
I am not willing to make. I think, in fact, that it can be shown cats are best avoided. These pieces of behaviour could, of
that we all believe in free will and that, as a consequence of course, be devices for pretending to the beliefs they are
their sharing this universal belief, Holbach et al. either do not normally manifestations of, just as is the case with verbal
mean what they say or are inconsistent. behaviour.
The argument I shall give in support of this contention is Deliberation, pace Hobbes, is a species of behaviour, and
adapted from an argument of Richard Taylor's,' though, it would therefore not be surprising if deliberation manifested
I believe, it is importantly different from Taylor's argument. certain beliefs of the deliberator. I think that this is indeed
This argument proceeds from the fact that we deliberate. To the case. In my view, if someone deliberates about whether
deliberate is to try to decide between (or, it may be, among) to do A or to do B, it follows that his behaviour manifests a
various incompatible courses of action.' Some philosophers, belief that it is possible for him to do A—that he can do A,
like Hobbes, believe that deliberation, while it ends in activity, that he has it within his power to do A—and a belief that it is
is not itself an activity but rather a state in which the agent possible for him to do B. Someone's trying to decide which
is a passive arena within which various hopes, fears, and desires of two books to buy manifests a belief with respect to each
contend for the prize of causing his next action. Other philo- of these hooks that it is possible for him to buy it just as
sophers see deliberation as activity par excellence. But all surely as would his holding it aloft and crying, "I can buy
philosophers who have thought about deliberation agree on this book". (More surely, in fact, for "trying to decide" is at
one point: one cannot deliberate about whether to perform least partly a description of the book-buyer's unobservable,
a certain act unless one believes it is possible for one to perform inner behaviour. What would be strictly analogous to saying
it. (Anyone who doubts that this is indeed the case may find "I can buy this book" would be appearing to be trying to
it instructive to imagine that he is in a room with two doors decide which book to buy.) I do not mean to deny that people
and that he believes one of the doors to be unlocked and the sometimes deliberate hypothetically. Someone may deliberate,
other to be locked and impassable, though he has no idea for example, about whether to break off his lecture if the
which is which; let him then attempt to imagine himself audience should become unruly, or about whether to cut
deliberating about which door to leave by.) Different accounts down his intake of gin if the doctor should order it. More to
have been given of why this is so. I believe the reason is a very the present point, someone may deliberate about whether to
straightforward one. Almost any piece of human behaviour marry if anyone should ask her, or about whether to buy
manifests certain beliefs of the agent who performs it. One a certain painting if it should come on the market. That is,
very special case of behaviour that manifests its agent's beliefs someone may deliberate about whether to perform a certain
is verbal behaviour: if I say, "The cat is on the mat", then, in act if he can. But the woman who deliberates about whether
the normal case, my utterance manifests a belief that the cat to marry if she should be asked, or who deliberates about
is on the mat, and, I should think, various other beliefs, such whether to buy a certain painting if it should come on the
as a belief that someone may care whether the cat is on the market, is not an example of someone whose deliberation
mat, and a belief that someone may not know that the cat is about a certain course of action fails to manifest a belief that
on the mat. But one's beliefs may also be manifested in one's it is within her power to pursue that course of action. This is
non-verbal behaviour. If I throw my forearm across my eyes, true simply because to deliberate about whether to marry if
then, in the normal case, this behaviour manifests a belief one should be asked is not to deliberate about whether to
that my eyes are in danger; if someone arranges for his wife marry, and to deliberate about whether to buy a certain
156 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 157

painting if it becomes available is not to deliberate about either Holbach never deliberated or else he believed in the
whether to buy that painting. I do not know how to convince case of some pairs of incompatible courses of action that each
someone who denies this that he is wrong. I suppose I might was within his power. Did he deliberate? Well, of course he
ask him to consider the following two propositions, which did. If he hadn't, he would be as notorious as Pyrrho. A man
seem to me to be both true and analogous: to order the guard who didn't deliberate would either move about in random
to shoot the prisoner if he should try to escape is not to order jerks and scuttles, or would withdraw into catatonia.' (That
the guard to shoot the prisoner; to believe that Smith will help some catatonics are people who have ceased to believe in their
us if he can is not to believe that Smith will help us. own free will is an interesting hypothesis.) Therefore he
If I am right about deliberation, then there is a very simple believed in free will, or, at least, in his own free will. Were his
reason why one cannot deliberate about what one does not beliefs therefore inconsistent? Or should we say that he didn't
believe (i.e. fails to believe) is possible:when one deliberates, really believe that there is no free will, but merely said he did?
one's behaviour manifests a belief that what one is deliberating I am not sure what to say about this. Let us consider a more
about is possible. But couldn't one's behaviour manifest a homely story that poses the same puzzle and see what we are
belief that one doesn't have? This strikes me as a very queer inclined to say about it. Suppose there is a man who goes to
question. To ask it is like asking whether a person's frown great lengths to praise his wife's fidelity and to express his
might be a manifestation of an emotion he is not experienc- absolute confidence in her devotion to him and him alone.
ing. Well, a person can pretend to feel a certain emotion he And let us suppose that the belief in her fidelity that he con-
does not in fact feel, and can frown "insincerely" as part of tinually and stridently professes is perfectly sincere in this
this pretence, can't he? This is certainly the case, but we sense: if he were filling out a questionnaire about current
should not say in such a case—if we knew what was going on— sexual mores, and was perfectly convinced of the anonymity
that the person's frown was a manifestation of, for example, of his answers, he would check the "absolutely certain" box
anger; we should say that it wasn't a manifestation of any beside the question, 'Is your spouse faithful to you?' But let
emotion at all but was merely meant to look as if it were. Well, us also suppose that he becomes upset if he discovers that
couldn't a person's deliberative behaviour fail to be a mani- she has been alone with another man, even under the most
festation of any belief? If by 'deliberative behaviour' is meant innocuous and socially acceptable of conditions; that he opens
the outward and visible signs of deliberation, of course. If by her letters; that he attempts to eavesdrop on her telephone
`deliberative behaviour' is meant deliberation, of course not. conversations; that, in fact, he hires private detectives to
Let us now return to Baron Holbach. What did he believe? report on her every move. What should we say about this
What he said is clear enough: that there is no free will: that man's beliefs? It seems clear that he manifests in his behaviour
of any two incompatible courses of action, at most one is a belief that his wife is unfaithful to him, or is at any rate
within the agent's power to carry out. Did he believe what he likely to be, Whatever he may say, his behaviour shows "what
said? There is at least some reason to suspect that he did not he really believes". But what about what he says? Does his
believe that he lacked free will. I have given arguments above verbal behaviour also manifest his beliefs? There are two ways
for the conclusion that no one could deliberate about whether in which we might describe his "doxastic situation". We might
to perform an act that he does not believe it is possible for say he lacks the belief that his wife is faithful and that his
him to perform. Even if these arguments arc wrong, their frequent utterance of sentences that express the proposition
conclusion has been accepted by everyone I know of who has that his wife is faithful is a sham. Or we might say that his
thought about deliberation. It would seem to follow from utterances do manifest the belief that his wife is faithful and
that conclusion, a proposition as near to being uncontroversial that his beliefs arc simply inconsistent. I do not know which
as y philosophically interesting proposition could be, that
-
of these is the right thing to say—though I favour the second
158 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 159

of them—and I have no idea of how to go about finding out. it was—that what one should do is to (attempt to) stop having
I can say no more than this about Baron Halbach. I do not any beliefs that contradict one's "immutable" belief. But if
know how to describe his doxastic situation. Does he lack the the only thing wrong with having contradictory beliefs is that
belief that he has no free will? Is his frequent utterance of being in that state ensures believing a falsehood, then there
sentences that express the proposition that he has no free will would seem to be no reason to attempt to arrange one's other
a sham? Or do his utterances manifest the belief that he has beliefs in such a way as to avoid conflict with the immutable
no free will? Are his beliefs simply inconsistent? While again belief. After all, that belief is ex hypothesi in conflict with
I favour the second alternative, I have no idea of how to go one's evidence. In all probability, therefore, it is false. If one's
about finding out which alternative is right. other beliefs are supported by one's evidence, then one might
Bufperhaps there is no need for us to find out which alterna- as well hold on to them, even though this would ensure that
tive is right. Suppose we were to bring to the attention of one have at least one false belief, for the immutable belief is
Ilolbach—or of anyone else who denies the existence of free very likely false and thus one's having at least one false belief
will—the fact, for fact it is, that in deliberating one manifests is highly probable in any event. Why give up any true beliefs
in one's behaviour a belief in one's free will with respect to simply to achieve consistency?
the act under deliberation. Suppose we were able to convince The argument of the preceding paragraph was very abstract.
him of this. What should he do? This, I think, is the important Let us underscore its point by looking at a particular case:
question. I once heard a story of a Japanese astronomer who seemed to succeed
There would seem to he two courses open to him (besides very well in treating the sun alternately as an inanimate natural body
changing his mind about free will): he might cease to deli- whose properties can be investigated by the techniques of mathematical
berate or he might simply decide to "live with" havingirreme- physics, and as a divinity, the ancestress of theJapanese imperial dynasty;
diably inconsistent beliefs. The former course is unattractive, when challenged about the matter by a European colleague, he said
'Here in Europe I know it's all nonsense, but in Japan I believe it'.'
implying as it does a life spent in catatonic withdrawal or
purely random activity. What about the latter? What precisely Let us interpret this story as entailing that the Japanese astro-
is wrong with having inconsistent beliefs? It's obviously not nomer simply has not got it within his power to stop believing
a wholly good thing, but is it a very bad thing? I think that that the sun is a divinity. What should hello? If his highest
Professor Geach has given the right answer: "The trouble value is consistency, then, clearly, he should give up his western
about inconsistency is that if our factual statements are incon- heliological beliefs, despite the fact that the evidence on which
sistent, one or the other of them is going to be false, and we they are based is as impressive as the evidence for the thesis
often wish statements made by or to ourselves to be true". 6 that the Japanese emperor is descended from the sun is un-
That is to say, having inconsistent beliefs is not "in itself" impressive. But if the only thing wrong with having incon-
a bad thing. What is bad in itself is having false beliefs. Having sistent beliefs is that inconsistent beliefs ensure false beliefs,
inconsistent beliefs is bad only because having inconsistent then his best course of action is to continue to believe that
beliefs ensures having at least one false belief. the sun is a divinity and to believe that the sun is an inanimate
Why is this point important? Well, it could be very impor- object. Following this course of action ensures his having a
tant. Suppose one had a certain belief that one found one false belief, but since, we are supposing, he can't give up his
couldn't rid oneself of, however much one tried. And suppose belief that the sun is a divinity, he is going to have a false
one had indisputable evidence—and regarded it as indisputable belief whatever he does. By retaining his western beliefs,
evidence—for the falsity of that belief. If it were true that however, he gains the advantage of having a body of true
there was something "intrinsically" bad about having contra- belief about the sun that, no doubt, is often useful to him.
dictory beliefs, then it might well be—depending on how bad I think that this is the way Ilolbach should look at his
160 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 161

doxastic situation. He has, or thinks he has, overwhelming approves of his reasons for rejecting free will. He is certainly
evidence for the thesis that he has no free will. Yet he repeat- likely to see himself that way.
edly engages in behaviour that is a sufficient condition for his I now turn to the question, "What does the thesis that we
believing he has free will, behaviour that he would have to be have no free will logically commit its adherents to?"
literally mad to give up. Moreover, he sees, or thinks he sees,
that there is no evidence whatever in favour of the thesis that 5.3 The answer to this question is a philosophical common-
he has free will. The best thing for him to do would seem to place. If we do not have free will, then there is no such thing
be simply to say something like, "I have inconsistent beliefs, as moral responsibility. This proposition, one might think,
but that doesn't trouble me. The only way for me to achieve certainly deserves to be a commonplace. If someone charges
consistency would be for me to give up my assent to the you with, say, lying, and if you can convince him that it was
obviously true proposition that we lack free will. This is too simply not within your power not to lie, then it would seem
large a price to pay for consistency, the only virtue of which, that you have done all that is necessary to absolve yourself of
after all, is that it does not logically necessitate having a false responsibility for lying. Your accuser cannot say, "I concede
belief. But, apparently, I must have a false belief whatever I it was not within your power not to lie; none the less you
do, so this feature of consistency, though a virtue in the ought not to have lied". Ought, as the saying goes, implies
abstract, would not benefit me." can. (Of course, it is unlikely that anyone would believe you
I said earlier that in the present section I should ask what if you said that it was not within your power not to lie, but
rejecting free will would mean for us without asking what the that is not the point.) Similarly, if someone charges you with
denial of free will logically entails. The answer is: to reject not having done something he maintains you ought to have
free will is to condemn oneself to a life of perpetuallogical done, he must withdraw his charge if you can convince him
inconsistency. Anyone who rejects free will adopts a general that you couldn't have done it. If, for example, he charges
theory about human beings that he contradicts with every you with not having spoken up when a word might have saved
deliberate word and act. This perhaps sounds worse than it is. Jones's reputation, he must withdraw his charge if you can
We have seen that it is the best course to adopt for one who convince him that you were bound and gagged while Jones
is convinced that he is in the possession of evidence that was being maligned. (These simple facts are actually a bit too
proves that we have no free will. When described abstractly, simple. An agent may have been unable to perform a certain
the man who rejects free will may seem rather a comic figure, act at a certain time, but—owing to his abilities with respect
a sort of metaphysical version of our suspicious husband. And to acts that were or might have been performed at earlier
indeed we do generally regard the man whose acts belie his times—he may once have been able so to arrange matters that
words as comic. But this is partly because we suppose that he would have been able to perform that act at that time. For
the contradiction of word by act is not an inevitable feature example, I may have been unable to contribute to a certain
of life. This supposition is sometimes not unreasonable. The charity yesterday because I was locked in a bank vault that
suspicious husband might have been different. It, might have can't be opened from the inside. But if it should transpire that
been the case that his words and his acts did not manifest I had shut myself into the vault in order to avoid the repre-
contrary beliefs about his wife's fidelity. There is nothing sentatives of the charity, few people would regard my having
impossible in this; in fact, it would seem to be the normal been locked in the vault as providing me with an adequate
case. But anyone who denies the existence of free will must, excuse for not contributing. The reason is easy to see: though
inevitably, contradict himself with monotonous regularity. there may be a sense in which it is true that I couldn't have
Far from finding such a person comic, one might very well contributed to the charity, there was none the less a time--
see I•im as a kind of philosophical hero, particularly if one before I shut myself in—at which I could so have arranged
162 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 163

matters that I should have been able to contribute to it when over, that there is nothing Gunnar can do about Cosser's in-
the time to do so rolled round. In the sequel, I shall ignore tentions or about the power Cosser has over his acts. It would
the possibility of cases like the "bank vault" case in order to seem therefore, that Gunnar has no choice about whether
avoid unnecessary detail in the statement of my argument.) he shoots Ridley. If he does not change his mind, he will
It would seem to follow from these considerations that shoot Ridley. If he does change his mind, he will shoot Ridley.
without free will there is no moral responsibility: if moral re- Every possible future that is open to him is a future in which
sponsibility exists, then someone is morally responsible for he shoots Ridley. These futures, it is true, are of two sorts:
something he has done or for something he has Ieft undone; those in which he shoots Ridley without having been caused
to be morally responsible for some act or failure to act is at to do so by Cosser and those in which he shoots Ridley only
least to be able to have acted otherwise, whatever else it may because Cosser caused him to. And perhaps he has a choice
involve; to be able to have acted otherwise is to have free will. about which of these sorts the actual future will belong to.
Therefore, if moral responsibility exists, someone has free will. But he has no choice about whether the actual future will
Therefore, if no one has free will, moral responsibility does belong to one of these two sorts, and hence has no choice
not exist. about whether he shoots Ridley.
It would be hard to find a more powerful and persuasive Time passes, we suppose, and Gunnar does not change his
argument than this little argument. Or so one would have mind and does shoot Ridley without having been caused to
thought. In a remarkable recent essay, however, Harry do so by Cosser. Could he have done otherwise than shoot
Frankfurt has cast grave doubts upon it.' Frankfurt's reason- Ridley? Obviously not: before he shot Ridley he had no
ing is essentially this. He points out that it depends essentially choice about whether he would shoot Ridley, and so we may
on the premiss: now say, after the fact, that he couldn't have done otherwise
than shoot him. Is he responsible for having shot Ridley? It
A person is morally responsible for what he has done
would certainly seem so, at least if anyone is ever responsible
only if he could have done otherwise.
for anything. Let us suppose, for the moment, that Cosser had
(Frankfurt calls this proposition the Principle of Alternate not devised his plan for ensuring that Gunnar shoot Ridley.
Possibilities.) He then shows how to construct counter- Let us consider Gunnar's wicked act and let us "build into"
examples to it. Here is a "Frankfurt counter-example" to the our description of the circumstances under which it was per-
Principle of Alternate Possibilities. The example is my own, formed whatever may be necessary for Gunnar's being re-
but it does not differ in any important way from Frankfurt's sponsible for it. (Someone might, for example, want us to
own examples. include in our description of the act that Gunnar was its
Suppose there is a man called Gunnar who has decided to immanent cause, a la Chisholm.) Now let us consider adding
shoot his colleague Ridley. Suppose a third man, Cosser, very to this description of the circumstances of Gunnar's act the
much desires that Gunnar shoot Ridley. Cosser is naturally statement that Cosser would have caused him to perform it
delighted with Gunnar's present intention to shoot Ridley, if he had changed his mind. Does this statement alter the fact
but he realizes that people sometimes change their minds. of Gunnar's responsibility? How could it? It is no more than
Accordingly, he devises the following plan: if Gunnar should a statement about certain of the unrealized causal dispositions
change his mind about shooting Ridley, Cosser will cause of the objects that constituted Gunnar's environment during
Gunnar to shoot Ridley. We may suppose that Cosser is able some short period that immediately preceded his act. No act,
directly to manipulate Gunnar's nervous system, and is thus no intention, no disposition of Cosser had the slightest effect
able, in the fullest and strongest sense of the word, to cause on Gunnar's act or on the deliberations that led up to it. The
Gunnar to act according to his wishes. Let us suppose, more- causal history of his act is just what it would have been if

164 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 165
Cosser had never existed. Now it would seem to be undeniable a principle about unperformed acts (things we have left
that if there are two possible worlds in each of which the same undone). In Sections 5.5 and 5.6, I shall consider two prin-
agent performs the same act, and if the causal history of this ciples about the consequences of what we have done (or left
act is the same in every detail in both worlds, then it cannot undone). In Section 5.7, I shall argue that if these three prin-
be that the agent is responsible for that act in one world and ciples are true then moral responsibility requires free will.'
not responsible for it in the other. If this general principle is
correct, however, then Cosser's unactualized disposition to 5.4 Consider the following principle (the Principle of Possible
cause Gunnar to shoot Ridley does not relieve Gunnar of his Action):
responsibility for having shot Ridley. It is doubtless true that PPA A person is morally responsible for failing to perform a
if Gunnar had changed his mind about shooting Ridley, and given act only if he could have performed that act.
had, as a consequence, been caused to shoot Ridley, he would This principle is intuitively very plausible. But the same might
not have been responsible for shooting Ridley. But this pro- have been said about the Principle of Alternate Possibilities.
position is consistent with the proposition that Gunnar is Can we show that PPA is false by constructing a counter-
responsible for shooting Ridley. example to it that is like Frankfurt's counter-examples to
It seems we must conclude that we have a genuine case in the Principle of Alternate Possibilities? An adaptation of
which an agent is morally responsible for having shot a certain Frankfurt's general strategy to the case of unperformed acts
man even though he could not have done otherwise than would, I think, look something like this: an agent is in the
shoot that man. This case shows that the Principle of Alter- process of deciding whether or not to perform a certain act a.
nate Possibilities is probably false. (I shall later explain why He decides not to perform a, and, owing to this decision,
I say 'probably%) Let us suppose that the Principle of Alter- refrains from performing a. 10 But, unknown to him, there
nate Possibilities is indeed false. What follows? It does not were various factors that would have prevented him from per-
follow that we might be morally responsible for our acts even forming (and perhaps even from deciding to perform) a. These
if we lacked free will; it follows only that the usual argument factors would have come into play if he had shown any ten-
for the proposition that moral responsibility entails free will dency towards performing (perhaps even towards deciding to
has a false premiss. But might there not be other arguments for perform) a. But since he in fact showed no such tendency, these
this conclusion? Might there not be other premisses from which factors remained mere unactualized dispositions of the objects
this entailment could be derived? I shall argue that there are. constituting his environment: they played no role whatever
I shall exhibit three principles that, I shall argue, have the fol- in his deciding not to perform or in his failure to perform a.
lowing features: (i) they entail that free will is necessary for Putative counter-examples to PPA prepared according to
moral responsibility; (ii) though they are clearly in some sense this recipe produce, in me at least, no inclination to reject
"similar to" or "variants on" the Principle of Alternate Possi- this principle. Let us look at one.
bilities, they are nevertheless immune to Frankfurt-style Suppose I look out the window of my house and see a man
counter-examples. (I shall call counter-examples to principles being robbed and beaten by several powerful-looking assail-
that are similar to but distinct from the Principle of Alternate ants. It occurs to me that perhaps I had better call the police.
Possibilities, and which are as strategically similar to Frank- I reach for the telephone and then stop. It crosses my mind
fui t counter-examples as possible, Frankfurt-style counter- that if I do call the police, the robbers might hear of it and
examples. I shall reserve the term 'Frankfurt counter-example' wreak their vengeance on me. And, in any case, the police
for counter-examples to the Principle ofAlternate Possibilities.) would probably want me to make a statement and perhaps
The Principle of Alternate Possibilities concerns performed even to go to the police station and identify someone in a
acts (things we have done). In Section 5.4, I shall consider line-up or look through endless books of photographs of
 WHAT OUR NOT HAVING FREE WILL WOULD MEAN 167
166 WHAT OUR NOT HAVING FREE WILL WOULD MEAN

thugs. And it's after eleven already, and I have to get up early affairs are particulars or universals. In order to avoid taking
tomorrow. So I decide "not to get involved", return to my sides in the debate about this, I shall adopt the following
chair, and put the matter firmly out of my mind. Now suppose strategy: I shall state a certain principle about excuse from
also that, quite unknown to me, there has been some sort of responsibility that seems to me to be plausible, provided the
disaster at the telephone exchange, and that every telephone events or states of affairs we hold people responsible for are
in the city is out of order and will be for several hours. particulars; and I shall state a similar principle that seems to
Am I responsible for failing to call the police? Of course me to be plausible, provided the events or states of affairs we
not. I couldn't have called them. I may be responsible for hold people responsible for are universals; for each of these
failing to try to call the police—that much I could have done— principles, I shall argue that it cannot be refuted by Frankfurt-
or for refraining from calling the police, or for having let style counter-examples. The first of these principles, which I
myself, over the years, become the sort of man who doesn't shall call principles of possible prevention, is:
(try to) call the police under such circumstances. I may be PPP1 A person is morally responsible for a certain event-
responsible for being selfish and cowardly. But I am simply particular only if he could have prevented it.
not responsible for failing to call the police. This "counter- This principle is about events; but if we were to examine a
example", therefore, is not a counter-example at all and PPA principle, otherwise similar, about "state-of-affairs-particulars"
is unscathed. (for example, the way secondary education is organized in
It is, of course, proverbially hard to prove a universal Switzerland)," we could employ arguments that differ from
negative proposition. Perhaps there are Frankfurt-style the following arguments only in verbal detail.
counter-examples to PPA. But I don't see how to construct What are events if they are particulars? They are items that
one. I conclude that Frankfurt's style of argument cannot be can be witnessed, at least if they consist in visible changes in
used to refute PPA. visible particulars, remembered, and reported.' They are
typically denoted by phrases like 'the fall of the Alamo', 'the
5,5 Both the Principle of Alternate Possibilities and PPA are death of Caesar', 'the unexpected death of Caesar', and 'what
principles about acts, performed or unperformed. But, in fact, Bill saw happening in the garden'." How shall we identify
when we make ascriptions of moral responsibility, we do not and individuate event-particulars (hereinafter, "events")?
normally say things like 'You are responsible for killing J ones' Individuating particulars, whether events, tables, or human
or 'He is responsible for failing to water the marigolds'. We beings is always a tricky business. (Consider the Ship of
are much more likely to say 'You are responsible for Jones's Theseus.) As Davidson says,
death' or 'He is responsible for the shocking state the mari- Before we enthusiastically embrace an ontology of events we will want
golds are in'. That is, we normally hold people responsible to think long and hard about the criteria for individuating them. I am
not for their acts or failures to act (at least explicitly), but myself inclined to think we can do as well for events generally as we
for the results or consequences of these acts and failures. can for physical objects generally (which is not very well) .. 14
.

What, ontologically speaking, are results or consequences of In a later paper than the one this quotation is taken from,
action and inaction? What sorts of thing are Jones's death Davidson tries to "do as well". He tells us that finding a satis-
and the shocking state the marigolds are in? The general terms factory criterion of individuation for events will consist in
`event.' and 'state of affairs' seem appropriate ones to apply providing "a satisfactory filling for the blank in:
to these items. But what are events and states of affairs? This If x and y are events, then x = y if and only if -- ."
question, like all interesting philosophical questions I know
of, has no generally accepted answer. Philosophers do not The "filling" he suggests for this blank is (roughly) `x and y
seem even to be able to agree whether events and states of have the same causes and effects'." The hiconditional so
168 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 169

obtained, is, I have no doubt, true. But this biconditional will vagueness. A necessary condition for x and y having the same
not be "satisfactory" for our purpose, which is the evaluation causal genesis is "their" having developed from the same
of PPP1. What we want to be able to do is to tell whether some sperm and egg." But this is not sufficient, or "identical"
event that would happen if what we earlier called "unactualized (monozygotic) twins would be numerically identical. This
dispositions of the objects constituting the agent's environ- criterion can be used to make sense of talk about what some
ment" were to come into play, is the same as some event—the particular person would have been like if things had gone
event responsibility for which we are enquiring about—that very differently for him." Can we devise a "counter-factual"
actually has happened; that is, we want to know how to tell criterion for events that is at least no worse than our criter-
of some given event whether it, that very same event, would ion for persons? I would suggest that we simply truncate
nevertheless have happened if things had been different in Davidson's criterion: x is the same event as y if and only if
certain specified ways. (For when we ask whether an agent x and y have the same causes. That is to say, if x is the product
could have prevented a certain event e by doing, say, x, we of certain causes, then, necessarily, an event y is the product
shall have to be able to answer the question whether e would of those causes if and only if y is x. (Note the similarity of
none the less have happened if the agent had done x.) this criterion to the causal-genesis criterion of personal
To see why Davidson's criterion cannot be used to answer identity.)
our sort of question about event-identity, consider the follow- I do not know how to justify my intuition that this criterion
ing formally similar criterion of individuation for persons: `x is correct, any more than I know how to justify my belief in the
and y are the same person if and only if x and y have the causal-genesis criterion. But, of course, arguments must come
same blood relatives, including siblings'. This criterion, while to an end somewhere. I can only suggest that since substances
true, does not help us if we are interested in counter-factual (such as human beings and tables) should be individuated
questions about persons. For, obviously, any given man might by their causal origins, and since we are talking about events
have had different relatives from those he in fact has; he might that, like substances, are particulars, the present proposal is
have had an additional brother, for example. Davidson's pro- plausible. Moreover, I am aware that this proposed criterion
posed criterion is of no help to us for what is essentially the is vague. It is not clear in every case of, say, a story about the
same reason: any given event might have had different effects events leading up to Caesar's murder, whether it would be
from the effects it has in fact had. For example, if an historian correct to say that the murder had "the same causes" in the
writes, "Even if the murder of Caesar had not resulted in civil story that it had in reality. But I think the notion of same
war, it would nevertheless have led to widespread bloodshed", event is clear just in so far as the notion of same causes is
he does not convict himself of conceptual confusion. But he clear. And this latter notion is surely not hopelessly unclear:
is certainly presupposing that the very event we call 'the if Cleopatra had poisoned Caesar in 48 Be, then, clearly, there
murder of Caesar' might have had different effects." would have happened an event that has not in fact happened,
The above considerations are not offered in criticism of an event it would have been correct to call 'Caesar's death',
Davidson's criterion, which is, after all, true, and may be a and which would have had different causes from the event
very useful criterion to employ, say, when we are asking that is called 'Caesar's death'. And, just as clearly, we cannot
whether a given brain-event and a given mental event are one say of the event we in fact call 'Caesar's death', "Suppose it
event or two. But Davidson's criterion is not the sort of cri- had been caused four years earlier by Cleopatra's poisoning
terion we need. We need a criterion that stands to Davidson's Caesar in Alexandria". Moreover, it is hardly to be supposed
criterion as 'x and y are the same human being if and only if that we should be able to devise a criterion that will resolve
x and y have the same causal genesis' stands to the above cri- all "puzzle cases", since we are unable to devise such a cri-
terion of personal identity. I use 'causal genesis' with deliberate terion for people, mountains, or tables."
170 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 171
While I think this criterion of counter-factual event-identity 5.6 Let us now turn to a principle about universals:
is true, I think it is worth pointing out that the argument of
the sequel will require only a certain principle that is entailed PPP2 A person is morally responsible for a certain state of
by but does not entail this criterion: if x is the product of affairs only if (that state of affairs obtains and) he
certain causes, then, necessarily, an event y is the product of could have prevented it from obtaining."
those causes if y is x. This weaker principle may also be stated The states of affairs "quantified over" in this principle are
in these words: no event could have had causes other than its universals in the way propositions are universals. Just as there
actual causes. The criterion I have proposed is equivalent to are many different ways the concrete particulars that make
the conjunction of this weaker principle with the proposition up our surroundings could be arranged that would he sufficient
that if a certain chain of causes and that chain alone has pro- for the truth of a given proposition, so there are many differ-
duced a certain event as its effect, then no other event could ent ways they could be arranged that would be sufficient for
have been the effect of that chain of causes and that chain the obtaining of a given state of affairs. Consider, for example,
alone. While I think that this proposition is true, we shall not the state of affairs that consists in Caesar's being murdered.
need to make any use of it. This state of affairs obtains because certain conspirators
Let us now return to PPPI. Can we devise a Frankfurt-style stabbed Caesar in Rome in 44 BC, 2 ' but, since it is a universal,
counter-example to this principle? Let us try. it, that very same state of affairs, might have obtained because,
Gunnar shoots and kills Ridley (intentionally), thereby say, Cleopatra had poisoned him in Alexandria in 48 BC. But
bringing about Ridley's death, a certain event, But there is this is a bit vague. In order the better to talk about "states of
some factor, F, which (i) played no causal role in Ridley's affairs", let us introduce "canonical" names for them. Such
death, and (ii) would have caused Ridley's death if Gunnar had names will consist of the result of prefixing 'its being the case
not shot him—or, since factor F might have caused Ridley's that'--hereinafter, 'C'—to "eternal" sentences.' Thus a can-
death by causing Gunnar to shoot him, perhaps we should say, onical name for the state of affairs referred to above would
"if Gunnar had decided not to shoot him"— and (iii) is such be `C (Caesar is murdered)'. And let us say that the result of
that Gunnar could not have prevented it from causing Ridley's flanking the identity-sign with canonical names of states of
death except by killing, or by deciding to kill, Ridley himself. affairs expresses a truth just in the case that the eternal sen-
So it would seem that Gunnar is responsible for Ridley's tences embedded in these names express equivalent proposi-
death, though he could not have prevented Ridley's death. tions, where propositions are equivalent if they are true in
It is easy to see that this story is simply inconsistent. What just the same possible worlds. I shall assume that every propo-
is in fact denoted by `Ridley's death' is not, according to the sition is equivalent to and only to itself. This assumption could
story, caused by factor F. Therefore, if Gunnar had not shot be dispensed with at the cost of complicating the syntax of
Ridley, and, as a result, factor F had caused Ridley to die, the sequel. A state of affairs will be said to obtain if the
then there would have been an event denoted by `Ridley's proposition associated with it—that is, the proposition ex-
death' which had factor F as (one of) its cause(s). But then pressed by the sentence embedded in any of its canonical
this event would have been an event other than the event in names—is true. Thus C(Caesar is murdered), C(Cacsar is
fact denoted by 'Ridley's death'; the event in fact denoted by stabbed), and C(Caesar is poisoned) are three distinct states
`Ridley's death' would not have happened at all. But if this of affairs, the first two of which obtain and the last of which
story is inconsistent it is not a counter-example to PPPI. And does not. To prevent a state of affairs from obtaining is to
I am unable to see how to construct a putative Frankfurt-style prevent its associated proposition from being true, or to see
counter-example to PPPI that cannot be shown to be incon- to it that or ensure that that proposition is not true.
sistent by an argument of this sort. Let us now, so armed, return to PPP2. Can we show that
172 WHAT OUR NOT HAVING FREE WILL WOULD MEIN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 175

PPP2 is false by constructing Frankfurt-style counter-examples In fact, it is arguable that C(Ridley dies) is the very same
to it? What would an attempt at such a counter-example look state of affairs as C(Ridley is mortal). Given our principle of
like? Like this, I think: Gunnar shoots Ridley (intentionally), identity for states of affairs, these "two" states of affairs are
an act sufficient for the obtaining of Ridley's being dead, one if the two eternal sentences 'Ridley dies' and 'Ridley is
a certain state of affairs; but there is some factor, F, which mortal' express the same proposition. And what proposition
(i) played no causal role in Ridley's death, and (ii) would could either of them express but the proposition also expressed
have caused Ridley's death if Gunnar had not shot him (or by 'Ridley does not live forever' and 'Ridley dies at some
had decided not to shoot him), and (iii) is such that Gunnar time or other'? So, it should seem, Gunnar is not responsible
could not have prevented it from causing Ridley's death for C(Ridley dies), and the attempted counter-example to
except by killing (or by deciding to kill) Ridley himself. So it PPP2 fails.
would seem that Gunnar is responsible for Ridley's being Nor do matters go differently if, somewhat implausibly, we
dead, though he could not have prevented this state of affairs think of 'Ridley's being dead' as denoting some more "specific"
from obtaining. state of affairs, such as C(Ridley is killed by someone). If
This case seems to show that PPP2 is false. But in fact it Gunnar is indeed responsible for C(Ridley is killed by some-
does not. Let us remember that if this case is to be a counter- one), we shall nevertheless have a counter-example to PPP2
example to PPP2 and not to some other principle, some prin- only if Gunnar could not have prevented this state of affairs
ciple involving particulars, the words `Ridley's being dead' from obtaining. Let us flesh out "Factor F" with some detail
that occur in it must denote a universal. What universal? to ensure that this is the case: suppose there is a third party,
Presumably, C(Ridley dies). But while it is indeed true that Pistol, who would have killed Ridley if Gunnar had not; and
Gunnar could not have prevented C(Ridley dies) from obtain- suppose Gunnar was able to prevent Pistol's killing Ridley
ing, I do not think it is true that Gunnar is responsible for only by killing Ridley himself. By these stipulations, we
C(Ridley dies). Why should anyone think he is? Well, Gunnar ensure that Gunnar could not have prevented C(Ridley is
did something—shooting Ridley—that was sufficient for killed by someone). But do we, in making these stipulations,
C(Ridley dies). What is more, he performed this act inten- absolve Gunnar from responsibility for this state of affairs,
tionally, knowing that it was sufficient for that state of affairs. or is his being responsible for it at least consistent with our
This argument, however, is invalid. For consider the state of stipulations?
affairs C(Ridley is mortal). When Gunnar shot Ridley, he It seems evident to me that we absolve him. C(Ridley is
performed an act sufficient for the obtaining of this state of killed by someone) would have obtained no matter what
affairs. But it would be absurd to say that Gunnar is respon- Glinnar had done, and the principle, 'If a certain state of
sible for C(Ridley is mortal). God, or Adam and Eve jointly, affairs would have obtained no matter what x had done, then
or perhaps no one at all, might be held accountable for x is not responsible for it' seems, to say the least, highly
Ridley's mortality; certainly not his murderer. (Unless, of plausible. Are there any features of the case that should lead
course, Ridley would have lived forever if he hadn't been us to reject this highly plausible principle? It would seem not.
murdered; let's assume that this is not the case.) We might The only reason the case presents us with for thinking that
even imagine that Gunnar was a madman and believed that Gunnar is responsible for C(Ridley is killed by someone) is
Ridley would live forever if he did not shoot him. In that case, this: Ridley performed an act sufficient for the obtaining of
Gunnar performed a certain act intentionally, knowing that it this state of affairs. But, as we have seen, this reason is not
was sufficient for C(Ridley is mortal), and we may suppose, good enough; if it were, then Gunnar would be responsible
in order to bring about C(Ridley is mortal). Even in this case, for C(Ridley is mortal).
however, Gunnar is not responsible for C(Ridley is mortal). Now it is possible to flesh out "Factor F" in a way that

174 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 175

evades this reply. We need only suppose that Factor F "works him to bring it about. That is to say, in a Frankfurt-style
through" Gunnar. That is, suppose Factor F, instead of in- counter-example to PPP2, the agent's inability to prevent
volving a third party, involves some mechanism that has the a certain state of affairs is secured by so constructing the case
following property: if Gunnar had decided not to shoot Ridley, that that state of affairs would have obtained no matter what
this mechanism would have caused him to change his mind he had done—or, at least, would have obtained no matter
and shoot Ridley despite his earlier decision. This stipulation, what choices or decisions he had made. But then, according
like our earlier stipulation about Pistol, will ensure that Gunnar to our two "no matter what" principles, the agent will not be
could ndt have prevented C(Ridley is killed by someone). responsible for the state of affairs.
Moreover, this revision will not allow us to argue that Gunnar It occurs to me that someone might object to our employ-
is absolved from responsibility for this state of affairs because ment of these principles on the ground that they "presuppose"
it would have obtained no matter what he had done. For, in PPP2. This, however, would be a mistake, since 's would have
the revised case, C(Ridley is killed by someone) would not obtained no matter what x had done' and 's would have ob-
have obtained no matter what Gunnar had done: it would tained no matter what choices or decisions x had made' are
not have obtained if Gunnar had not shot Ridley. We can, stronger statements than (i.e. neither one is entailed by) 'x
however, use a similar principle to the same purpose: 'If a could not have prevented s'. Suppose, for example, that
certain state of affairs would have obtained no matter what Swallow is a dipsomaniac: he cannot refrain from drinking if
choices or decisions x had made, then x is not responsible for drink is present. Suppose drink is present; then Swallow
it'. This principle seems at least as evident as the "no matter cannot prevent C(Swallow drinks). But it is not true that this
what he had done" principle, and it obviously entails that, in state of affairs would have obtained no matter what Swallow
the revised case, Gunnar is not responsible for C(Ridley is had done, or true that it would have obtained no matter what
killed by someone). Moreover, this second principle could be choices or decisions he had made. If he had not drunk, it
applied in the third-party case that we considered in the pre- would not have obtained; if he had steadfastly chosen not to
ceding paragraph, in case anyone wanted to split hairs about drink it would not have obtained. But if `x could not have
`no matter what he had done' like the hairs that were split in prevented s' does not entail (is a weaker statement than)
Chapter II, note 3. either 's would have obtained no matter what x had done' or
If we had chosen to examine instead of C(Ridley is killed `s would have obtained no matter what choices or decisions
by someone) some even more specific state of affairs, such as x had made', then it would seem that neither of our two prin-
C(Ridley is shot to death at 3.43 p.m. 12 January 1949 in ciples "presupposes" PPP2.
Chicago), this would have made no difference to our argument, These two principles may be false, but they certainly seem
which in no way depended on the "degree of specificity"— to be true. If someone were to construct a Frankfurt-style
whatever, precisely, that might mean—of C(Ridley is killed by example in which an agent was clearly responsible for a state
someone). If we attempt to construct a Frankfurt-style of affairs that would have obtained no matter what he had
counter-example to PPP2, we must construct it in such a way done or had chosen or had decided, then, of course, we should
that the agent who figures in it could not have prevented have to reject our two principles, plausible though they may
a certain state of affairs. In a Frankfurt-style counter-example, be. But none of the Frankfurt-style examples we have con-
this feature is achieved by the positing of an "unactualized sidered, or any other that I know of, has this feature. The
disposition of the agent's environment", which, if the agent Gunnar-Ridley examples—and I know of no Frankfurt-style
had acted (or had decided or chosen) differently, would have putative counter-examples to PPP2 that are essentially dif-
brought about that state of affairs independently of his will, ferent from the Gunnar-Ridley exarnples—may initially impress
or else would have affected his will in such a way as to cause one as implying that Gunnar is responsible for some baneful
176 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 177

state of affairs that would have obtained no matter what he that the passage of a runaway horse through the streets of
had done or chosen or decided. This impression, however, Rome will result in the injury of some of her detested citizens.
seems to depend on the assumption that if an agent performs Unknown to Ryder, however, all roads lead to Rome: Dobbin's
a certain act (and is responsible for performing it) and if that career would have led him and Ryder to Rome by some route
act is sufficient for the obtaining of a certain state of affairs, no matter what Ryder had done. Therefore, Ryder could not
and if the agent performs that act knowing it is sufficient for have prevented C(Ryder passes through Rome on a runaway
the state of affairs (or perhaps even performs it in order to horse). Is Ryder responsible for this state of affairs? It is
cause that state of affairs to obtain), then that agent is re- obvious that he is not. And it seems obvious that he is not
sponsible for that state of affairs. But, as we have seen, this responsible for this state of affairs just because it would have
assumption is false, for, if it were true, Gunnar would be been the outcome of any course of action he might have
responsible for its being the case that Ridley is mortal. elected. Our conclusion—that PPP2 cannot be refuted by
It seems, therefore, that Frankfurt's ingenious style of Frankfurt-style counter-examples—is therefore not only true
argument, however effective it may be against the Principle but intuitively plausible.
of Alternate Possibilities, cannot be used to refute PPP2. This The universals that PPP2 is "about" are states of affairs;
result is not implausible. Its intuitive plausibility can be seen but if we had examined a principle, otherwise similar, about
if we think in terms of the following picture: in attempting to "event-universals" (for example, "its coming to pass that
construct Frankfurt-style counter-examples to PPP2, we have Caesar dies"), we could have employed arguments that differed
been imagining cases in which an agent "gets to" a certain from the above arguments only in verbal detail.
state of affairs by following a particular "causal road", a road It has been suggested to me" that these arguments appear
intentionally chosen by him in order to "get to" that state of less plausible if one reflects on the fact that essentially similar
affairs; but, because this state of affairs is a universal, it can arguments could be used to show that Gunnar did not bring
be reached by various causal roads, some of them differing about C(Ridley is killed) and that Gunnar's pulling the trigger
radically from the road that is in fact taken; and, in the cases did not cause this state of affairs. It is certainly true that if
we have imagined, every causal road that any choice of the the above arguments are sound, then similar arguments can
agent's might set him upon leads to this same state of affairs. be used to show that Gunnar did not bring about C(Ridley is
This is why the agent in our attempts at Frankfurt-style killed) and that his bodily movements did not cause this state
counter-examples always turns out not to be responsible for of affairs to obtain. But these conclusions appear to me to be
the state of affairs he is unable to prevent. simply true. Let us concentrate on
Perhaps the point of this fanciful talk about "causal roads"
will be clearer if we look at a case in which an agent is unable (I) Gunnar did not bring about C(Ridley is killed).
to prevent a certain state of affairs that involves roads secun-
dum litteram. Suppose Ryder's horse, Dobbin, has run away Why should anyone think (I) is false? It would clearly be
with him. Ryder can't get Dobbin to slow down, but Dobbin invalid to argue that (I) is false because Gunnar did something
will respond to the bridle: whenever Ryder and Dobbin come logically or causally sufficient for C(Ridley is killed), for by
to a fork in the road or a crossroad, it is up to Ryder which the same argument we could establish the falsity of the (true)
way they go. Ryder and Dobbin are approaching a certain proposition that Gunnar did not bring about C(Ridley is
crossroad, and Ryder recognizes one of the roads leading away mortal). Or consider the case of Ryder and Dobbin. In turning
from it as a road to Rome. Ryder has conceived a dislike for into a certain road, Ryder did something causally sufficient
Romans and so, having nothing better to do, he steers Dobbin for his passing through Rome on a runaway horse, but would
into the road he knows leads to Rome, motivated by the hope anyone want to say that Ryder brought about the (for him
178 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 179

inevitable) state of affairs C(Ryder passes through Rome on So, it would seem, we are unable to devise a Frankfurt-style
a runaway horse)? counter-example either to PPP1 or to PPP2. If our attempts
The states of affairs we have been considering are universals. at counter-examples looked initially plausible, this, I think,
There are many ways the concrete particulars that make up was due to a confusion. When we hear the Gunnar-Ridley
our surroundings could be arranged that would be sufficient story, it seems correct to say that it follows from the story
for their obtaining. What Gunnar and Ryder can bring about that Gunnar is responsible for Ridley's death and that Gunnar
is which of these possible arrangements of particulars—which could not have prevented Ridley's death. But 'Ridley's death'
murderer, which road—the universals will be "realized in"; is ambiguous. If we are using this phrase to denote a universal,
that some arrangement or other of the particulars will realize then we may say that Gunnar could not have prevented
these universals in something totally outside their control; it Ridley's death, but not that he was responsible for Ridley's
is riot something they bring about. Here is an analogy involving death. If we are using this phrase to denote a particular, then
another sort of universal, properties: Chisel is a sculptor, we we may say that Gunnar was responsible for Ridley's death,
suppose, and sculpts the heaviest statue that ever was or will but not that he could not have prevented it.
be, The Dying Whale; thus Chisel brings it about that a certain This result might lead us to wonder whether Frankfurt's
particular, The Dying Whale, exemplifies the property of being counter-examples to the Principle of Alternate Possibilities
the heaviest statue;' but he does not bring it about that this rest on a similar confusion. Suppose we were to split the
property is exemplified, since, no matter what he had done, Principle of Alternate Possibilities into two principles, one
this property would "automatically" have been exemplified about "act-particulars" (event-particulars that are voluntary
by something or other. He causes something to exemplify movements of human bodies) and one about "act-universals"
this property, but he does not cause this property to be (that is, things that could be done by distinct agents, such as
exemplified. murder, prayer, or killing Jones at noon on Christmas Day,
In affirming (I), I do not mean to affirm the falsehood 1953); should we then see that Frankfurt's alleged counter-
(II) Gunnar did not bring about Ridley's death, examples to the Principle of Alternate Possibilities depend
for their plausibility on treating one and the same act as a
where 'Ridley's death' denotes an event-particular (the iden- particular at one point in the argument and a universal at
tity of which is determined by its causal antecedents), one another?
that is also perhaps denoted by 'Ridley's untimely death', I do not think that Frankfurt is guilty of any such confu-
`the only death Gunnar ever caused', and so on. Anyone who sion. The "acts" that figure in his counter-examples seem to
feels inclined to reject (I) should make sure that this inclination me to be treated consistently as universals. If this is the case,
does not arise from a failure to distinguish between (I) and it raises two questions. Let us split the Principle of Alternate
(II). To revert to our sculpture example, (I) and (II) stand to Possibilities into two principles in the way suggested in the
each other roughly as do the following: preceding paragraph: PAP1 (about particulars) and PAP2
Chisel did not cause the property of being the heaviest (about universals). The first question: if indeed Frankfurt's
statue to be exemplified "acts" are universals, he is arguing against PAP2; can his argu-
ment be met by considerations like those we raised in defence
and of PPP2? The answer seems to me to be "No", but I am not
Chisel did not cause (the particular thing that is) the at all sure about this. The considerations raised in defence of
PPP2 depended on our having at our disposal a fairly precise
heaviest statue to exist.
notion of "state-of-affairs universal", and I am not at present
The former is, as I argued above, true, and the latter false.' able to devise an equally precise notion of "act-universal"
180 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 181
that I find satisfactory. The second question: what about In order to deduce the free-will thesis from the thesis that
PAP1? I do not find this question interesting, since I do not there is such a thing as moral responsibility we shall need two
think that "event-particulars that are voluntary movements premisses in addition to our three principles.
of human bodies" are what we are talking about when we The first of these is:
talk about one's responsibility for one's acts. I shall not,
however, defend this thesis here. An adequate defence of it If (i) no one is morally responsible for having failed to
would be fairly complex, and I do not think my reasons for perform any act, and (ii) no one is morally responsible
thinking what I do on this matter are worth developing merely for any event, and (iii) no one is morally responsible for
to establish a negative conclusion, a conclusion, moreover, any state of affairs, then there is no such thing as moral
upon which none of our arguments depends. responsibility.

5.7 We have seen that three principles relating ability and This premiss is not a tautology, for there are arguably other
responsibility cannot be refuted by Frankfurt-style counter- things for which one can be held morally responsible than
examples: unperformed acts, events, and states of affairs. There are, if
nothing else, performed acts. (Of course, these may be a
A person is morally responsible for failing to perform a
species of event, but I should not want anything I say to
given act only if he could have performed that act;
depend on the premiss that acts are simply a type of event.)
A person is morally responsible for a certain event-parti- Could it not perhaps be that Gunnar is responsible for having
cular only if he could have prevented it; shot Ridley, though he is responsible for no event or state of
affairs? It might be thought that if he is responsible for having
A person is morally responsible for a certain state of shot Ridley, then it follows that he is responsible for C(Gunnar
affairs only if (that state of affairs obtains and) he could shoots Ridley). But this does not follow. For suppose that
have prevented it from obtaining. Gunnar has shot Ridley and is responsible for having done so,
but suppose that if Gunnar had decided not to shoot Ridley,
I shall argue that these three principles entail that moral something outside his control would have caused him to shoot
responsibility requires free will. That is, I shall deduce from Ridley. Then in every future that is open to him, Gunnar
these three principles the following conclusion: the thesis shoots Ridley, and, as we have seen, is therefore not respon-
that there is such a thing as moral responsibility entails the sible for its being the case that he shoots Ridley. Even in this
free-will thesis. By 'the thesis that there is such a thing as outré case, however, it would seem that Gunnar is responsible
moral responsibility', I mean the thesis that someone is for some state of affairs: he is responsible for C(Gunnar shoots
morally responsible for something—for some act, some event, Ridley without having been caused to do so by something
some state of affairs, the publication of some scurrilous pam- outside his control), for example. Moreover, he would seem
phlet, the currency of some malicious rumour, or the decline to be responsible for the event-particular, Ridley's death. I
of classical studies at some university. By 'the free-will thesis', am unable to construct a case in which it seems plausible to
it will be recalled, I mean the thesis that most people, if not say that an agent is responsible for some act but not for any
all people, are very often, if not always, in the following posi- state of affairs, and a fortiori, I am unable to construct a case
tion: when they are faced with having to choose among various in which it seems plausible to say that an agent is responsible
incompatible courses of action, each of these courses of action for some act but not for any state of affairs or any event or
is such that they can (are able to, have it within their power any unperformed act. I conclude that our first premiss is true
to) choose it. and in fact necessarily true.
182 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 183

The second premiss we shall need is: for the conclusion that this sense of 'free will' is the
sense this term has in other parts of your book. Mightn't
If (i) someone could have performed (i.e. had it within it be that some key term that figures in the definition of
his power to perform) some act he did not in fact perform, `free will'—`could have done otherwise', say—has more
or (ii) someone could have prevented some event that in than one sense? And mightn't it be that you have shown
fact occurred, or (iii) someone could have prevented that determinism entails that one "couldn't have done
some state of affairs that in fact obtains, then the free- otherwise" in sense A and that moral responsibility
will thesis is true.' requires that one "could have done otherwise" in sense B?
This premiss would be hard to deny. If any of the disjuncts of
its antecedent is true, then, surely, someone has been faced In Section 3.11 I briefly considered what was essentially
with incompatible courses of action that were severally open this objection. I attempted to meet it by the following device.
to him. Consider, for example, the first disjunct. Suppose that I invited the reader to consider the Third Argument—the First
Gunnar could have shot Ridley but didn't; then the course of Argument would have done as well--and to consider the
action shooting Ridley was open to him owing to the fact phrase 'No one has a choice about whether p', which occurs
that he could have shot Ridley; and the course of action not in that argument. I asked the reader to substitute the phrase
shooting Ridley must have been open to him since he took `No one has a choice about whether p, in just the sense of
that course of action. Similar arguments could be given for having a choice that is relevant to questions of moral responsi-
the other two disjuncts. bility' at each occurrence of the former phrase in the premisses
It is easy to see that our two premisses and our three prin- of the Third Argument. I then asked the reader to consider
ciples together entail the proposition that the existence of carefully whether any of these premisses appeared less
moral responsibility entails the free-will thesis. I am aware plausible on this new interpretation. It was my contention
that the argument of the present section that is intended to that the "new" premisses appeared no less plausible than
establish this thesis is trivial and might even be described as the "old".
pedantic. It has, however, one interesting feature: it does not In the present section, I shall attempt to meet this charge of
rely on the Principle of Alternate Possibilities but only on possible equivocation in another way. I shall argue "directly"
"similar" principles that cannot, so far as anyone knows, be for the incompatibility of determinism and moral responsi-
refuted by Frankfurt-style counter-examples. Therefore, we bility. That is, I shall argue that determinism is incompatible
may conclude that even if Frankfurt has shown that the Prin- with responsibility, and my argument will make no mention
ciple of Alternate Possibilities is false, it is nevertheless true whatever of free will: notions like could have done otherwise
that moral responsibility requires free will. and had a choice about will play no part in the argument of
Our "philosophical commonplace", then, retains its status, the present section. This argument will indeed be formally
and a major part of the answer to the question, "What would identical with one •of our arguments for the incompatibility
our not having free will mean?" must be: our not having free of free will and determinism—the Third Argument of Chapter
will would mean that moral responsibility does not exist. III—but the concept of free will will not figure in it. Now if
I am successful in showing that moral responsibility is incom-
patible with determinism and if my argument for this conclu-
5.8 To this argument someone might reply as follows: sion does not make any use of the concept of free will, it will
You have shown at most that in one sense of `free will', not follow from my success in this undertaking that my earlier
moral responsibility requires free will. But that is not arguments about "free will" and determinism, on the one
very interesting unless it is supplemented by an argument hand, and about "free will" and responsibility, on the other,
184 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 185

involve the same concept of free will. Nevertheless, it would human beings had ever been. The latter is obviously true,
be highly implausible to suppose otherwise and I doubt since no human being is responsible for the laws of nature.
whether anyone who accepted the argument I am about to (For example, if it is a law of nature that nothing travels faster
give for the incompatibility of responsibility and determinism than light, then no human being is morally responsible for
would feel any temptation to think that the term 'free will' the fact that nothing travels faster than light.) Our argument
has been used in the earlier sections of the present chapter in is inscriptionally identical with the Third Formal Argument.
a sense different from the sense in which it was used in earlier We begin with the consequence of determinism with which
chapters. the Third Formal Argument began:
The argument, pretty clearly, is sound if and only if the
Third Argument is sound. We shall proceed by redefining the (1) ❑ (Po & L. D P).
modal symbol 'N' that figured in that argument. `Np' was read ('P', we remember, is a dummy letter for which any sentence
'p and no one has, or ever had, any choice about whether p'. expressing a truth may be substituted.) From (1) we may
We shall now read `Np' as 'p and no one is, or ever has been, deduce
even partly responsible for the fact that p'. 27 Our argument
will depend upon two inference rules that are inscriptionally (2) ❑ (Po D (L D P))
identical with the rules (a) and (i3) of Section 3.10:
by elementary modal and sentential logic. Applying rule (A)
(A) op Np to (2), we have
(B) 1\1(p Dq), Npi—Nq.
(3) N(13 3 (L P)).
0
The validity of Rule (A) seems to me to be beyond dispute.
No one is responsible for the fact that 49 X 18 = 882, for the We now introduce a premiss:
fact that arithmetic is essentially incomplete, or, if Kripke is (4) NP,,.
right about necessary truth, for the fact that the atomic
number of gold is 79. (Doubtless it is true in some sense of From (3) and (4) we have by rule (B)
`because' that (A) is valid because (a) is valid; doubtless the
fact that no one has any choice about the truth-value of a (5) N(L J P).
necessary truth is the reason no one is responsible for the We introduce our second premiss:
truth-value of a necessary truth. But whatever the reason for
the validity of (A) may be, it is unquestionably valid.) The (6) NL.
validity of (B) is a more difficult matter. Let us return to (B) Then, from (5) and (6) we have by (B)
after we have presented our analogue of the Third Formal
Argument. (7) NP.
This argument will require two premisses, '1\111 ' and 'NL'.
0

(As in Section 3.10, the symbol 'Po' abbreviates a sentence I have called this an argument. More precisely it is an argu-
that describes the state of the world at some time before ment-form. We may derive indefinitely many arguments from
there had ever been any human beings, and the symbol it by substituting arbitrary sentences for 'T. If the sentence
abbreviates a sentence that expresses the conjunction into substituted for is a truth, and if determinism is true, the
a single proposition of the laws of nature.) The former of substitution-instance of (1) so obtained will be true and the
these premisses is obviously true, since no human being is argument so obtained will be sound (assuming, of course,
morally responsible for anything that occurred before any that it is valid). This fact about our argument-form amounts
186 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 187
to a proof of the following proposition: substitute any truth The prospect of showing (B) to be valid by reducing it to
you like for '13 ' in the following schema generally accepted inference rules seems, if anything, even
If determinism is true, then no one is, or ever has been, bleaker than the prospect of showing (0) to be valid by that
morally responsible for the fact that P method. It is hard to see how an inference rule essentially
involving the concept of moral responsibility could be reduced
and you will get a truth. For example, if you substitute to inference rules involving only non-moral concepts. And it
`Kennedy was assassinated', The US used atomic weapons is hard to think of generally accepted rules of inference that
against Japan', or 'Nixon received a pardon' for 'P', you will essentially involve moral concepts. The familiar principle that
get a truth. This result, I think, may be properly summarized ought implies can may be such a rule, but there are few if any
in these words: determinism is incompatible with moral re- others. And while this familiar principle might be of some use
sponsibility. in proving the validity of (B) if we were allowed to assume the
We have proved this result provided that the reasoning validity of (0), this is not something we can properly assume.
employed in our argument-form is valid; that is, provided I think the only thing I can do to persuade the reader that
that both (A) and (B) are valid; that is—since the validity of (B) is valid is to ask him to consider carefully such instances
(A) is beyond dispute—provided (B) is valid. Let us now turn of it as
to the question of the validity of (B).
Since (B) is formally identical with (0), we might expect N John was bitten by a cobra on his thirtieth birthday
that anyone who attempted to answer this question would N (John was bitten by a cobra on his thirtieth birthday)
encounter obstacles similar to those we encountered in our John died on his thirtieth birthday)
attempt in Section 3.10 to answer the question whether (3) is
valid. This is indeed the case. hence, N John died on his thirtieth birthday
The purely formal argument for the validity of (0) presented
in 3.10 will, of course, apply to (B): if there is any set of and
possible worlds W such that 'Np' can be thought of as making N Plato died in antiquity
the assertion that p is true in every member of W, then (B)
will "come out" valid; if there is any set of worlds that stands N (Plato died in antiquityD Plato never met Hume)
to 'Np' as the set of all worlds in which the laws of nature hence, N Plato never met Hume
hold stands to 'it is physically necessary that p' and as the set
of all morally permissible worlds stands to 'it is morally re- and to attempt to construct counter-examples to it. (I am
quired that p', then the validity of (B) follows. And it is not asking, of course, for counter-examples that can be seen to be
wholly implausible to suppose that there is such a set; perhaps such without assuming that determinism is compatible with
the set containing the actual world and all and only those moral responsibility.) I believe that the plausibility of (B),
worlds such that human beings can be held morally responsible like the plausibility of (p), is best appreciated by someone
for their non-actuality. But I have no way of showing that who has made a serious and sustained attempt to show that it
there is some set of worlds—whether it is the one mentioned is invalid. 28
in the preceding sentence or some other—that bears the re- One might at this point want to raise the question whether
quired relation to `Np'. It seems to me obvious that there my use of (B) begs the question against the proponent of the
must be some such set, whether or not I have correctly named compatibility of moral responsibility and determinism. But
it, but the compatibilist will presumably deny that there is what I should say in response to this question so perfectly
any such set, and I have no argument to use against him. parallels what I have said in response to the question whether
188 WHAT OUR NOT HAVING FREE WILL WOULD MEAN WHAT OUR NOT HAVING FREE WILL WOULD MEAN 189
my use of (0) begs the question against the proponent of the to reopen the question of compatibilism only if we can
compatibility of free will and determinism, that I shall simply discover arguments for determinism whose premisses are
refer the reader to Section 3.10 and ask him to read it mu tatis more plausible than the premisses of our arguments for
mutandis. I will, however, risk wasting space and write out the incompatibilism.
result of making a certain set of mechanical substitutions in In Chapter VI, we shall take up the question "What argu-
the final paragraph of 3.10. I have now given an argument for ments can be given in support of determinism?" and we shall
the incompatibility of moral responsibility and determinism. compare the premisses of these arguments with the premisses
If the compatibilist (sc. with respect to responsibility and from which we have deduced incompatibilism, and with the
determinism) wishes to refute this argument—and, of course proposition that there is such a thing as moral responsibility.
nothing obliges him to do this—here is what he will have to If we do not reject the reality of moral responsibility, then
do: he will have to produce some set of propositions intui- we shall have to reject either incompatibilism or determinism.
tively more plausible than the validity of (B) and show that I shall try to discover which of these courses is the best.
these propositions entail the compatibility of moral responsi-
bility and determinism, or else he will have to devise a counter-
example to (B), a counter-example that can be evaluated
independently of the question whether moral responsibility
and determinism are compatible.

5.9 I take the arguments of Sections 5.3 through 5.8 to

This book is a defence of the thesis that free
establish that moral responsibility .requires free will in just will and determinism are incompatible, and
that sense of free will in which it was argued in Chapters III
an exploration of some of the consequences
and IV that free will is incompatible with determinism.
The main conclusion of the present chapter, then, is that of this thesis. Free will is understood as the
power to act otherwise than one in fact does,
our not having free will would mean that we are morally re-
sponsible for nothing. The strongest argument for the existence and determinism is understood as the thesis
that the past and the laws of nature together
of free will seems to me to be this: moral responsibility
requires free will and we are responsible for at least some of determine a unique future. The author
the things we have brought about. In Chapter VI, I shall argue argues that determinism is incompatible
with free will because determinism entails
that this is the only strong argument for the existence of free
that one's present acts are determined by
will. But we have also seen that free will is incompatible with
factors outside one's control.
determinism. Suppose there were arguments for determinism Several arguments in favour of the
that were rationally more compelling than this argument for compatibility of free will and determinism
free will. If this were so, then we should be rationally com- are examined and rejected, the most
pelled to reject the thesis of the reality of moral responsibility. important being the argument that free will
Such a conclusion might lead us to reopen the question of
the compatibility of free will and determinism. For, it might in fact entails determinism, since if one's acts
were undetermined by one's past, they
he argued, however plausible the premisses of our various would be mere random occurrences. The
arguments for incompatibilism may be, their denials are not
author goes on to argue that moral
nearly so implausible as the thesis that there is no such thing
responsibility requires free will; and that,
as moral responsibility. But this consideration should lead us since the reality of moral responsibility is not
in doubt, and since there is no .good reason to
 THE TRADITIONAL PROBLEM 191

be carefully distinguished from the thesis of universal causa-

Chapter VI tion, the thesis that every event—or state of affairs or what
have you—has a cause. Even if I am wrong in my belief that
The Traditional Problem universal causation does not entail determinism,' that this
entailment holds is certainly a substantive philosophical thesis.
And I doubt whether anyone would think that universal
6.1 In Chapter I, I .argued that the problem of free will and causation was incompatible with free will unless he thought
determinism was best looked upon as two problems: the that universal causation entailed determinism.) It will be con-
Compatibility Problem and the Traditional Problem. Chapters venient to present our enquiry as a commentary on two as-
III and IV were devoted to the Compatibility Problem and its sertions: 'Science shows determinism to be true' and 'Reason
ramifications. That is, in those two chapters we attempted to shows determinism to be true'.
answer the question whether free will and determinism are
compatible. We concluded that they are in fact incompatible. (0 Science shows determinism to be true. This simply does
This conclusion has the consequence that the Traditional not seem to be the case. If anything, quite the opposite is
Problem exists, a proposition that all the most acute thinkers true, since the standard interpretation of quantum mechanics
about free will, from the seventeenth century to the present, is indeterministic. This bald statement requires a variety of
have been at great pains to deny. (I am not happy about qualifications and comments.
enlisting in a war in which the enemy forces are commanded (a) One sometimes hears it said that quantum-mechanical
by Hobbes and Hume and Mill. But I must follow the argu- indeterminacy applies only to unobservable, submicroscopic
ment where it leads.) In the present chapter I shall do what events and not to the observable events of everyday life. This
I can about the Traditional Problem. I do not know how to is quite false: quantum mechanics is a physical theory--as
solve it. That is, I do not know how to find out whether we opposed, say, to a biological or geological theory—and is thus
have free will. There is, I believe, one good argument for the as legitimately applied to the behaviour of an elephant or a
existence of free will, and no good argument against it unless glacier as to the behaviour of a neutrino or a positron. It is
there should be some good argument for determinism. I believe, true that the predicted indeterminacy in the behaviour of an
moreover, that there are no good arguments for determinism elephant or a glacier is smaller by many orders of magnitude
and that there are some rather good arguments against it. The that anything that could be measured by the most marvellous
preceding two sentences sum up what I believe to be the instruments that are dreamt of in our science fiction. Never-
"dialectical situation" in which the philosopher who con- theless, it is there.
fronts the Traditional Problem finds himself. The body of this More important than this, however, is the fact that the
chapter will be a defence of the thesis that this description of domain of the submicroscopic and the domain of the observ-
such a philosopher's dialectical situation is an accurate one. able are not causally isolated from each other. This is, in a
sense, a truism, since the behaviour of the observable presum-
6.2 Let us turn to the question of determinism. That is, let ably supervenes on the behaviour of the submicroscopic.'
us ask what reasons there are for thinking that determinism is But there is a stronger and more interesting sense in which
true and what reasons there are for thinking that determinism these two domains are not isolated from each other: individual
is false. Determinism, we remember, is the thesis that the past submicroscopic events can "trigger" observable events. One
and the laws of nature jointly determine a unique one among device that actually effects such a "triggering" is a Geiger
the possible or internally consistent futures to be the future, counter. The passage of a single alpha particle or electron or
the actual future. (Let us remember that determinism must
burst of high-energy electromagnetic radiation through a tube
192 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 193
of gas can sufficiently ionize the gas in the immediate vicinity even with the more modest doctrine that the past states of
of the path of the intruder to produce an effect that—after the universe, whether observable or unobservable, determine
suitable amplification, which a Geiger counter is designed to a unique succession of future observable states.
provide—registers as an audible click; and any such particle or (b) In classical physics, the momentary state of a physical
burst will have been produced by a change in the nucleus or system (or, at any rate, of a point mass) is represented by a set
the orbital electrons of a single atom. These facts may, I think, of vectors—say, the set containing the position and momentum
be fairly summarized by saying that an individual submicro- vectors of the system. In quantum physics, however, the
scopic event is the cause of the click that the Geiger counter momentary state of a system is represented by a function,
produces. Given this feature of Geiger counters, it would be the 1,U-function of the system. Just as the laws of classical
easy enough to construct devices whose behaviour is undeter- mechanics determine the way in which a classical system will
mined (provided, of course, that the quantum theory is change its state with the passage of time (in response to im-
actually right). We might, for example, build a time bomb pressed forces), so the laws of quantum mechanics determine
incorporating a radioactive source, a Geiger counter, and the way in which a quantum-mechanical system will change
a firing mechanism designed to take a Geiger counter's "out- its state with the passage of time. I use the word 'determine'
put"—clicks or whatever—as "input". Such a bomb might be quite deliberately: given the quantum-mechanical state of
designed to explode if the counter, for example, clicked five a closed system (such as the entire physical world) at a given
times within any ten-second interval. Suppose we were to time, only one "future"—one sequence of quantum-mechanical
build such a bomb and arm it; suppose the obliging God of states—is possible for that system.' Thus, it would seem,
the Philosophers were to provide us with a thousand perfect quantum mechanics is a deterministic theory.
duplicates of it—the finest duplicates that omnipotence could There is some sense in describing matters this way. But if
contrive, this duplication, of course, being understood to we describe them this way, then we must distinguish between
include the setting of the arming mechanism. What would the question whether quantum mechanics is a deterministic
happen? What would not happen, save on the barest chance, theory in the sense of the preceding paragraph and the question
is the simultaneous explosion of all the bombs. Or at least this whether quantum mechanics entails determinism in the sense
would not happen unless the very basis of current physical of Chapter III. Though quantum mechanics may be a deter-
theory is a ghastly illusion. To say this is not to place limits ministic theory, it seems to me to be wholly implausible to
on the possible accomplishments of omnipotence, or not suppose that its truth entails determinism. This would be
unless one would place such limits if one said that even an a plausible supposition if all the properties of a physical system
omnipotent being could not make a functional aeroplane at a moment were determined by its quantum-mechanical
entirely out of cheese and whipped cream. For just as cheese state at that moment. This, however, does not seem to be the
and whipped cream are materials intrinsically unsuited to the case, since the properties a macroscopic object can actually
needs of aeronautical engineering, any physically possible be observed to possess at a given moment do not in general
material is intrinsically unsuited to the task of building a time follow from the quantum-mechanical state of that object
bomb that is both deterministic and constructed in the way (that is, the sum of the 1P-functions of all its constituents)
specified earlier in this paragraph. Actual matter, matter that at that moment. Rather, there is in general only a statistical
obeys the rules of quantum mechanics, is intrinsically incap- correlation between an object's being in a certain quantum-
able of carrying within itself the perfectly determinate dispo- mechanical state—whether that object is an observable or
sitions to future behaviour that strict determinism requires. a submicroscopic object—and its possessing a determinate
It would seem then, that the truth of quantum mechanics observable or measurable property. Consider Laplace's omni-
is incompatible not only with determinism generally, but scient Intelligence. Even if it were possible for this Intelligence
194 THE TRADITIONAL PROBLEM THE TRADITIONAL PRLIBLEM 195

to know the quantum-mechanical state of the entire world at by a deterministic theory. (Einstein, for example, felt this
the present moment, the most that follows from the fact that way.) There are two ways in which such a supersession might
quantum theory is a deterministic theory is this: the Intelli- come about.
gence would know the quantum-mechanical state of the world First, quantum mechanics might be superseded by some
at all future moments. But from the supposition that the theory that is incompatible with it. Presumably this would
Intelligence will know the quantum-mechanical state of the have to be a novel and revolutionary theory, the postulates of
world in, for example, the year 2000, one cannot infer that which were unlike those of quantum mechanics "from the
the Intelligence will know whether I shall be alive or dead in ground up". No one, of course, can say that this couldn't
that year or where I shall be or what philosophical problems happen. But it is worth noting that no such theory has
will be exercising me. At most, I think, the Intelligence would appeared; quantum mechanics has no serious competitor.'
be able to assign probabilities to various alternative hypotheses Secondly, quantum mechanics might be superseded by some
about my condition in that year. As to whether any given theory that is compatible with it but more comprehensive.
reasonably specific hypothesis would receive a significantly Einstein's treatment of the statistical theory of Brownian
higher probability-assignment than those equally specific motion provides an example of the sort of possibility I am
hypotheses with which it is in competition, I have no idea. thinking of. In a classic paper, Einstein showed that the
Certainly the Intelligence could not make any very detailed statistical laws describing the odd path (a "drunkard's walk")
predictions about the undetermined time bombs we considered taken by A dust particle suspended in a fluid (Brownian
earlier; he would be able to tell us only things like this: "The motion) could be derived from certain plausible theses con-
probability that the bomb will explode within the next hour cerning the agitation of the particle by the thermal motion of
is .7362". Thus even the Intelligence who Laplace thought the molecules that fluid was composed of; he showed this,
would have no use for the concept of probability would, in moreover, quite independently of the question whether these
a world that quantum mechanics gives a correct description thermal motions were themselves deterministic. In this sense,
of, find this concept indispensable. then, Einstein showed how a statistical theory could be "em-
That is to say, he would find this concept indispensable if bedded in" a deterministic theory (the essentially Newtonian
he were interested in properties other than those that belong theory of the mechanics of thermal motion). Cases of this
to a system in virtue of its quantum-mechanical state. It is sort might lead one to wonder whether something of the
just possible that someone might maintain that, in some sense, same sort might be the case with quantum mechanics. Might
only quantum-mechanical properties are "real": that a correct there not be a deterministic micro-microcosm that underlies
picture of ultimate reality would have room only for quantum- the apparently indeterministic behaviour of the microcosm in
mechanical properties and must exclude any such properties a way analogous to the way in which molecular motion, which
as being red, being in motion, or thinking about quantum may be presumed for the sake of the example to be deter-
mechanics. A philosopher who took this position could con- ministic, underlies the apparently indeterministic Brownian
sistently say that quantum mechanics'is a correct theory of motion? Or, to put the question more formally, might it not
physical reality and that determinism is true. But such an be possible to embed quantum mechanics in some deter-
extreme position is unlikely to win many followers, and I ministic theory,' in a way analogous to the way in which
shall not discuss it further. Einstein embedded the statistical theory of Brownian motion
(c) Many philosophers and many scientists have had a hard in classical mechanics? (Deterministic theories in which
time accepting quantum mechanics as anything more than quantum mechanics might be embedded are often called
a stopgap, a theory that works well and is certainly better than "hidden variable" theories, the "hidden" variables being
nothing, but which somehow must eventually be superseded the variables of the theory that represent the underlying
196 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 197
deterministic features of a physical system that go unrepre- whose possible existence is being investigated. The assumptions
sented in quantum mechanics.) that have been made are attractive and seemingly self-evident;
The issues this question involves are very difficult. As I but attractive and seemingly self-evident assumptions have
understand them—and I understand them hardly at all—the had to be rejected on several notable occasions in twentieth-
answer seems to be "Probably not". Von Neumann was once century physics. But even if it is possible to embed quantum
widely believed to have shown that the mathematical structure theory in a deterministic theory, no one at any rate has
of quantum theory—unlike the mathematical structure of the done this in any way that is generally agreed by experts in
statistical theory of Brownian motion—is such as to prevent quantum theory to be both logically satisfactory and "of
its being embedded in a deterministic theory. That is, he was physical interest". Therefore, it would seem, the most reason-
widely believed to have shown that any theory treating of the able thing for the non-expert to believe at the present time is
same subject-matter as quantum mechanics, that could be that it is at least unlikely that quantum mechanics will be
embedded in a deterministic theory, must make different pre- embedded in a deterministic theory, and that it may well be
dictions about observables from quantum mechanics, predic- impossible for this to happen.
tions that are known empirically to he false. The history of
the reception of von Neumann's argument has been a tortuous These qualifications and comments having been made and
one, which I shall not attempt to summarize.' The argument duly noted, it still seems reasonable to say that science shows
was never accepted by Einstein or by de Broglie, and has often determinism to be false and it seems plainly impossible to say
been received with rather more rhetoric than is usual with a that science shows determinism to be true. But some philo-
piece of mathematics. The following passage by two disciples sophers depreciate this conclusion. I have heard philosophers
of de Broglie argue more or less as follows:
.. , the great mathematician von Neumann ... even produced a theorem Quantum mechanics is all very well, But no one seriously
stating that it would be impossible to picture any hidden determinism suggests that "trigger" mechanisms like your time bombs
which would reproduce the statistical laws of [quantum] mechanics. are among the antecedents of human action. And if no
This strange theorem claims that theory will not only explain all the such mechanisms are parts of us, then the degree of in-
observed phenomena, but will also, by way of its own internal logic,
preclude any other explanation. However, de Broglie carried out a determinacy that quantum mechanics assigns to human
critical analysis, and showed that von Neumann's theorem was based thoughts, words, and deeds is so small as to be experi-
upon an ad hoc hypothesis which in itself rendered the whole theorem mentally meaningless. A human being might, therefore,
valueless. We still cannot rule out the possibility of a hidden determinism' be as much a deterministic system as, say, a digital com-
represents an extreme but far from unique reaction. (I doubt puter, even if quantum mechanics is correct and essen-
that many of the people competent to judge these issues tially indeterministic. And a computer (unless its de-
signers deliberately incorporate a randomizing mechanism
would agree without reservation with the claim the authors
into it) is certainly the sort of thing that is "for all
make on behalf of the analysis carried out by their maitre.)
practical purposes" deterministic, albeit it may, for
Current thinking on these matters is hard to pin down.
theoretical reasons, be held to be subject to some im-
I believe it comes to something like the following: (i) there measurably small degree of indetermination. But surely
are mathematical lacunae in von Neumann's argument;
the great achievements of brain physiologists in under-
(ii) other, rigorous arguments for the same conclusion have
standing the functioning of the brain (our "hardware")
been constructed; 8 (iii) even the most rigorous of these argu- and the equally great achievements of psychologists in
ments necessarily depends upon certain assumptions about
understanding our cognitive functioning (our "soft-
measurement and about the nature of the "hidden variables" ware") make only one conclusion possible:_ a human
198 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 199

being, like a computer, is "for all practical purposes" of such a thesis. If astrology should turn out to be true, then
a thing whose future behaviour is wholly determined by the world would have to be a very different sort of world from
its past states and its current "input". the world that science forces those of us who attend to it to
think it is. (I am not talking about the possibility that there
I accept, and regard as important, the distinction between might turn out to be a grain of truth in astrology; I am talking
strict determination and determination "for all practical about its turning out to be true that there is nothing bul truth
purposes". And I have no doubt that if strict determination in astrology.) The ultimate truth of astrology is a possibility
is, as I have argued, incompatible with free will and moral that is perhaps logically consistent with the totality of my
responsibility, then determination in this very slightly weaker sensory experience to date; nevertheless, if astrology should
sense is also incompatible with them. It would, in fact, be an turn out to be true, then I could only say that the world was
easy task to adapt the arguments of Chapter III to show that an epistemic cheat, a world that had been rigged, like a crooked
"near determinism" is incompatible with free will, though roulette wheel or a magician's props, to effect a most unfair
the adapted arguments would be comparatively untidy. But divorce of appearance from reality. But suppose it should turn
as for Ijile "great achievements" argument, I can only call it out that the behaviour of human beings is not determined.
a bluff. I have a very hard time seeing why so many philoso- Should we have any reason to regard the world as a•cheat?
phers seem to think that the results of the empirical study of Suppose, to make this question a bit less abstract, that it
human beings lend support to the hypothesis that human turns out that the electrical resistances across certain synapses
behaviour is determined. (In the sequel, I shall use 'determined' in the brain occasionally change in undetermined ways and
to mean 'always determined, for all practical purposes, by past that these changes are of sufficient magnitude and sufficiently
states and present stimuli'.) I do not mean that I believe that frequent to render false the hypothesis that human behaviour
empirical investigations have shown that human behaviour is is determined.' What physiological theories would have to be
not determined. My difficulty is simply this: the human discarded? What well-established experiments would have to
organism and human behaviour are such terribly complex be regarded as having produced misleading results? What parts
things, and so little is known about the details of that terrible of our textbooks on the structure and functioning of the brain
complexity (in comparison with what there is to be known), would have to be rewritten? The answer, it seems to me, is
that it is hard to see why anyone should think that what we that the discovery I have imagined would reveal no defects
do know renders a belief that human behaviour is determined whatever in our theories and our textbooks, simply because
reasonable. I can only conclude that these philosophers are no scientist has ever claimed to be able to predict the actual
convinced on a priori grounds, or perhaps on no real ground course of events in a living brain, just as no meteorologist has
at all, that human behaviour is determined, and, owing to this ever claimed to be able to predict next week's weather in
conviction, are predisposed to regard very nearly anything as complete detail.
evidence in support of it. (If this is so, it should not surprise Let us consider. an analogy. Suppose we should somehow
us. Similar cases abound. Many theists are predisposed to find out that demons have for centuries been influencing
overrate the evidential value of apparent design in nature, and human behaviour by introducing minute changes in human
many atheists to overrate the evidential value of pain and brains—by introducing, in fact, changes in the electrical resist-
suffering.) ances across certain synapses. I think it should be pretty clear
Let us look at it this way. There are various theses, which, that none of our investigations into the functioning of the
if they were to turn out to be true, could only lead the edu- brain renders this supposition in the slightest degree improb-
cated person to protest that the world was, epistemically able. If we learned this we should indeed know something
speaking, very unfairly arranged. Astrology is a good example new about the sources of human behaviour, but none of our
200 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 201
painfully gathered information about the functioning of the seventeenth century: all the gleanings of the empirical sciences
brain would have to be discarded or have to be modified in of man are simply irrelevant to the question whether it is
any way. No one would have any right to exclaim, "But this a cogent argument.
goes contrary to what we believed we had discovered about I do not say that in no case could we have decisive empirical
the brain!" Now it is doubtless true that we are perfectly well evidence for the thesis that a thing was "for all practical pur-
justified in assigning a negligible subjective probability to the poses" determined. If I had never before seen a watch and one
hypothesis that demons have been affecting our brains in the day pitched my foot against one when crossing a heath, I
way I have imagined. But this justification does not rest on daresay I could easily enough discover that its behaviour was
our empirical discoveries about the brain. (The question determined by its prior states and the momentary external
"Where does it rest, then?" is one I shall leave to epistemolo- influences to which it was subject. I should think that the
gists.) If we should learn that there are many intelligent races results of such an investigation would be as reliable as any
in the universe, exactly half of which are afflicted with mani- results ever are that require high-level abstractions like 'deter-
pulative demons, all our enormous learning about the structure mination' and 'state' for their statement, and that they could
and functioning of the brain would be of no help whatever to be called empirical investigations with complete propriety.
us in deciding whether our own race was so afflicted, for the But we are not watches. Determined or not, we are too com-
proximate effects of the demons' manipulations would be plicated for our internal workings to be surveyed like those
lost in the immeasurably vast living wilderness of the brain. of a watch.
Surely the case is the same if we are asking not about I do not say that we could not have empirical evidence for
demonic influence but about undetermined change. Suppose the thesis that a man was "for all practical purposes" deter-
God were to tell us that the brains of half the races in the mined. Perhaps in the future it will be possible to construct
universe were determined and that the brains of the other half computer models of particular brains that will enable us to
were subject to minute undetermined changes of the sort I predict how a particular person will react to various stimuli.
have imagined. How should we find out which sort of race If these stimuli were of complexity comparable with that of
we were? There seems to be no a priori reason to think that the stimuli we encounter in everyday life, and if the predic-
there would be any difference between the brains of the two tions were always, or almost always, right, then we should
sorts of race that could be detected in practice. It therefore have very good reason indeed to believe that human beings
seems quite unreasonable to say that we have any empirical were determined. And, of course, there are almost certainly
reason for saying that we are determined. other sorts of evidence, sorts of evidence I cannot imagine,
I do not say that we have no reason of any sort for saying that would render highly probable the hypothesis that we are
we are determined. I am saying that any reason we do have determined. But no such evidence ever has been collected,
must be at least partly an a priori or conceptual reason. For and, even if human beings are determined, it looks like being
example, someone might say that we are determined because, a very long time before anyone ever does collect it.
if we weren't, our behaviour would be a meaningless, random To recapitulate: there would seem to be two ways in which
series of jerks and jumps and our speech mere babble. I have scientific evidence could convince us that we are determined;
said in Chapter IV why I think this contention is wrong. My first, we might believe this, as Laplace did, on the basis of our
present point is that it is not based upon any empirical investi- most general, physical theories (which apply to all physical
gation of our functioning, but rather upon an identification systems and hence to us); secondly, we might believe this on
of an undetermined sequence with a random sequence that is, the basis of the empirical study of man. But our most general
presumably, made on purely conceptual grounds. This argu- physical theories are no longer deterministic. And the empiri-
ment is no better and no worse today than it was in, say, the cal study of man has a long way to go before it will be in
202 THE TRADITIONAL PROBLEM
THE TRADITIONAL PROBLEM 203
a position to tell us anything about whether we are or are not called by this name must have. First, if x is a sufficient reason
determined. for y, then x must entail y. That is, it must be impossible for
x to obtain without y's obtaining. For if it were possible for
(ii) Reason shows determinism to be true. There are philoso- x to obtain and y to fail to obtain, how could the obtaining
phers who believe that determinism is a truth of reason. That of x be a sufficient reason for the obtaining of y I Secondly,
is, there are philosophers who claim to see, by some sort of no contingent state of affairs may be its own sufficient reason.
conceptual or ontological insight, that the laws of nature must This would seem to be an essential feature of the concept of
be sufficiently "tight" that, at any given moment, there is only a sufficient reason. (I introduce the qualification 'contingent'
one physically possible future. I do not share this insight and in order to accommodate those who hold that a necessary
I doubt whether it really exists. Moreover, I know of no very state of affairs is its own sufficient reason. Whether or not
plausible argument that proceeds from prcmisses known a this is so will make no difference to our argument.)
priori to the conclusion that the world is wholly determined!' We may now show that PSR leads to the collapse of all
There is one principle from which determinism can be de- modal distinctions. Let P be the conjunction of all contin-
duced, however, about which I have something more useful gently true propositions into a single proposition.' (In what
to say than "I don't believe it". This is the famous Principle follows, I shall use `P' indifferently to denote the conjunction
of Sufficient Reason (PSR): of all contingently true propositions and the state of affairs
For every state of affairs that obtains, there is a suf- that consists in the truth of this proposition. In the language
ficient reason for its obtaining. of Section 5.6, this vast state of affairs would be denoted by
the result of prefixing 'its being the case that' to some sentence
(In this principle, 'state of affairs' is to be understood in the expressing P. This state of affairs is what Plantinga calls `the
sense of Section 5.6.) This principle may plausibly be held to actual world'. What I call 'the actual world' in Section 3.6 is
entail determinism. For suppose determinism is false. Then the possibility that P be true.) It is evident that P itself is
there is a time following which there are at least two physically a contingent proposition, for a necessary proposition may
possible futures. But one of these must eventually come to not have even a single contingent conjunct. Now, according
pass; and it is hard to see what the "sufficient reason" could to PSR, there exists a state of affairs S that is a sufficient
be for the coming to pass of that particular future. Now this reason for P. S must be contingent or necessary. But it cannot
reasoning is far from being watertight. Someone might suggest, be either. It cannot be necessary, for, if it were necessary then
for example, that the actual future became actual not for any P (which, by our first principle, is entailed by S) would be
reason to be found in the natural world but rather because necessary. It cannot be contingent, for if it were contingent,
God chose that it should, God's choice being in that case the it would be a conjunct of P; and if it were a conjunct of P it
sufficient reason demanded by PSR. It is clear, therefore, that would be entailed by P; and if it were entailed by P, it would
we should need certain supplementary premisses to deduce both entail and be entailed by P; and if it both entailed and
determinism from PSR. However this may be, PSR must be were entailed by P, it would be P (by the criterion of identity
rejected, for it has an absurd consequence: the collapse of all for states of affairs given in Section 5.6'); and if S were P,
modal distinctions. then a contingent state of affairs would be its own sufficient
In order to see this, we must take a brief look at the concept reason, contrary to our second principle. Since S cannot be
of a sufficient reason. I do not know how to give an adequate either necessary or contingent, it cannot exist, and PSR is false.
account of this concept—if indeed any concept attaches to This result follows provided we assume that there is such
the words 'sufficient reason'—but I think I can see certain a thing as P—that is, the conjunction of all contingently true
features that any coherent concept that could reasonably be propositions. And there is such a thing as this if there arc any
204 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 205

contingently true propositions. Hence if PSR is true, there proposition the unthinking acceptance of which is a presup-
are no truths but necessary truths: there is no distinction to position of that activity, then reflective, theoretical assent to
be made between truth and necessity." Now many great that proposition will be common, if not universal, among the
philosophers have believed this—Spinoza, for example. But members of that race. Philosophers belonging to that race
I doubt whether many present-day philosophers could bring may, consistently with my thesis, occasionally raise questions
themselves even to consider this thesis seriously. I know I about how these propositions that are presupposed by very
couldn't. (For one thing, given the rule of inference that was nearly every aspect of their lives are to be justified; some
called (a) in Section 3.10, it would entail that no one had philosophers may even, consistently with my thesis, verbally
free will.) We must therefore reject PSR, and with it the only reject these propositions!'
plausible attempt to show that determinism is a truth of reason. It is because the proposition that we have free will is insep-
Since, as we have seen, neither empirical science nor pure arably bound up with our deliberative life, in my view, that
reason shows determinism to be true, or even provides us most of us are certain we have free will. But if this is indeed
with any good reason for thinking determinism true, we can the correct explanation for our certainty, then this certainty
only conclude that a belief in determinism is—at least at the is without evidential value. Our certainty about our own con-
present time—wholly unjustified. scious mental states may proceed from the fact that each of
us is, necessarily, in a privileged position in any dispute about
6.3 What reasons are there for thinking that we have or that his mental states; .but no one is necessarily in a privileged
we lack free will? position in a dispute about whether he has free will. If each
I do not think that there is any way we can simply find out of us has implanted in his brain a Martian device like the one
whether we have free will in the sense in which we can find we have imagined, then it is the Martians, and not we, who
out whether there is life on Jupiter. Some philosophers be- are the experts on the question whether we have free will.
lieve that it is possible to find out whether we have free will Is there any way other than introspection for us to find
by introspection. But this seems just obviously wrong, since, out whether we have free will? Well, if determinism is true,
if it were right, we could find out by introspection whether then we might find out that we have no free will by finding
we were fitted out with Martian devices like those we imagined out that we are determined. But the difficulties that attend
in our discussion of the Paradigm Case Argument in Chapter IV. finding out whether we are determined have already been
And, of course, we cannot do this. discussed. Moreover, even if determinism is false, it seems to
It is certainly true that most of us are perfectly certain that be mere wishful thinking to suppose that we shall find out
we have free will. But there is no reason to think that our that we have free will by finding out that determinism is false.
perfect certainty on this matter derives from our having, in First, it is unlikely that we ever shall find out that we are
some sense, "direct access" to the springs of action in the undetermined (in the sense of Section 6.2), even supposing
way some philosophers believe we have direct access to our that we are undetermined: we are simply too complicated for
own mental states. It rather derives, I should think, from our such a discovery to be likely. Secondly, if we did find out
knowledge, in most cases inarticulate, that one cannot deli- that we were undetermined, this discovery would not show
berate without believing in one's own free will. It seems to that we have free will. I have argued in various parts of this
me to be reasonable to suppose that for any race of rational book that our being undetermined is a necessary condition
beings, and for any type of activity that is inextricably en- for our having free will and is not a sufficient condition for
twined with very nearly every aspect of their lives—as deliber- our not having free will. But I have never said that our being
ation about future courses of action is inextricably entwined undetermined is a sufficient condition for our having free will.
with very nearly every aspect of our lives—if there is some I have not said this because it is false: the proposition that it
206 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 207

is undetermined whether I shall raise my hand at a certain and that he could have helped doing it—we react in certain
time does not entail the proposition that I have it within my characteristically human ways: we blame, we remonstrate, we
power to raise my hand at that time. Suppose, for example, hate, we reflect on the futility of hate, we plan revenge, we
that it is physically possible that I shall raise my hand one remind ourselves that the desire for revenge is a desire to usurp
minute from now. Suppose also that I am tied to an undeter- God's prerogative. Which among these things we do will pre-
mined time bomb of the sort we considered in the previous sumably be partly a function of our constitution and our
section. Suppose the bomb is in fact going to go off in 30 education. That we shall do at least some one of them follows
seconds. Then, though there arc futures consistent with both from our being human, if not simply from our being rational
the present state of the world and the laws of nature in which beings. And to react in any of these ways is to demonstrate
I shall raise my hand one minute from now, I have no choice more surely than any high-minded speech ever could that we
about whether I shall raise my hand one minute from now, believe in moral responsibility.
since I shall be in bits one minute from now and have no I have listened to philosophers who deny the existence of
choice about this." moral responsibility. I cannot take them seriously. I know a
So, it would seem, there is no hope of finding out whether philosopher who has written a paper in which he denies the
we have free will by finding out whether determinism is true. reality of moral responsibility. And yet this same philosopher,
And there is no way other than this (and introspection) that when certain of his books were stolen, said, "That was a
I can think of that even seems relevant to the question whether shoddy thing to do!" But no one can consistently say that a
we have free will. Is our position therefore hopeless? I think certain act was a shoddy thing to do and say that its agent
not. Let us return for a moment to the consequences of reject- was not morally responsible when he performed it: those
ing the free-will thesis, a subject we discussed at length in who are not morally responsible for what they do may perhaps
Chapter V. We saw in that chapter that the denial of the free- deserve our pity; they certainly do not deserve our censure.
will thesis entails that there is no such thing as moral respon- What I have said in the last few paragraphs about our belief
sibility. And this was hardly a surprising conclusion: there is in moral responsibility is in some respects similar to what I
hardly anyone who has supposed that we could be held said in Chapter V about deliberation and our belief in free
morally accountable for what we do if we have no choice will. But there is a difference. The philosopher who denies
about what we do. But if the reality of moral responsibility free will continually contradicts himself because his non-verbal
entails the existence of free will, then, I would suggest, we behaviour continually manifests a belief in free will. But, I
have a perfectly good, in fact, an unsurpassably good, reason would suggest, the philosopher who denies moral responsibility
for believing in free will. For surely we cannot doubt the speaks words that contradict his theories, words like "That
reality of moral responsibility? was a shoddy thing to do". It is not only that his deeds belie
There are, perhaps, people who not only doubt but reject his words (though of course that is true too), but that his
the thesis that they are morally responsible agents. (I am words belie his words. I suggested in Chapter V that it would
thinking of psychopaths. I say 'perhaps' because it is not be impossible for us to cease behaving in ways that manifest
clear to me what goes on in the mind of, say, the man who a belief in free will. But I don't think it is impossible for us to
rapes and murders a little girl and afterwards feels no remorse.) cease talking in ways that manifest a belief in moral responsi-
But few people if any will react to an act of gratuitous injury bility. It would be merely very, very difficult. I ask you to try
deliberately done them by a human being in the way that to imagine what it would be like never to make judgements
they would react if that same injury were caused by a bolt of like 'What a perfectly despicable way for him to behave' or
lightning or a bough broken by the wind. When some person 'You'd think a person with her advantages would know better
injures us—at least if we believe he knew what he was doing than that' or 'I can never think of what I did without feeling
208 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 209

sick'. If you try to imagine this, perhaps you will experience responsible for what they do. We all believe that responsibility
what I do when I try the experiment. I find that one difficulty exists. And, I think, if we examine our convictions honestly
I anticipate in giving up making such judgements is very much and seriously and carefully, we shall discover that we cannot
like a difficulty I should anticipate in giving up making judge- believe that this assent is merely something forced upon us
ments like 'That car is dangerous because of its bad brakes' by our nature and the nature of human social life, as our be-
and The mushrooms growing under that tree are poisonous'. havioural manifestations of assent to the proposition that we
This difficulty arises from two facts: such judgements are have free will are forced upon us by the sheer impossibility of
often right and they are extremely important for getting along a life without deliberation. I think that we shall discover that
in the world. Think of some piece of behaviour you have we cannot but view our belief in moral responsibility as a
witnessed that you really would call 'perfectly despicable'. justified belief, a belief that is simply not open to reasonable
Isn't that the right thing to call it? Doesn't it describe it? Isn't doubt. I myself would go further: in my view, the proposition
it just as "objective" a description (whatever that means) as that often we are morally responsible for what we have done
`dangerous' or as 'poisonous'? is something we all know to be true.
Many philosophers, I suspect, will say that they use suet That we are convinced that we know something does not,
words as these to describe people's behaviour, but that in of course, prove that we do know it or even that it is true.
doing so they are not ascribing to those people moral responsi- But it is true that we are morally responsible, isn't it? And we
bility for that behaviour. This seems to me to be wrong. do know it to be true, don't we?
Suppose that there is a certain man who did a thing that led If we do know that moral responsibility exists, then we
us to say, "That was a perfectly despicable thing for him to should have no doubt about whether we have good reason to
do". Suppose that we later discover that he did that thing believe we have free will. It is this and only this, I think, that
shortly after he had been given, without his knowledge or provides us with a reason for believing in free will. It may well
consent, a drug that is known to alter human behaviour in be that, ironically enough, we believe that we are free be-
radical and unpredictable ways. Suppose this discovery led us cause we have no choice about what we believe about this
to decide that he had "not been responsible for what he was (owing to the necessity, for one's deliberations, of a belief in
doing" at the time he performed the act. It seems to me that one's own free will). But this fact cannot be anyone's reason
we could not then go on saying, "That was a perfectly despic- for believing in free will: at most it could be someone's excuse
able thing for him to do", not even if we qualified this asser- if he were charged with believing in free will without having
tion by adding, "though he wasn't responsible for his acts any reason that supported his belief. If someone were asked
when he was doing it". That additional clause, in fact, does to defend his belief in free will, he could not reply by saying
not seem to me to be a coherent qualification of the original that neither he nor anyone else had any choice about what he
assertion (unless, perhaps, "That was a perfectly despicable believed about free will. But it is as adequate a defence of the
thing for him to do' is taken to mean, 'Normally what he did free-will thesis as has ever been given for any philosophical
would be a perfectly despicable thing for someone to do'; but position to say, "Without free will, we should never be morally
that is not the case we are considering). The reason is simple. responsible for anything; and we are sometimes morally re-
To call an act 'despicable' is to censure its agent for perform- sponsible".
ing it, while to say of an agent that he was not responsible for I will now consider an objection to this argument (or, more
what he was doing when he performed an act is to excuse exactly, an objection to the employment of this argument by
him for performing it; and one cannot simultaneously excuse an incompatibilist) which, in some moods, I find very powerful
and censure. and which, I suspect, many philosophers find absolutely con-
We all, therefore, believe that people are sometimes morally vincing:
210 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 211

If what you say is correct, then, because we know that have good reason to believe that there exists no Cartesian
moral responsibility exists, we know that we have free Universal Deceiver—no being who deceives us all about
will. But, according to you, the free-will thesis entails almost everything. (For if there were such a being, then
indeterminism. And, presumably, you think that your almost every proposition we think we know to be true
arguments present the attentive philosopher with a good would be false.) Now the obvious validity of the paren-
reason for believing that the free-will thesis entails inde- thetical deduction clearly provides the attentive philo-
terminism. Therefore, if all your arguments are correct, sopher with a good reason to accept its corresponding
then our (alleged) knowledge of the existence of moral conditional:
responsibility, coupled with certain arguments a priori,
can constitute a good reason for believing that determin- If most of the propositions we think we know to be
true are true, then there exists no Universal Deceiver.
ism is false. But these things are not the sorts of things
that can be a good reason for believing in indeterminism. But if we have good reason to accept both this condi-
Indeterminism is, to put it bluntly, a thesis about the tional and its antecedent, then we have good reason to
motion of particles of matter in the void; the more accept its consequent; that is, we have good reason to
special thesis that we are undetermined (in the sense of accept the proposition that there exists no Universal
Section 6.2) is a thesis about the structural details and Deceiver. But these considerations are not the sorts of
the minute workings of our nervous systems. Only scien- considerations that can provide us with good reasons for
tific investigations are relevant to the truth or falsity of believing that there is no Universal Deceiver. The thesis
such theses. (This assertion is true even if you are right that there is no Universal Deceiver is, to put it bluntly,
in saying that it is unlikely that science will be able to a thesis about the features of a part of the world that is
show that we are undetermined.) Therefore, your argu- inaccessible to any possible human investigation. Your
ments represent just one more attempt by a philosopher contention that we have good reason to believe that
to settle by intellectual intuition and pure reason a ques- most of our knowledge-claims arc correct is just another
tion that should be left to empirical science. And if it case of a philosopher claiming to have good reason to
should prove that this question can't be settled by em- believe something no one could possibly have good
pirical science, owing perhaps to "the obscurity of the reason to believe.
matter and the shortness of human life", then we should
simply elect to have no opinion about the right answer The anti-sceptic who replies to this argument has, I think,
to it. two types of response available to him.
First, he can simply reply that the deduction the sceptic
I have said that in some moods I find this argument very
powerful. Nevertheless, my considered opinion is that it ought has presented—coupled with his own thesis about the general
correctness of our claims to knowledge—does show that he has
to be rejected. Its weak point can best be brought out by
comparing it with a certain argument that often figures in dis- good reason for believing that there is no Universal Deceiver.
cussions of scepticism. Here is a version of this argument. The This might be a surprising result, he will concede, but the
argument seems quite inescapable. As to the sceptic's conten-
sceptic speaks:
tion that the thesis of the non-existence of a Universal Deceiver
You say that most of the propositions we unreflectively is not the sort of thesis that human beings could have good
assume we know to be true are true. You, say, moreover, reason to accept (the anti-sceptic continues), the sceptic's
that we know this, or, at least, have good reason to be- deduction shows this initially plausible contention to be false.
lieve it. But we can deduce from this thing you say we Secondly, the anti-sceptic can point out that the sceptic's
212 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 213
argument depends essentially on the rule of inference reasons. And any philosopher who is willing to say that, since
he has good reason to believe that our knowledge-claims are
RP generally correct, he therefore has good reason to believe that
R(p q)
hence, Rq there is no Universal Deceiver, should feel no qualms about
saying that since he has good reason to believe that moral
(where 'R' stands for 'one has good reason to believe that') responsibility exists, he therefore has good reason to believe
and deny that this rule is valid. that determinism is false. Moreover, any philosopher who
I myself favour the first response. But this is not a book accepts the "determinism" argument will be hard-pressed
about epistemology, and I shall say no more about the ques- to find a way consistently to reject the sceptic's argument.
tion of the proper response to the sceptic's argument. My Assume that this philosopher accepts the rule of inference
purpose in setting forth the sceptic's argument is to compare `Ftp, R(p q) Rq', which is common to the two arguments;
it with the argument about determinism that is our present how then is he to reply to the critic who takes him to task as
topic. These two arguments are very much alike. They are so follows:
much alike, in fact, that the philosopher who rejects the con-
clusion of the argument about determinism has two types of You say that our having good reasons for believing in
response available to him that are exactly parallel to the two moral responsibility is not the sort of thing that could
types of response available to the anti-sceptic. provide us with good reasons for believing that determin-
First, he can simply reply that the incompatibility of moral ism is false. But you also say that our having good reasons
responsibility and determinism—coupled with his thesis that for believing that our knowledge-claims are generally
we know, and hence have good reason to believe, that moral correct is the sort of thing that can provide us with good
responsibility exists—does show that he has good reason to reasons for believing that there is no Universal Deceiver.
believe that determinism is false. As to the contention that But isn't it true that you are responding to philosophical
determinism is not the sort of thesis that we could have good problems that are essentially the same in arbitrarily dif-
reasons for thinking false unless those reasons• were provided ferent ways? If reflection on human knowledge can pro-
by empirical science, he can reply that the arguments he vide us with good reason for accepting a thesis about
accepts for the incompatibility of determinism and moral re- what goes on beyond the limits of any possible observa-
sponsibility show this initially plausible contention to be false. tion, then why can't reflection on human moral respon-
Secondly, lie can point out that the "determinism" argu- sibility provide us with good reason for accepting a thesis
ment depends essentially on the inference rule displayed about the motion of particles of matter in the void?
above, and deny that this rule is valid. I can see no very convincing response to this. I conclude that
I favour the first response, but I shall not directly defend the incompatibilist who believes that the existence of moral
the thesis that this response is the better of the two. I shall responsibility is a good reason for accepting the free-will thesis
instead argue that any philosopher who rejects the sceptical ought not to be troubled by the charge that his views commit
argument we have examined should also reject the argument him to the thesis that the existence of moral responsibility is
we are considering, and for the same reason, whatever that a good reason for believing in indeterminism. Or, at any rate,
reason may be. Any philosopher who rejects the sceptic's he ought to be no more troubled by this charge than he is
argument on the ground that 'one has good reason to believe troubled by "Universal Deceiver" arguments for scepticism.
that' does not behave like a necessity operator, should, of
course, reject the "determinism" argument, since that argu- 6.4 We have seen that there is every reason to think deter-
ment depends on the same assumption about the logic of good minism false, and no good reason to think true the weaker
214 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 215

thesis that our behaviour is determined in the way the be- I am convinced that in a large number of cases the answer is
haviour of computers is determined. We have seen that there that the people who regard my central thesis as simply in-
is at least one excellent reason for believing that we have free credible are victims of scientism. Scientism, as I use the word,
will: that we are at least sometimes morally responsible. We is a sort of exaggerated respect for science—that is, for the
have seen that free will is incompatible with determinism. We physical and biological sciences—and a corresponding dis-
are therefore in a position to put forward, tentatively perhaps, paragement of all other areas of human intellectual endeavour.
an answer to the question, "Do human beings have free will It hardly need be pointed out that scientism is the primary
or are they determined?" Human beings have free will and ideology of our age. It hardly need be pointed out that the
they are not determined. illusions scientism engenders are so pervasive and so insidious
This answer may be difficult for some people to accept. (I that it is practically impossible to get anyone who is subject to
am not, of course, saying that it is so obviously the right them to consider the possibility that they might be illusions."
answer that it should be accepted without argument. But I (I hope the following disclaimer is unnecessary: if I deprecate
have given arguments. They are this book.) I am thinking of scientism, I do not thereby depreciate science. To deny that
people who find this answer so initially implausible. so in- Caesar is due divine honours is not to belittle his generalship.)
trinsically incredible, that they will respond to the arguments Now it is not my purpose to enter into an extended discus-
I have given either by ignoring them or by setting out to find sion of or to offer an analysis of scientism. I wish only to call-
something wrong with them. (And, of course, a. philosopher attention to two effects of scientism on contemporary philo-
who sets out to find something wrong with an argument will sophical thought.
generally find some feature of it that he will consider a mistake. The first of these effects might be described as a revolt against
If he finds nothing else, he can always point to the fact that holism. The following sort of reasoning is common enough:
the premisses of the argument entail its conclusion and charge
the argument with begging the question.) A human being is, in the last analysis, a system of atoms,
This book is not really addressed to such philosophers as an enormously complex material thing the properties of
these. It is addressed to those who have kept an open mind which supervene entirely upon the properties of the
about the question of free will and determinism and who individual atoms that are its parts and upon the arrange-
would like to see what the arguments are. I am not one of ment of these individual atoms. But the kind of indeter-
those people who regards "keeping an open mind" as, per se, minism you ascribe to human beings could not supervene
a virtue. To have kept an open mind about Nazi racial theories upon the properties of a system of atoms. To say that if
or about astrology would not be a mark of virtue, either intel- all the parts of a system are determined then the whole
lectual or moral. But, it seems to me, the course of the free- system is determined is not to commit the fallacy of
will debate since the seventeenth century has been sufficiently composition, for determination is clearly the sort of
confusing, and twentieth-century physics and psychology, and property that "carries over" from parts to whole. Admit-
investigations of the human nervous system, have been suf- tedly the case is complicated by the fact that the in-
ficiently complex, that having kept an open mind about free dividual atoms are undetermined in some respects owing
will and determinism has indeed been consistent with the to quantum-mechanical considerations. But quantum
dictates of virtue. Why, then, do some philosophers feel com- mechanics has nothing whatever to do with free will,
pelled to react to the thesis of the last paragraph but one by and hence you must be postulating some other source of
calling its proponents "superstitious" or "believers in magic" indetermination, one having no basis in the behaviour of
or "armchair physiologists"? atoms and thus one that can have no place in a scientific
Perhaps there is no general answer to this question. But world-view.
216 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 217
I really don't know what to say about this sort of argument, Moral responsibility exists even though no one ever has
except that there must be something wrong with it. It proceeds any choice about anything;
from certain premisses (such as anti-holism and the thesis that
quantum-mechanical indeterminacy has nothing to do with There exist true propositions p and q such that someone
free will) and these premisses must be examined for plausi- has no choice about whether p is true and has no choice
bility. I myself am inclined to think that the two premisses about whether if p is true, then q is too, and yet has a
choice about whether q is true.
I have mentioned are quite plausible. I don't think that it is
very likely that we have any properties that don't in some If any of these propositions were true, that, in my view, would
sense supervene upon the properties of atoms of which we be a very great mystery indeed, a mystery much greater than
are composed; I am certainly not much tempted to think that that which would attend the truth of either
quantum mechanics has anything to do with free choice." We have properties that do not supervene upon the
But even if these were the only two premisses of the argument, properties of the atoms that we consist of
and even if the argument were absolutely clear in its termino-
logy, and indisputably logically valid, I should not find the Or
argument convincing. Its premisses may be plausible, but Free will involves quantum-mechanical indetermination.
I regard them as far less plausible than the proposition that
we are sometimes morally responsible for our acts and the Those under the influence of scientism will doubtless not see
proposition that rule (a) is valid. And if the argument of this it that way. In their view, no mystery could be greater than
book is correct, then it follows from these two premisses that the truth of a proposition that is in conflict with "the scientific
there is something wrong with the argument we are considering. world-view". And at least the former of these two proposi-
I should like to know just what is wrong with it, but I don't. tions is in conflict with that world-view, since it denies the
I have never pretended to understand "how free will works". anti-holism that is an essential part of the "scientific world-
If I knew I would tell you, but I don't know. The questions view"."
raised by the above argument are deep ones that I have no Here, I think, those philosophers whom I describe as
way of answering, just as I have no way of answering the victims of scientism and I have reached bedrock. We have
questions that I confessed myself puzzled by in Section 4.4. nothing more to say to each other; or, at any rate, though we
I have no liking for unresolved mysteries in philosophy. But may call each other names we have no more arguments.
it is no good trying to pretend that mysteries do not exist if The second effect that scientism has on some philosophers
they quite plainly do exist. Moreover, I prefer small mysteries is to produce in them a profound fear of being caught out by
to large mysteries. One way to look at this book is as an the progress of science." I believe that many philosophers
attempt to present a position on free will that commits its would find the following words an accurate expression of
adherents to smaller mysteries than does any available com- their thought. "Suppose I were to accept the argument that
peting position. Anyone who rejects the central theses and since moral responsibility exists, determinism is false, and
arguments of this book—I include under this heading the then suppose the physicists were to decide that determinism
theses of the present chapter, including the thesis that we are is true after all. Or suppose the neurophysiologists or the
not determined—must, if he has any well-worked-out theory cognitive psychologists or someone should provide us with a
of free will, determinism, and moral responsibility, accept at really good reason for believing that we're deterministic
least one of these three propositions: systems in at least as strong a sense as the sense in which
computers are deterministic systems. I should look pretty
Moral responsibility does not exist; silly then, shouldn't I? I should be laughed at for the same
218 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 219

reason we laugh at Hegel, who deduced the necessary exis- Despite what I have said, there is one question I shall very
tence of exactly six planets a priori. My argument and I should likely be asked by philosophers who think I have overstepped
be mentioned in the same breath as Kant and Euclidean the bounds of philosophy that I think is a good question and
geometry or Paley and the Argument from Design. Well, that's which I am bound to answer. That question might be put like
not for me." this:
Perhaps I am being unkind; perhaps the words I have attri- Very well. You admit that your theory is in principle
buted, or nearly attributed, to "many philosophers" could be empirically refutable. If physics were to become once
replaced with nobler-sounding words. But I do think that the more deterministic, or if the empirical sciences of man
ideas expressed in this imaginary quotation have more influ- were to provide us with really good reasons for believing
ence than has been generally recognized, and I do not consider that a human being is a deterministic system, then (you
this influence a healthy one. While it would be silly for philo- concede) your rejection of determinism would be ren-
sophers—or for anyone else, including scientists—to set out to dered untenable by science. Well, suppose this did happen
predict the future course of the sciences, I see no reason for despite your prediction that it won't. What would you
philosophers to shrink from holding positions that entail that say then?
science will never be able to do this or that. Such theories
would be empirically refutable, of course, but I see no reason I am not quite sure what I would say, but I believe I would
to regard that as a bad thing. If a philosophical theory is say that (0) was, after all, invalid. (I shall presently touch on
empirically refutable and gets empirically refuted, then there the reasons for my hesitation on this point.) This response is
will be one fewer philosophical theory about, and that would not purely ad hoc; I have not simply picked some one of the
be no bad thing. possible reactions to a scientific validation of determinism
Now some philosophers may say that a theory that calls and then arbitrarly embraced that reaction. Consider the
itself philosophical ought not to be empirically refutable, since following list of propositions:
any theory that is empirically refutable is "automatically"
not philosophy but science. According to this view, Kant and (1) We are sometimes morally responsible for the conse-
quences of our acts;
Paley—I pass over the fib about Hegel' —were, when they were
discussing geometry and design in nature, doing science (very (2) The validity of ((3) entails that our having free will entails
bad science, incidentally) rather than philosophy. According indeterminism;
to this view, I too am guilty of presenting science in the guise (3) If (1) is true, then we have free will;
of philosophy. This meta-philosophical thesis seems to me to
have nothing whatever to recommend it. The theories and (3a) We have free will;
arguments of this book are paradigmatically philosophical, (4) (0) is valid;
whatever else they may be. If someone wants to say that
they're also "scientific", that's a matter of indifference to me. (4a) Our having free will entails indeterminism;
Of course I should not want anyone to say that my theories (5) Indeterminism is true.
and arguments were bad science. Or not unless it is possible
for bad science to be good philosophy, something that seems In this book I have argued for, or, at any rate, asserted each
to me to be highly improbable. I think that what I have pro- of these propositions. Nevertheless, I do not regard them as
duced is good philosophy, and what I should like to see from equally well established. Though all of them are, in my view,
my adversaries is arguments for the conclusion that it is bad very likely true—I mean I regard their conjunction as very
philosophy. likely true, and a fortiori I regard them individually as very
220 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 221

likely true—some of them seem to me to be what I shall call would seem to he no reason to suppose that he should regard
preferable to others. (Let us say that a person regards p as none of them as preferable to the others. There is no reason
preferable to q if (i) he believes that it is much more likely why a philosopher should not say something of this form:
than not that both p and q arc true, and (ii) he believes that "Propositions A and B at present constitute the foundations
if either p or q is, after all, false, then p is true and q is false,) of my philosophy; I have no argument for either, but I believe
I have indicated my preferences among these propositions by I can simply see that they're true. Nevertheless, if either of
the way I have numbered them: I regard (1) as preferable to them is false, it's B."
(2), (2) as preferable to (3), and so on. (But I regard neither Now given that the above list accurately represents the
of (3) and (3a) as preferable to the other and neither of (4) ordering of my preferences, it is easy for me to say what I
and (4a) as preferable to the other.) Why do I have this set of should do if science were to provide me with an indisputable
preferences? Well, there are certain purely formal constraints reason for believing in determinism: I should have to reject
at work. For example, since (1) and (3) jointly entail (3a), (5). But (5) is entailed by (3a) and (4a), so I should have to
and since I know this, it would be rather odd of me to regard reject either (3a) or (4a). Since I prefer (3a) to (4a), I should
.

both (1) and (3) A9grthrable to (3a). But, mainly, this just is therefore have to reject (4a). But (4a) is entailed by (2) and
the preference-ordering I find myself with. I do not know (4), so I should have to reject either (2) or (4). Since I prefer
how to defend it, though, of course, I should be willing to (2) to (4), I should therefore have to reject (3). And that
consider carefully any argument for the conclusion that I would seem to be the end of the matter. (In this reasoning,
ought to order my preferences differently. I do not, of course, I of course presuppose that an "indisputable reason for believ-
think that my ordering of preferences is entirely arbitrary. ing in determinism" would not require me to accept without
I suppose I think that it is based on various conceptual insights proof any proposition that I should prefer (p) to.)
and that if you do not share my preferences you lack my I have defended (P) entirely on a priori grounds. But it
insights. would not surprise me too much to find that this proposition,
I am not in the least embarrassed by admitting this. Every which at present seems to me to be a truth of reason, had
philosopher's positions, however much he may argue, must been refuted by the progress of science. Such refutations
ultimately be based on certain propositions that simply seem have happened many times. And it does not follow from the
to him to be true. There is no way round this; it is a simple fact that they have happened that there is anything wrong
consequence of the fact that every argument has a finite with accepting on a priori grounds a principle that later turns
number of steps. Moreover, every philosopher must think out to be empirically refutable. One must simply realize that
that those of his colleagues whose "ultimate premisses" are a priori convictions are as corrigible as any others.
incompatible with his own have either not considered those I have earlier expressed some hesitancy about saying that
premisses with sufficient care or else lack some sort of insight I should reject (f3) if science were to force me to accept deter-
into how things are. What else is he to think? If he thinks minism. I hesitate because this counter-factual supposition is
that the ultimate premisses of those who disagree with him very abstractly described. I wish to leave open at least the
are "just as good" as his own, then he should admit that he bare possibility that there is a way in which science might
has no business holding his own premisses and should, at the force me to accept determinism that would also force me to
very least, retreat to a position in which he accepts nothing alter the ordering of preferences that I have expressed in the
stronger than the disjunction of his former premisses with all above list. I do not regard this as very likely; but who can be
those propositions that arc "just as good". certain about what the future holds?
One more meta-philosophical point is in order. Given that
a philosopher must have a set of "ultimate premisses", there
222 THE TRADITIONAL PROBLEM THE TRADITIONAL PROBLEM 223
6.5 In this, the final section of this book, I shall outline its able to "handle" the well-known counter-examples to simpler
central argument. conditional analyses. This is the case simply because such
The free-will thesis is the thesis that, very often, when we analyses are too complex to be grasped as wholes and thus
are faced with having to choose between incompatible courses cannot even seem to be intuitively evident.
of action, each of them is such that it is within our power to We should, therefore, be incompatibilists unless there is
choose it. Or, more idiomatically, the free-will thesis is the something objectionable about incompatibilism, something,
thesis that we very often have a choice about what we are moreover, that is more objectionable than the rejection of the
going to do. apparently self-evidently valid inference rule (33). And indeed
Deterininism is the thesis that, given the past and the laws there is something objectionable about incompatibilism. The
of nature, there is only one possible future. incompatibilist must either reject the free-will thesis or else
The free-will thesis and determinism are incompatible. That accept the thesis that an agent often has a choice about
is, incompatibilism is true. This can be seen on the basis of whether a certain one of his inner states is followed by a
several detailed arguments, all of which are elaborations of certain overt act (the state being the cause of the act), even
the following simple argument (the Consequence Argument): though neither the inner state nor anything else determines
that act to occur." I do find each of these alternatives objec-
If determinism is true, then our acts are the consequences tionable, but the second seems to me to be far less objection-
of the laws of nature and events in the remote past. But able than the rejection of (13). I therefore counsel accepting
it is not up to us what went on before we were born, and incompatibilism.
neither is it up to us what the laws of nature are. There-
fore, the consequences of these things (including our If incompatibilism is true, then either determinism or the
free-will thesis is false. To deny the free-will thesis is to deny
present acts) arc not up to us. the existence of moral responsibility, which would he absurd.
One of these three detailed arguments—The Third Argument Moreover there seems to be no good reason to accept deter-
of Chapter III—is a particularly useful tool in a discussion of minism (which, it should be recalled, is not the same thesis as
compatibilism because it isolates, as a separate premiss, the the Principle of Universal Causation). Therefore, we should
most doubtful thesis the incompatibilist must accept. That reject determinism.
thesis is the validity of the following rule of inference: This conclusion is, at least in principle, open to scientific
refutation, since it is conceivable that science will one day
(13 ) p and no one has, or ever had, any choice about whether p ; present us with compelling reasons for believing indeterminism.
Then, and only then, I think, should we become compati-
If p then q, and no one has, or ever had, any choice bilists, for, in the case imagined, science has ex hypothesi
about whether if p then q; shown that something I have argued for is false, and the
hence, q and no one has, or ever had, any choice about weakest of my arguments are those that support incompati-
whether q. bilism.

The proposition that this rule is valid, when carefully con-

sidered, disposes the mind to assent; it seems to be a truth of
reason. But, for all that, it may be false. Still, it seems more
likely to be true than certain of the incompatibilist's premisses.
In particular, it seems more likely to be true than any condi-
tional analysis of ability that is sufficiently complex to be
 Notes

PREFACE
1 In Ethics and the History of Philosophy (New York: 1952).
2 Mind (1934).
3 In Sidney Hook, ed., Determinism and Freedom in the Age of
Modern Science (New York: 1950.
4 In Keith Lehrer, ed., Freedom and Determinism (New York:
1966).
5 Englewood Cliffs, NJ: 1966.
6 The Journal of Philosophy (1969).
7 Published as a pamphlet by Cambridge University Press, 1971,

CHAPTER I
1 Four very important problems about free will that we shall not
consider are (i) the problem of free will and divine foreknowledge, (ii) the
problem of free will and psychoanalysis, (iii) the problem of discovering
whether &tide's incompleteness results imply anything about human
freedom, and (iv) the problem of discovering whether the conceptual
impossibility of predicting one's own decisions implies anything about
human freedom.
2 In my usage, law of nature and the physical modalities are inter-
definable. I assume that, as etymology would suggest, the laws of nature
and the laws of physics are the same, or, at least, that the former super-
vene upon the latter (that is, the laws of nature could not possibly have
been different without the laws of physics also having been different).
3 Psychological determinism, the thesis that one's acts are determined
by one's strongest motive, will not be discussed in this book. I can get
no grip on the notion of a "strongest motive" and consequently have
nothing to say about it.
4 The doctrine of immanent causation will be discussed in secs. 4.4
and 4.5.
5 To say that Tom has no cause is not to say that such things as
Tom's coming to be and the fact of Tom's existence have no cause; nor
is it to deny that Tom has an aitia.
6 Causality and Determination (Cambridge, 1971).
7 Jan -Lukasiewiez "On Determinism", in Storrs McCall, ed., Polish
Logic (Oxford: 1967). Perhaps an example would make Lukasiewicz's
-

point clearer. Suppose t is noon and t, is 1,00 p.m. Suppose the cause
 227
226 NOTES TO PAGES 5-23 NOTES TO PAGES 24 - 32

of A occurred at 12.30, the cause of the cause of A at 12.15, the cause arguments in a narrow sense (I resist the temptation to pun): those argu-
of the cause of the cause of A at 12.07.30, and so on, ad infinitum. ments that depend on the notions of time and truth. The question of
8 See sec. 4.4. the compatibility of free will with determinism and indeterminism will
9 I doubt whether truth is necessary for lawhood, See sec. 1.5. be considered in Chapters III and IV.
10 Perhaps there is more to a proposition's "supporting its counter- 2 The Collected Plays of W. Somerset Maugham, (London: 1931),
factuals" than this. If so, what more is there? 298-9. The quotation is from "Sheppy", Act III.
11 Or so most people now believe. Assume they're right for the sake 3 There is a problem about defining no matter what I do (he does,
etc.). Suppose you were to say "She would have died no matter what
of the example.
he had done", and a carping critic replied 'That's not true! She wouldn't
12 The remainder of this section derives mainly from Richard Taylor's
have died if he'd prevented her death, made an effective medicine out
classic paper "I Can", The Philosophical Review (1960).
of the materials at hand, transported a doctor to her bedside by magic,
13 In "Freedom to Act", in Ted Honderich, ed., Essays on Freedom
or if he'd done any of a great variety of things." The obvious way of
of Action (London: 1973), Donald Davidson defends the view that the
dealing with this critic is to stipulate that 'no matter what one does'
power to act is a "causal power". His thesis is consistent with mine. In
means 'no matter which of the things one can do one does'. But if we
Davidson's sense of 'causal power', the phrase 'if he wanted to speak
accept this stipulation, then it will follow from fatalism that the man
French, he would' predicates a causal power of the person to whom it is
who delayed seeing a doctor about his cough till it was too late would
applied. In my sense of 'causal capacity' it does not. For suppose a cer-
have died of the disease the cough signalled no matter what he had done
tain man would speak French if he wanted. It doesn't follow that there
(since, according to fatalism, what he could have done and what he did
is any set of external circumstances such that if he were plunked down
coincide). And, in general, if we accepted this stipulation, it would be a
in those circumstances he would speak French (or do any other particular
consequence of fatalism that all events are strongly inevitable for every-
thing), for perhaps he doesn't want to speak French (or to do any other
particular thing) and wouldn't want to under any circumstances. And, one. That this is a consequence of fatalism is a thesis that one of the
a fortiori, it doesn't follow that there is some set of circumstances in most prominent contemporary authorities on fatalism has been at pains
which he would speak French (or whatever) willy-nilly. I do not believe to deny. (See Steven Cahn, Fate, Logic, and Time, New Haven and
that the power to act is a "causal power" in Davidson's sense, but London: 1967, ch. 2, passim.) It would hardly do, therefore, to make
whether it is or not is a question that has nothing to do with the theses this thesis true by definition. Perhaps the solution to this problem is to
I have been advancing. See sec. 4.3 for a discussion of the power to act stipulate that no matter what one does means no matter what choices
and conditionals whose antecedents involve states internal to the agent. or decisions one makes.
14 There seems to be no customary name for the conjunction of 4 To believe that God plays the role of a cosmic puppet-master would
cornpatibilism with the thesis that we have free will. be to have a superstitious belief about God.
15 R. E. Hobart, "Free Will as Involving Determination and Incon- 5 This is the infamous "idle argument" of antiquity, considered,
ceivable Without It" (1934). P. H. Nowell-Smith, "Free Will and Moral according to Cicero, by Chrysippus.
Responsibility" (1948), J. J. C. Smart, "Free-Will, Praise and Blame" 6 For a detailed treatment of these issues, see Chapter V.
(1961). 7 The one exception I know of is "Time, Truth, and Modalities" by
16 See Richard Foley, "Compatibilism and Control Over the Past", "Diodorus Cronus" (Steven Cahn and Richard Taylor), Analysis (1965).
Analysis (1979) and "Reply to Van Inwagen", Analysis (1980). 8 This is true in the same sense as that in which it is true that I use
17 For more about "begging the question" and "where the burden 'cardinal number' as a general term for the things people count with.
of proof lies", see sec. 3,10. But, of course, most numbers are too large for us to count with, and in
my view—most propositions are too complex for us to entertain. It would
be more accurate, therefore, to say that I use proposition as a term for
CHAPTER II a certain class of objects, some of the simpler members of which are the
things people assent to, etc. If anyone finds the "further" or "unenter-
1 C. D. Broad was a fatalist by the terms of this definition. The taMable" propositions mysterious, I ask him to lay aside his objections
argument of his inaugural lecture, "Determinism, Indeterminism and for the moment. Unentertainable propositions will play no role in the
Libertarianism" (in Ethics and the History of Philosophy, New York: argument of the present chapter, though they will figure in our attempts
1952), might be summarized as follows: free will is incompatible with to define determinism in Chapter III.
both determinism and indeterminism and is therefore impossible (though 9 For the sake of convenience, I shall frequently use 'believe' and
this bald summary does not do justice to Broad's beautifully finished 'say' transitively when talking about the relations people bear to propo-
lecture). In the present chapter we shall be concerned with fatalistic sitions, despite the fact that sentences of the form 'S says p' and 'S

228 NOTES TO PAGES 33-38
NOTES TO PAGES 40 -56 229
believes p' are usually too odd-sounding for me to feel at all comfortable
about them, except for the case in which `p' is replaced by certain 'wh'- or earlier. Then, it would seem, it was true in 10,000,000 BC that no
nominalizations. For example, we certainly can't say 'John said Newton's propositions had yet been asserted; or, at least, this would seem to be
First Law of Motion' and my ear doesn't much care for 'John believes the right thing to say if the temporal qualification of the possession of
Newton's First Law of Motion'—but 'John believes what Newton postu- truth makes sense. But, of course, if anyone had asserted this proposition
lated about motion' is perfectly all right. in 10,000,000 BC he would have said something false.
10 Thus, sentences containing indexical terms can, strictly speaking, 17 This argument-form is essentially the principle that will be called
be said to express propositions only in or relative to a situation or a 'Rule (p)' in the parts of this book that deal with free will and deter-
"context of utterance", just as denoting phases containing indexicals can minism. I have no wish to dispute the validity of this argument-form,
be said to denote objects only in or relative to a situation or context of for the validity of (0) comes very close to being the single premiss upon
utterance. We may therefore say that a sentence containing indexical which the argument of this book is based. For a point about the relation
terms expresses a given proposition in a given context of utterance, of Rule (0) to fatalism see note 32, Chapter III,
provided that the result of concatenating 'the proposition that' and that 18 Metaphysics (Englewood Cliffs, NJ: 1963), ch, 5.
sentence denotes that proposition in that context of utterance. 19 Note that this point does not depend on our artificially extended
11 We. shall return to the topic of truth and falsity in sec. 2.7. sense of in the absence of. It therefore seems unlikely that Taylor's argu-
12 I use 'sound' and 'valid' in what have become their usual technical ment is defective owing to some incoherency in the notion of one's per-
senses: a valid argument is an argument whose conclusion follows from forming an act "in the absence of" conditions that, if they obtained,
its premisses; a sound argument is a valid argument with true premisses. would obtain at times different from the time at which one performs
13 The remainder of this section owes a great deal to A. J. Ayer's the act.
essay "Fatalism", in The Concept of a Person (London: 1963). 20 Taylor has told me that he intended Principle (A) to apply only
14 I take it that when someone speaks these words, he uses the demon- in the case of conditions causally but not logically necessary for one's
strative pronoun to refer to the proposition expressed by the sentence acts. But what is the point of this restriction? Surely if it's plausible to
the person he is speaking with has uttered, and not to the sentence suppose that I can't do a thing in the absence of a condition causally
uttered. To deny this would be like saying that in the following fragment necessary for ray doing it, then it's even more plausible to suppose that
of conversation, I can't do a thing in the absence of a condition logically necessary for
my doing it. That is to say, if there were any good reason to reject the
"How many have you invited to the wedding?" principle
"Four hundred."
"That's too many.", No agent is able to perform an act in the absence of a condition
logically necessary for its accomplishment,
the demonstrative pronoun refers not to the number four hundred but
to the words 'four hundred'. that reason would be an even better reason for rejecting the principle
15 This paraphrase, I believe, captures Aristotle's view of truth-at-a- No agent is able to perform an act in the absence of a condition
time. At any rate it is suggested by the language he uses in De Interpre- causally but not logically necessary for its accomplishment.
tatione, IX, particularly at 18b. A typical and especially suggestive passage
is "... if a thing is white now, it was true before to say that it would be 21 Richard Taylor, "I Can", The Philosophical Review (1960), 81.
white, so that of anything that has taken place it was always true to say 22 This is as close as I can come to making sense of the syntactical
'it is' or 'it will be'." (W. D. Ross (ed.), The Works of Aristotle Trans- lusus naturae (or artis) `(p) (p V '-p)'. One might try 'Everything is such
lated into English, vol. I (Oxford: 1928), tr. E. M. Edghill.) About that either it or it is not the case that it'.
twenty lines later, in discussing the alleged necessity of the events referred 23 This argument-form certainly does not in any clear sense "pre-
to in a correct prediction, he says, "... a man may predict an event ... suppose" LEM.
and another predict the reverse; that which was truly predicted at the
moment in the past will of necessity take place in the fullness of time.
Further, it makes no difference whether people have or have not actually
made the contradictory statements ...". Steven Cahn employs essentially CHAPTER III
this Aristotelian conception of truth-at-a-time. Sec his Fate, Logic, and
Time (cited in note 3), 33 note 15. 1 "Freedom to Act" in Ted Honderich, ed., Essays on Freedom of
Action (London: 1973), 139.
16 This definition faces a great many purely technical difficulties. Sup-
2 I know of only one fallacy that incompatibilists have been accused
pose, for example, that no propositions were asserted in 10,000,000 BC
of that is not childish. In Will, Freedom and Power (Oxford: 1976),
 NOTES TO PAGES 69 - 80 231
230 NOTES TO PAGES 58-68
to 32, left to 5; (ii) it is not possible that he should do this and the past
175-6, Anthony Kenny charges that incompatibilists make use of the
be as it was and the safe remain locked. Thus, by the strict terms of our
following rule of inference, which he contends is invalid:
definition, he can render false the proposition that the safe is locked.
I cannot do so-and-so Therefore, it might be argued, what our definition really captures is
To do such-and-such is, in this case, to do so-and-so something that is more like 's could, if he were lucky enough, render p
hence, I cannot do such-and-such. false' than it is like 's can render p false'. But if this is so, it will not
(The incompatibilist, Kenny says, proceeds by substituting 'violate a materially affect our argument. Our argument will, loosely speaking,
law of nature' for 'do so-and-so' and expressions denoting unperformed proceed by deducing 'It is not the case that s can render p false', for
acts for 'do such-and-such'.) Whether or not this inference form is valid, arbitrary values of p, from determinism. If this thesis deduced from
and I do not think Kenny presents a clear counter-example to it, I can- determinism is really best read 's could not, no matter how lucky he
not see that any of the arguments of the present chapter depends on it, was, render p false', this fact would hardly undermine my claim to have
demonstrated the incompatibility of free will and determinism. More
despite the fact That these arguments do depend on our inability to
over, the present definition has the useful consequence that 's can render
render false any proposition that is a law of nature.
p false' is a purely extensional context, a feature that might very well
3 In sec. 3.6 I shall explain in detail what I mean by possible world
have to be sacrificed in order to produce a more intuitive result in cases
and related notions like that of truth "in" a world.
like the case of 'He can render the proposition that the safe is locked
4 This point has nothing to do with the so-called "many-worlds"
interpretation of quantum mechanics. What I call possible worlds are false'.
10 Narveson, op. cit.
abstract objects: ways the universe might be. The "worlds" of the many-
11 Andrl Gallois, "Van Inwagen on Free Will and Determinism",
worlds interpretation of quantum mechanics, however, are concrete
objects: universes. Philosophical Studies (1977).
12 Many philosophers think that a "double temporal reference" is
5Not that I believe that laws of nature are by definition true. See
required to make ascriptions of ability fully explicit. That is, they think
sec. 1.5. But I shall from now on assume that lawhood entails truth.
that the "real form" of ascriptions of ability is something like this: .c
This assumption could be dispensed with at the cost of minor compli-
could at t, have done A at 1 2 . I am not convinced by their arguments. I
cations.
find 'Tom could have raised his hand at noon' ambiguous but clear--that
6 This thesis would seem to entail that there are at least as many
is, I think it's clear what the two individually clear things this sentence
propositions as there are moments of time; and, presumably, there are
might be used to say are—and 'Tom could at eleven o'clock have raised
as many moments of time as there are real numbers. To postulate an
his hand at noon' sounds rather strange to me. I think the best way to
indenumerable infinity of propositions, most of which are not things
understand sentences of this latter sort is like this: 's could at 1 1 have
that could possibly be thought of, owing to their unimaginable com-
plexity, may seem to some people to be extravagant. But I do not see done A at t 2 ' df 's could have (done A at t 2 ) and at t, it was not yet
too late for s to do A at t i '. That is, I prefer to take sentences of the
how to state the thesis of determinism without some "extravagant"
form 's could have done A at t' as "basic" (whatever precisely that
assumption. Compare Richard Montague's criticism, in the opening para-
means) and to define sentences containing a "double temporal refer-
graphs of "Deterministic Theories", of the characterization of deter-
ence" in terms of these basic sentences, disambiguating brackets, and
minism that he attributes to Laplace and Ernest Nagel. (Formal Philos-
ophy (New Haven: 1974), 303 f.) the notion of "not yet too late". I do not believe that my argument
requires this rather special notion.
7 See Jan Narveson, "Compatibilism Defended",Phdosophical Studies
(1977). 13 Cf. Narveson, op. cit. The possibility of this line of argument was
first pointed out to m.e in conversation by Raymond Martin and Michael
8 The phrase 'possible in the broadly logical sense' is Alvin Plantinga's.
See The Nature of Necessity (Oxford: 1974), 1 f. Gardner.
14 The speed of light is just under 3 X 10 1 metres per second. The
9 There is one other way in which this definition may diverge from
difference is minuscule—about ninety times the muzzle velocity of a
our intuitive notion of one's having the truth-value of a proposition
high-speed rifle bullet.
within one's power. Consider the proposition that the safe is locked. Even
15 Oxford: 1974.
if we were normally in the habit of talking of rendering propositions
false, we should not normally say that an agent had it within his power 16 By 'possibility' I mean something like what philosophers have
traditionally called 'logical possibility'. But I dislike this term intensely,
to render this proposition false if he did not know, and had no way of
since, in my view, many important possibilities and impossibilities do
discovering, the combination. But it may well be that a certain agent,
not owe their modal status to "logic", or not in any clear sense. It is,
who has no way of finding out the combination, is in the following
for example, as impossible as anything could he for the moon to be made
situation: (i) he has it within his power to turn the dial left to 26, right
232 NOTES TO PAGES 80-82 NOTES TO PAGES 82-99 233

of cheese. This is just as much an impossibility as a round square; there has two main senses: `the region that includes all regions' (the cosmos)
is no sense in which the one is possible and the other impossible. But in and 'the state of affairs that includes all states of affairs'. And, surely, a
what sense is the impossibility of the moon's being made of cheese (possible) state of affairs and a possibility are very similar things.
"logical"? If anyone does think it's "possible" in any sense for the moon 22 I prefer 'at' to In', since the former preposition is less likely to
to be made of cheese, I recommend he read George Seddon's "Logical reinforce the widespread incoherent practice of thinking of possible
Possibility", Mind (1972). See also my "Ontological Arguments", Noiis worlds as things with insides (that is, cosrnoi). But my usage is not con-
(1977), and (for a discussion of the "green cheese" case) my review of sistent.
R. G. Swinburne's The Coherence of Theism, in The Philosophical Review 23 An equivalent definition: an object exists at a world if that world
(1979). includes the possibility that that object exist. More generally, a propo-
17 I use 'object' simply as the most general count-noun. Thus, in my sition is true at a world if that world includes the possibility that that
usage, everything is an object. I mean the word to have no implications proposition be true. These definitions are equivalent because a counter-
beyond those that are, in the very strictest sense, consequences of this factual conditional having a protasis of the• form If w were actual
definition. is equivalent to the corresponding strict conditional.
18 Assuming, that is, that our ordinary modal beliefs are correct, an 24 The first conjunct of the definiens is unimportant; it is there
assumption I shall feel free to help myself to when I am giving illustrative merely to restrict the application of `Dx' to possible worlds, a function
examples. But if Spinozism--a thesis strikingly at variance with our `( 3y) (Nxy)' would have performed equally well. The second conjunct
ordinary modal beliefs—were true, the possibility that Socrates teach does the work.
Plato would include the second of these possibilities and would preclude, 25 I shall be forced to talk about the individuation of events in
for example, the possibility that the Axis powers win the Second World Chapter V.
War. Thus, if Spinozism were true, we should owe even more to Socrates' 26 The bars are, of course, as much a piece of imagery as the system
pedagogy than we had suspected! of corridors. My use of them in this model is not meant to suggest that
19 Thus, those philosophers who go about saying loudly and defiantly, an agent is unable to bring about an event only in the case that some
"There's only one possible world, the actual one!" are either Spinozists tangible and immovable barrier stands between him and the means neces-
or fools. sary for bringing it about.
20 This definition obviously entails that 'actual' does not mean 27 Perhaps there is only one "empty" world, only one world that
`existent'. All non-actual worlds exist, which is hardly surprising, since contains no contingently existing objects. If so, this world is not relevant
everything exists. (The golden mountain cannot be said to be a counter- to our present concerns.
example to this thesis, since there isn't any such thing as the golden 28 I doubt whether anyone has access, by the terms of this definition,
mountain; no one can put forward the golden mountain as a counter- to any set having a merely possible world as its single member.
example to any thesis whatever, since there is no golden mountain to be 29 I shall presently suggest a semantics for `N', but only as a way of
put forward.) The thesis that non-actual worlds exist is simply a specifi- making the point that any very interesting semantics for this operator is
cation of the general truth that unrealized possibilities exist: (To deny going to be controversial.
this is to embrace Spinozism.) Some philosophers I have propounded 30 In the remainder of this paragraph, I am going to be less careful
this unremarkable proposition to have looked puzzled for a moment about use and mention than is my habit.
and then brightened and said, "Ah, I see—you mean they exist as possi- 31 Thomas McKay has proposed an interesting possible counter-
bilities." And yet if I had told these same philosophers that New York and example to (0). We should normally say that if someone throws a fair
London exist, they would not have looked momentarily puzzled and die in an honest game, then no one has any choice about the outcome
then have brightened and said, "Ah, I see—you mean they exist as cities." of the throw. And most of us think that most gamblers have a choice
Perhaps this odd reaction has its roots in a confusion between possi- about whether they practise their vice. But then the following instance
bilities and possibilia. If there were any such things as mere possibilia- of (p) could easily have true premisses and a false conclusion:
which there aren't—they would be denoted by phrases like 'the golden
N Alfred throws a six
mountain'. But a possibile is not the same as a possibility. Possibilities N (Alfred throws a six Alfred plays dice)
are denoted by phrases like 'the possibility that there be a golden moun-
hence, N Alfred plays dice.
tain'. This phrase, of course, denotes something if and only if it's possible
that there be a golden mountain. (The second premiss is true because the conditional embedded in it is a
21 My use of 'world' to denote comprehensive possibilities is not necessary truth.) Respondeo: Strictly speaking, Alfred does have a choice
utterly at variance with ordinary English usage. In his article, "World", about whether he throws a six, at least provided he has a choice about
in Studies in Words (Cambridge: 1967), C. S. Lewis tells us that `world' whether he plays dice. He can avoid throwing a six by avoiding playing

234 NOTES TO PAGES 101-109 NOTES TO PAGES 110-115 235

dice. What Alfred has no choice about is whether, given that he plays intentions plainer. The "table of instructions" consists entirely of con-
dice, he throws a six. That is—supposing that no one else has any choice ditionals, such as 'If you're in early middle age and a superior humiliates
about this either, and that Alfred does throw a six—N (Alfred plays you by repremanding you in front of your subordinates, attempt to get
dice Alfred throws a six). revenge by spreading scurrilous rumours about his personal life, taking
32 There is an argument about fatalism that parallels this argument extreme care to see to it that these rumours can't be traced back to
about genetic determinism. What is essentially the argument for fatalism you'. The presence of the tiny device is causally sufficient for the sub-
that we examined in sec. 2.5 can be stated in such a way as to make ject's satisfying each of these conditionals; but it is time and chance that
explicit use of (p). Let S be the (true) proposition that I shall shave determine which of them the subject satisfies non-vacuously and which
tomorrow, and let t be some moment in the remote past. Availing him- he satisfies only vacuously. Thus, (M) does not attribute omniscience to
self of Rule (p) , the fatalist argues: the Martians. They do not in general know what situations anyone is
N S was true at t going to find himself in. But their "programs" for us are sufficiently
comprehensive that every important contingency of human life is rep-
N (S was true at t I shall shave tomorrow)
hence, N I shall shave tomorrow. resented by the antecedent of one the conditionals that the program
comprises.
In sec. 2.5, I argued that the first premiss of this argument was either 4 Certain followers of Wittgenstein might say that if we made such
meaningless, or, given a certain stipulation about what it might mean, a discovery we should no longer know what to say. But no one, I think,
doubtful. (Obviously no such move is open to us in relation to the argu- would feel tempted to say that we should simply continue to call people
ments of the present chapter. The proposition that P c, really is about the `free'.
past and hence about what is outside our control. The proposition that 5 "An Empirical Disproof of Determinism?", in Keith Lehrer, ed.,
S was true at t is made to look by a linguistic trick as if it were about Freedom and. Determinism (New York: 1966).
the past. That is why the problem of fatalism is a problem to be dissolved 6 I have attempted in this brief discussion of Lehrer's reasoning to
by careful attention to the language of time and truth, while the prob- present an argument only for the conclusion that that reasoning is defec-
lem of free will and determinism is deep and intractable and madden- tive; I have not attempted a diagnosis. For a diagnosis, see my paper
ing.) This is a thesis about which the fatalist and I might have a serious "Lehrer on Determinism, Free Will, and Evidence",Philosophical Studies
philosophical discussion. But suppose that instead of saying this, I had (1972).
simply denied that the above instance of (p) was valid. Then, I think, 7 The most serious of these is that 'x can do y if he chooses' is no
the fatalist would rightly have accused me of declining to make a serious better a definition of 'x can do y' than 'x is beautiful if you look at it
contribution to the discussion, and so would any anti-fatalist who was correctly' is of 'x is beautiful', or than `x is poisonous if ingested' is of
really interested in getting to the bottom of the fatalist's puzzle. `x is poisonous'. J. L. Austin has shown that 'he can if he chooses' can-
33 In Chapter IV, the reader will remember, we shall examine two not be regarded as the straightforward conditional it appears at first
other arguments for compatibilism, the "Paradigm Case Argument" glance to be. ("Ifs and Cans", Philosophical Papers (Oxford: 1961).) I
and the "Mind Argument". I shall not compare the premisses of these believe that the correct analysis of 'he can if he chooses' shows it to be
arguments with (13). The Paradigm Case Argument and two versions of the a disguised conjunction: 'he can and if he doesn't this is only because
Mind Argument will be seen to suffer from defects more serious than he doesn't choose to and not because he can't'. Since this statement
an appeal to premisses less plausible than the validity of (13). The re- contains 'he can' as a conjunct, it is not going to be of much use to any-
maining version of the Mind Argument—there are three—will be seen to one who is interested in what 'he can' is compatible with.
have the validity of (0) among its premisses. 8 See his review of Austin's Philosophical Papers in Mind (1964),
34 I shall answer this charge in another way in sec. 5.8. and his "He Could Have Done Otherwise", The Journal of Philosophy
(1967). I cannot resist quoting in its entirety a footnote to the latter:

Presumably it was for reasons such as these that George Washing-

ton was said to be unabte to tell a lie. The point was, not that he
CHAPTER IV
lacked the wit or skill or opportunity to do it, but that he was so
1 "Divine Omnipotence and Human Freedom", in A. Flew and A. good that he couldn't bring himself to deceive. Bayle quotes a
McIntyre, eds., New Essays in Philosophical Theology (London: 1955). seventeenth-century Walloon theologian, one de Wolzogue, who
2 Ibid., 150. pointed out that, although God would have no difficulty in deceiv-
3 W. P. Alston has suggested to me that this statement of (M) may ing if he chose to deceive, nonetheless he cannot deceive since he
be misinterpreted, so I shall say a few words more to try to make my cannot choose to deceive. De Wolzogue wrote: "God can deceive
236 NOTES TO PAGES 115-121 NOTES TO PAGES 121 129 - 237

if he will . . . but it is impossible for him to have such a will to cited in note 16, Chapter I. I might mention in passing that there is a
deceive; it is also impossible for him to endeavor to employ his tendency among some compatibilists (including, I think, Foley) to treat
power for the execution of a deceit, whence I conclude that it is their favourite conditional analyses as part of the thesis they call 'corn-
impossible for him to deceive." See Pierre Bayle, A General Dic- patibilism' ; that is, there is a tendency to treat this word not as a name
tionary, Historical and Critical, article "Rimini (Gregorio de)," for the thesis that free will and determinism are compatible, but rather
note C. According to some Christians, an important point of dif- as a name for something of this form: 'Because the following conditional
ference between Mary and Jesus was that, while Mary could sin analysis of can is correct . . ., free will is compatible with determinism'.
but never did, Jesus "has not merely not actually sinned, but also This tendency should be resisted, if for no other reason, because it is
could not sin," the point being, again, that he could not undertake possible to accept the compatibility of free will and determinism and to
(choose, will, try, set out) to sin. Compare Ludwig Ott, Funda- reject conditionalism. For an actual case of this, see Lehrer, "An Empiri-
mentals of Catholic Dogma (Cork: Mercier Press, 1952), p. 169. cal Disproof of Determinism?", cited above.
Compare St. Thomas' treatment of the question, "Can God Do 16 For a conditional analysis of ability that does not have the con-
What Others Do?" in On the Power of God, Question II, Article 4. sequence that any of the premisses of the First Formal Argument is
false, see Carl Ginet, "The Conditional Analysis of Freedom", in P. van
9 "Cans Without Ifs", Analysis (1968-9).
Inwagen, ed., Time and Cause: Essays Presented to Richard Taylor
10 As R. M. Chisholm showed in his review of Austin (cited in note
(Dordrecht: 1980). From now on, I shall generally use 'conditional analy-
8), this definition could be improved by writing the definiens as follows:
sis' to mean 'conditional analysis that supports compatibilism'.
There is something such that if x had chosen to do it, then x 17 The reader should be able to verify for himself that given the fairly
would have done y, and x COULD HAVE chosen to do it. precise notion of 'render false' that was introduced in sec. 3.4, the
This definition would seem to avoid the difficulty illustrated by the Analysis is incompatible with the truth of premiss (6).
Napoleon-Waterloo case mentioned earlier. But let us leave the definition 18 We might expect from the preceding note that the correctness of
as the text has it, since we shall be interested in other sorts of difficulties the Analysis would entail the falsity of the premiss 'NU of the Third
than those Chisholm's emendation is designed to obviate, and the emen- Formal Argument—which is the premiss of that argument that seems
dation would cause our discussion of these other sorts of difficulties to intuitively to "correspond" to premiss (6) of the First Formal Argument
be more complex than it need be. —but this appears not to be the case. I do not think that this undermines
11 Some philosophers have denied this, on grounds that are not clear my contention that the three arguments of Chapter III are essentially the
to me. For a discussion of the question whether one's choosing to per- same and stand or fall together. The three arguments represent three
form a certain act is itself an act, see Myles Brand's introduction to his ways of "tightening up" the Consequence Argument and are in that
collection, The Nature of Human Action (Glenview, Ill.: 1970), 14-16. sense essentially the same; but the three sets of technical devices used
12 The desires that figure in the present example are "occurrent" and to tighten up the Consequence Argument differ markedly from one
not "standing". This morning I wanted to visit the library to verify a another and this had led to certain asymmetries among the premisses
reference and I also wanted to live to a ripe old age; but the latter desire, that superfically "correspond" to one another, with respect to their
unlike the former, was never present to my consciousness, never before logical relations to the Analysis. I have chosen the Third Argument for
my mind. Thus the former desire (or "want", as too many philosophers detailed examination because I believe that the fact that the validity of
insist on saying) was occurrent and the latter standing. I do not mean to (0) is the point at which the Analysis comes into conflict with that argu-
imply that an unconscious person has no standing desires. ment is of real philosophical interest, while the fact that the Analysis
13 See his 'Can' in Theory and Practice: A Possible Worlds Ap- is in conflict with premiss (6) of the First Formal Argument—rather than
proach", in M. Brand and D. Walton, eds. Action Theory (Dordrecht: with (5) or (4)—is pretty much an accidental result of the way in which
1976). Lehrer's analysis is not a conditional analysis, but it is sufficiently I chose to define 'render false'.
like one that various of its features may be "transferred" to typical con- 19 Cf. P. A. Nowell-Smith, Ethics (Oxford: 1957), 282. The main
ditional analyses with little modification. I should not want to give the sources of the arguments we shall be examining in the remainder of the
impression that the analysis that follows in the text is equivalent to chapter are Nowell-Smith's book, the Mind articles by Hobart, Nowell-
Lehrer's. Smith, and Smart cited in note 15 to Chapter I, and A. J. Ayer's "Free-
14 Can we define 'advantage' without using 'can' or some essentially dom and Necessity" in his Philosophical Essays (London: 1954).
equivalent term? (Does a 'can' grin residually up at us?) I am inclined 20 Except possibly 'unplanned' or 'unintended'. But this meaning
to think that 'advantage' hides a 'can', but I will not press the point. will be of no help to the proponent of the first strand. The thief's repent-
15 For an example of a compatibilist who, unless I have misunder- ance was certainly unplanned and he certainly had not intended to
stood him, has failed to grasp this simple point, see the papers by R. Foley repent, but these facts have no tendency to show that his act was not free.
238 NOTES TO PAGES 130-145 NOTES TO PAGES 146-150 239

21 / might want to, in fact. See sec. 5.5. it would be very hard to spell out in any precise form what is meant by
22 Cf. Hobart, 7. the words in scare quotes. But we can give examples of other sorts of
23 Of course the incompatibilist could make use of slippery-slope causal linkage—`deviant', I shall call them—between inner state and overt
arguments, too. Consider, for example, the hypothesis (M) that we con- act. Suppose, for example, that a certain thief desires to carry out his
sidered in our discussion of the Paradigm Case Argument in sec. 4.2. It mother's last wish and believes that repentance will best secure this
would be easy enough to construct a sequence of stories according to end; suppose that this mixture of desire and belief causes him to snarl
the pattern provided by the second strand of the Mind Argument, in at his own softness; suppose that his snarl is so violent that it precipitates
which the Martian device of that hypothesis is gradually turned into a a stroke, which in turn causes a temporary but frightening paralysis in
natural part of the agent's brain. the arm extended toward the poor-box; suppose the fear of God's
24 W. P. Alston has pointed out to me that many of the medievals, wrath thus engendered causes the thief to repent. If causation is transi-
most notably Aquinas, applied the notion of immanent causation to the tive, then a mixture of desire and belief has caused an act, but not in
human agent as well as to God. that natural or normal way that is what is intended by the adherents of
25 See R. M. Chisholm, "Freedom and Action", in Keith Lehrer, ed., a Davidsonian theory of the causes of action.
Freedom and Determinism (New York: 1966). Chisholm's later writings 32 ‘Np', it will be remembered, stands for 'p and no one has, or ever
on free will, such as the first chapter of Person and Object (La Salle, Ill.: had, any choice about whether p',
1976), continue to make use of the concept of immanent causation, but 33 By a "theory of choice" I mean a theory about what it is to have
this concept ceases to play a central role in Chisholm's exposition of his a choice, and not a theory about what it is to make a choice. A moment's
topic. See also R. Taylor, "Determinism and the Theory of Agency", reflection will show that these are not at all the same thing and that it
in Sydney Hook, ed., Determinism and Freedom in the Age of Modern is not obvious what the relation, if any, between them is. Consider the
Science (New York: 1958), and ch. 9 of Action andPurpose (Englewood man who is locked in a room and who does not know it. He may cer-
Cliffs, NJ: 1966). tainly make a choice about staying, despite the fact that he has no choice
26 "Actions, Reasons, and Causes", The Journal ofPhilosophy (1963). about staying.
27 Causality and Determination (Cambridge: 1971). 34 Alvin Plantinga has suggested to me that the thief may have had
28 The second strand of the Mind Argument may be looked at as an a choice about whether to repent owing to his having had a choice about
attempt to show that apparent acts that are undetermined are really whether, on the one hand, DB caused R, or, on the other, his desire for
acts if and only if apparent acts that are caused by a freakish demon are money and his belief that the poor-box contained money (DB*) jointly
really acts. It is interesting to note that the purely "Davidsonian" com- caused the event his robbing the poor box (R*). We should note that the
-

ponent of our second model entails that it is possible for a freakish two desire-belief pairs, DB and DB*, both actually obtained; according to
demon to cause what are literally our acts: the demon need only produce the theory Plantinga has proposed, what the thief had a choice about was
in us the appropriate beliefs and desires. A similar point was made in which of these two potential causes became the actual cause of an effect
sec. 4.2, in connection with the story of the Martian devices, though it appropriate to it. This may for all I know be the correct account of the
was not really necessary in that context (discussion of the Paradigm Case "inner state" of a deliberating agent who has a choice about how he is
Argument) for us to suppose that the Martian devices caused our acts: going to act. But if this account is correct, then there are two events its
it would have sufficed to suppose that they caused what seemed to be coming to pass that DB causes R and its coming to pass that DB * causes
or felt like our acts. R* such that, though one of them must happen, it's causally undeter-
29 A similar point can be made about the first model, since, pre- mined which will happen; and it will have to be the case that the thief
sumably, I can be the immanent cause only of events that are changes has a choice about which of them will happen. If this were so, I should
in natural parts of me. find it very puzzling and I should be at a loss to give an account of it.
30 Cf. R. M. Chisholm, "Comments and Replies", Philosophia (1978), 35 Strictly speaking, this puzzle arises only for someone who accepts
628-30. both what I accept and also the second model of action. (The reader
31 An internal state of a certain type might cause an act of a certain will remember that I have been pretending to accept the second model,
type in a way very different from the way in which states of that type as a literary convenience, though I in fact neither accept nor reject it.)
normally cause acts of that type. Suppose, for example, that mixtures But I am fairly sure that if I did accept (p) and the existence of free will,
of belief and desire cause our acts. Consider our thief, whose repentance as I do, and also accepted some particular theory of action, then I should
was caused by the co-presence of a desire to carry out his mother's last be forced to choose between the puzzling—though the puzzle might be
wish and a belief that repentance would secure this end. When one is told a different one—and the inconceivable. The words to which this note is
a story like the story of our thief, one naturally takes the causation of appended may be taken to represent what I should say if were able to
act by inner state to be of a certain "natural" or "normal" sort—though see my way clear to accepting some particular theory of action.

240 NOTES TO PAGES 150-165 NOTES TO PAGES 167-169 241

36 I wish to thank W. P. Alston for his advice about sec. 4.4, which refrains from performing that act: he may never even have considered
has led to changes in the structure of the argument of that section that performing that act. This distinction between two ways of failing to
I believe to be great improvements. Alvin Plantinga's comments on a perform a given act is of no importance for our present purposes. The
draft of sec. 4.4 have saved me from making several blunders. points made in the text would be equally valid if we had chosen to
37 Suppose a philosopher believes in free will and in "immanent examine a case in which the agent failed even to think of performing
causation" in the sense of the present section. How will he reply to the the act whose non-performance we are considering holding him respon-
argument that was presented in the text (p. 145) for the conclusion sible for.
that an appeal to immanent causation merely "sweeps the incompati- 11 Perhaps it is debatable whether this phrase designates a particular.
bility problem under the carpet"? I do not believe that this argument 12 But I doubt whether they can be anticipated. The objects of antici-
creates any new problems for him. I do not see why he should not pation and other "future-directed" attitudes would seem to be universals.
simply reply as follows: "The thief was the immanent cause of R. There- 13 Perhaps the last of these phrases could also be used to name an
fore he had a choice about whether DB was followed by R. Therefore event-universal. We seem to be using it this way if we say, "What Bill saw
he had a choice about whether R would occur, Therefore he had a choice happening in the garden happens all too frequently," but, I think, we
about whether his being the immanent cause of R would occur—simply use it to name a particular when we say, "What Bill saw happening in
because he had a choice about whether R would occur." The philos- the garden last night will live in infamy", or "could have been prevented
opher I have imagined appeals to the principle, 'If x and y have actually with a little foresight". The phrases "the fall of the Alamo" and "the
occurred and if x caused y, and if a certain agent had a choice about death of Caesar", however, seem to be suited only for denoting particu-
whether y would occur, then that agent had a choice about whether x lars: even if the Alamo had fallen twice, even if Caesar, like Lazarus, had
would cause y'. This principle seems quite reasonable. died twice, we could not say, "The fall of the Alamo has happened
twice" or "The death of Caesar has happened twice." This is not due,
or not due solely, to the presence of the definite article in these phrases,
for we can say, "The thing Bill fears most has happened twice."
14 From Davidson's contribution to a symposium on events and event-
CHAPTER V descriptions in Joseph Margolis, ed., Fact and Existence (Oxford: 1969),
1 Systeme de la nature X, xxi (tr. H, D. Robinson). 84.
2 This is in a way unfair. They believed they were giving arguments. 15 "The Individuation of Events" in Nicholas Rescher, ed., Essays
But their arguments were merely emphatic elaborations of the thesis in Honor of Carl G. Hempel (Dordrecht: 1969), 225.
that free will is an illusion. 16 Might logically, that is. A determinist could admit this point and
3 "Deliberation and Foreknowledge", American Philosophical Quar- consistently maintain that it was physically impossible for that event to
terly (1964). have had different effects.
4I use the term 'incompatible' very loosely. In the present loose 17 Or so it seems to me. Other philosophers—Cartesians, for example
sense, two contemplated acts may be incompatible simply because the —will have a different view of the matter.
agent has already decided not to perform both. 18 Cf. Saul Kripke, "Naming and Necessity", in Donald Davidson and
5 Or so it seems to me. W. P. Alston has told me that he doubts Gilbert Harman, eds., Semantics of Natural Language (Dordrecht:
whether the effects of a total abstention from deliberation would be as 1972), 312-14.
radical as that. They would certainly be pretty drastic, though. 19 A theory of event-particulars that is inconsistent with the view
6 Logic Matters (Oxford: 1972), 279. presented in the text is held by R. M. Martin and Jaegwon Kim. See
7 P. T. Geach, Reason and Argument (Oxford: 1976), 9. Martin's contribution to the symposium referred to in note 14, and, for
8 "The Principle of Alternate Possibilities", The Journal of Philo- Kim's latest published views on events, "Causation, Nomic Subsumption,
sophy (1969). and the Concept of Event", The Journal of Philosophy (1973). If we
9 These sections are a somewhat simplified version of my paper abstract from the particular twists that each of these authors gives to his
"Ability and Responsibility", The Philosophical Review (1978). Readers own account of events, we may say that, on the "Kim-Martin" theory,
interested in the fine points of the argument may wish to consult that the class of events is the class of substance-property-time triples. For
paper. example, Caesar's death is the triple (Caesar, being dead, 15 March 44
10 This schema and the instance of it that follows involve the agent's BC). (Strictly speaking, the term '15 March 44 BC' in the preceding
intentionally refraining from performing a given act. Of course not every sentence should be replaced with a term designating the precise instant
case in which we might want to consider holding an agent responsible at which Caesar died.) A Kim-Martin event happens just in the case that
for failing to perform some act is a case in which the agent intentionally its first term acquires its second term at its third term. However useful
 243
242 NOTES TO PAGES 171-178 NOTES TO PAGES 178-191
Kim-Martin events may he in certain contexts of discussion, I do not 25 I do not mean to give the impression that one never brings about
think it is correct to think of them as particulars. They are, rather, highly any state of affairs. For example, granting the correctness of the Warren
specified universals, just as the property being the tallest man is a highly Commission Report, Lee Harvey Oswald brought about C(Kennedy dies
specified -in fact, "definite"—universal (cf. note 24). This property, on 22 November 1963). But it is not true that Oswald brought about
though only one man can have it, is none the less such that it could have C(Kennedy dies). That state of affairs was brought about by God or by
been possessed by someone other than the man who in fact has it. Simi- Adam and Eve or by no one at all. Moreover, it is true that Oswald
larly, any Kim-Martin event that happens could have been caused by brought about the event-particular, Kennedy's death.
quite different antecedent events from those that in fact caused it. To 26 Strictly speaking, the antecedent of this conditional entails only
suppose that event-particulars have this feature is to violate my intuitions, the "minimal free-will thesis" (see sec. 3.9). But I suppose no one would
at any rate, about particulars. An additional problem: every Kim-Martin seriously maintain that the minimal free-will thesis was true and the
event is such that there is one particular moment (its third term) such free-will thesis false.
that the event must happen just at that moment if it happens at all. But 27 The function of 'even partly' will be explained in note 28.
surely Caesar's death might have happened at least a few moments earlier 28 If the words 'even partly' were omitted from the sentence-form
or later than it in fact did, just as a given man might have been born, or that 'Np' abbreviates, then (B) might be open to counter-example, Sup-
even conceived, at least a few moments earlier or later than he in fact was. pose, for example, that Smith kills the elder of the Jones twins and that
20 Nothing in PPPI corresponds to the parenthetical qualification the younger is killed by a bolt from the blue. It is at least arguable that
'that state of affairs obtains and' in this principle. So far as I can see, to in that case neither Smith nor anyone else is responsible for the fact that
say of a given event-particular that it "happens" is equivalent to saying both the Jones twins are dead. But then the following argument has true
that it exists. And, of course, there exist no events that do not exist. premisscs and a false conclusion if the words 'even partly' arc omitted
Thus there exist no events that do not happen. But states of affairs may from the reading of 'Np':
exist without obtaining, just as propositions may exist without being
N Both the Jones twins are dead;
true or properties without being instantiated. Cf. note 20, Chapter III.
21 Or, at least, this may he true. We shall later see that if Caesar N (Both the Jones twins arc dead j the elder of the Jones twins
would have been murdered by someone else if the conspirators hadn't is dead);
murdered him, then this state of affairs does not, strictly speaking, hence, N the elder of the Jones twins is dead.
obtain because the conspirators murdered him.
But it seems evident that, in the case imagined, Smith is at least partly
22 For a discussion of the propriety of applying the term 'universal' responsible for the fact that both the Jones twins are dead. A perhaps
to "states of affairs" in the present sense, see note 24. more troublesome case arises if we substitute "Tom throws a six' for
23 By Carl Ginet and Nicholas Sturgeon.
'p' and 'Tom plays dice' for 'q' in (B). I should be inclined to say that,
24 Perhaps some philosophers would be disinclined to call the prop- if Tom is indeed responsible for playing dice, then he is partly respon-
erty of being the heaviest statue there ever was or will be a universal, on sible for the fact that he throws a six, 'since he could have avoided throw-
the ground that a universal must be "sharable", must be capable of being ing a six by avoiding playing dice. Anyone who finds this response
exemplified by more than one object. And, for similar reasons, it might artificial and contrived may wish to replace the reading of 'Np' in the
be held that what I have called 'states of affairs' are not true universals,
text with 'p and no human being or group of human beings is respon-
since each of them either obtains or fails to obtain without further quali- sible for the fact that p or for any of the logical consequences of the
fication, whereas a state of affairs that was truly a universal should be
capable, say, of obtaining in 1943 but not in 1956, or of obtaining in fact that p' If we think of a logical consequence of a fact as a "part" of
that fact, then this reading may be regarded as a refinement of the read-
both Britain and the United States but not in France. Well, let us say
that our "states of affairs" and properties like being the heaviest statue ing given in the text. And it is clear that in this sense Tom can be respon-
are, if not "true" universals, at least cross-world universals. A property sible for a "part" of the fact that he throws a six, to wit, for the fact
or other abstract object is a cross-world universal if there are worlds w 1 that he plays dice.
and w, such that x falls under it in w i and y falls under it in to, and
x # y. (I use the words 'fall under' with deliberate vagueness; what
"falls under" a property is whatever has it; what "falls under" a state
of affairs is whatever arrangement of particulars realizes it.) If this CHAPTER VI
usage is an extension of traditional philosophical usage, it is a very 1 See sec. 1.3.
natural one; 1 call, for example, C(Gunnar kills Ridley) a 'universal' 2 Although, as we shall see, this supervention is merely statistical
hecause it is not "tied to" any given arrangement of particulars. if quantum mechanics is right.
244 NOTES TO PAGES 193-203
NOTES TO PAGES 204-223 245
3 Or, at any rate, this is as likely to be true as the corresponding to which it is logically equivalent as its sufficient reason. And this seems
thesis about classical states is in the case of classical mechanics. Cf. Richard plausible enough.
Montague, "Deterministic Theories", in Richmond H. Thomason, ed., 14 It follows from this result that PSR does indeed entail determinism,
Formal Philosophy: Selected Papers of Richard Montague (New Haven: for the simple reason that determinism is a thesis that, conceived very
1974), 332-6. abstractly, asserts only that certain true propositions are entailed by cer-
4 Cf. N. R. Hanson, "Quantum Mechanics, Philosophical Impli- tain other propositions; but if truth and necessary truth coincide, then
cations of ", in Paul Edwards, ed., The Encyclopedia of Philosophy (New any set of true propositions is entailed by anything whatever.
York and London: 1967), vol. 7, p. 45. 15 Cf. the discussion of Baron Holbach in sec. 5.2.
5 That is, in a theory according to which observable variables like 16 For another example, see the discussion of the relation between
position and momentum are determined. We have already seen that physical possibility and the power to act in sec. 1.4.
quantum mechanics itself is in a sense a deterministic theory, though it 171 use the word illusion in Freud's sense (Die Zukunft einer Il-
does not in general yield deterministic predictions of the values of ob- lusion, ch. 6): the alchemists' belief that lead could be turned to gold
servable variables. and Columbus's belief that a fifteenth-century ship could reach land by
6 See Max Jammer, The Philosophy of Quantum Mechanics (New sailing west from Europe were illusions, despite the fact that they were
York: 1974). true. For an example of a fine mind in the grip of scientism, read Carnap's
7 J. Andrade e Silva and G. Lochak, Quanta, tr. from the French intellectual autobiography in P. A. Schilpp, ed., The Philosophy of
by Patrick Moore (New York: 1969), 162. Rudolf Carnap (La Salle, Ill.: 1963).
8 By Jauch and Piron, for example. For a reasonably accessible pres- 18 I hope that the fact that I have discussed quantum mechanics at
entation of their argument, see ch. 5, 6, and 7 of Josef M. Jauch, Foun- some length has not given the opposite impression. My purpose in dis-
dations of Quantum Mechanics (Reading, Mass.: 1968). cussing quantum mechanics was simply to make it clear that, if it was
9 Cf. C. D. Broad, The Mind and its Place in Nature (London: 1925), ever reasonable to believe that science supported determinism (the gen-
113. eral thesis, that is, not the thesis that human beings are "for all practical
10 This is not to say that there are no plausible a priori arguments purposes" determined), it no longer is. Thus one reason our philosophical
for the conclusion that every event has a cause. Perhaps there are. forebears may have had for believing in determinism is not a reason we
11 Leibniz, of course, believed that this world's being the best of all have,
possible worlds was a sufficient reason for its being actual, despite the 19 This, of course, is a tendentious term. I would describe myself as
fact that this world is the best world does not entail this world is actual holding a "scientific world-view" if the term had not become associated
(for if that entailment did hold, then the other worlds would not be with a certain metaphysic (one having no real connection with the
possible; that is, they would not be possibly actual). achievements of science) that I reject.
This would seem to be a flatly incoherent position. If Leibniz is 20 One sometimes finds another fear mixed with the fear of being
right, then the possibility that the best of all possible worlds not be caught out: the fear of being swept into the dust bin. This is the fear
actual exists, though this possibility is, in his view, unrealized. But what that the progress of science—or perhaps changing social attitudes attri-
is the sufficient reason for this possibility's being unrealized?
butable to the progress of science—will leave one's views not so much
12 Someone might worry about whether this arguably impredicative refuted as irremediably old-fashioned.
definition leads to paradoxes of self-reference. It does not. The follow- 21 See Bertrand Beaumont, "Hegel and the Seven Planets", Mind
ing model shows this. Identify a proposition with the set of worlds in (1954).
which it is true, as we did in sec. 3.3. The conjunction of two propositions 22 This dilemma, strictly speaking, faces only the incompatibilist
will then be their intersection and a conjunct of a proposition will be who accepts our second model of action. But some such dilemma will
any proposition of which it is a subset. Thus, every proposition is a con- face any incompatibilist who holds any very explicit theory of action.
junct of itself. On this model, `P' denotes the set whose only member is
the actual world (provided there are at least two possible worlds: if there
is only one world, all propositions are necessary and 'P' fails to denote),
a set that can be singled out without paradox. Cf. this identification of
P in our model with the identifications made in the parenthetical remark
that follows in the text.
13 That criterion of identity is not really necessary for the argument.
If we rejected it, we should have to replace our second principle with
the principle that no contingent state of affairs can have a state of affairs
 INDEX 247

Ginet, Carl vi, 23 7n, 242n Maugham, W. S. quoted 24, 27, 227n
Gddel incompleteness results 225n McCall, Storrs 225n
McIntyre, Alasdair 234n
Hanson, N. R. 244n McKay, Thomas 233n
'hard determinism' 13f Melden, A. I. v
Harman, Gilbert 241n Mill, John Stuart 190
Hegel, G. W. F. 218, 245n Mind Argument, The 16f, 106, 126-52,
Hobart, R. E. vi, 134, 226n, 237n, 234n, 238n
238n miracle, the concept of a 141
Index Hobbes, Thomas 154, 155, 190 Montague, Richard 230n, 244n
Holbach, Paul-Henri Thiry, Baron d' Moore, Patrick 244n
153-60 passim, quoted 153, 245n moral responsibility 191, 291, 1041,
holism 2151, 217 161-89, 206ff, 243n
actuality 232n Darrow, Clarence 153
Honderich, Ted 226n, 229n
agent causation see immanent causation Davidson, Donald quoted 55, 137f,
Hook, Sidney 225n, 238n Nagel, Ernest 230n
Alston, William P. vi, 234n, 238n, 240n 140f, quoted 167, 168, 169, 226n, Hume, David 190 Narveson,Jan 230n, 23 In
An drade c Silva, J. quoted 196, 244n 241n Naylor, Margery vi
Anscombe, G. E. M. vi, 4, 139 deliberation 19, 30, 154-61, 2041,
immanent causation 4, 135-7, 145, necessary truth 96, 184
Aristotle 31, quoted 33, quoted 34, 209, 240n
1511, 225n, 238n, 240n - Neumann, J. von 196
quoted 228n determinism 2-8, 20-22, 58-65, 851,
astrology 198f incompatibilism 13, 15ff, 55-105,1111, Nowell-Smith, P. H. 226n, 237n
94, 95, 100f, I 11f, 183-8, 188f,
126-52 passim, 2221, see also corn-
Austin, J. L. 235n 190-204, 205-21 passim, 222f,
patibilism O'Connor, D. J. v
Ayer, Sir Alfred 228n, 237n 245n, see also 'hard determinism';
Ayers, M. R. v indeterminism 3, 126ff, 148, 191-7, 'of his own free will' 17
indeterminism; psychological deter-
205-21 passim, 2221
minism; 'soft determinism'
inevitability 23-9 Paley, William 218
Diodorus Cronus 31
Beaumont, Bertrand 245n introspection, knowing of one's free Paradigm Case Argument, The 16, I 06-
divine foreknowledge 225n
Brand, Myles 236n will by 2041 12, 150, 204, 234n, 238n
Broad, C. D. vi, 226n, 244n physical possibility and impossibility 3,
Edwards, Paul 244n
Broglie, Louis Victor de 196 Jammer, Max 244n 9, 85, 225n, 245n
Einstein, Albert 195
Epictetus 31 Jauch, Josef M. 244n Piron,,C, 244n
events, individuation of 167-70, 241n, Plantinga, Alvin vi, 79, 203, 230n,
Cahn, Steven 227n, 228n Kant, Immanuel 218 239n, 240n
242n
'can' 8-13
existence 232n Kenny, Anthony v, 229n possibilia 232n
Carnap, Rudolf 245n Kim, Jaegwon 241n possibility and impossibility 231n, see
causation 65, 138-41, see also im- Farrer, Austin v Kripke, Saul 241n also physical possibility and impos-
manent causation; Principle of Uni- fatalism 1, 23-54, 234n sibility
versal Causation, The Flew, Antony 106-12, quoted 107, Laplace, Pierre Simon, Marquis de 143, possible worlds 59, 78-82, 84, 85, 96f,
chance 1281, 144 234n 193f, 201, 230n 186, 230n, 232n, 233n
Chisholm, Roderick M. vi, 115, 136, Foley, Richard 226n, 236n Law of the Excluded Middle, The 50-4 predicting one's own decisions 225n
152, 163, quoted 235n, 236n, 238n Frankfurt, Harry vi, 19, 162-82 laws of nature 3, 5-8, 9, 141, 60-5, 85, Principle of Alternate Possibilities, The
Chrysippus 227n Franklin, R. L. v 92, I111, 185, 202, 225n, 230n 162-82
Cicero 31, 227n free will 8-13, 19-22, 86-91, 153-89, Lehrer, Keith vi, 113-14, .115, 119f, , Principle of Sufficient Reason, The 21,
compatibilism 13, 74-8, 101, 102-5, 204-21, 222f, see also free-will 225n, 235n, 236n, 238n 202-4, 244n, 245n
106-52, 222f, see also Compati- thesis, the; 'of his own free will' Leibniz, G. W. 244n Principle of Universal Causation, The
bility Problem, The; incompatibilism free-will thesis, the 14, 65-8, 87, 91, Lewis, C. S. 232n 3, 191, 223
Compatibility Problem, The 2, 16, 17, 243n 'libertarianism' 131 propositions 31-4, 53, 581, 227n, 228n
18, 19, 190 Frege, Gottlob 79 Lochak, G. quoted 196, 244n psychoanalysis 225n
compulsion 17 Freud, Sigmund 153, 245n Lucas, J. R. v psychological determinism 225n
Conditional Analysis Argument, The future contingencies see fatalism -Eucasiewicz, Jan 5, 225n psychological laws 63f
16, 19, 106, 108, 114-26, 150
Consequence Argument, The v, 16, 56, Gallois, Andre 231n Margolis, Joseph 241n quantum mechanics 52, 60, 191-7,
222, 237n Gardner, Michael 231n Martin, Raymond 231n 215, 216, 217, 230n, 243n, 244n,
'contra-causal freedom' 141 Geach, Peter 81,158, quoted 159, 240n Martin, Richard M. 241n 245n
248 INDEX

Reicher, Nicholas 241n

resporwibilit If: see moral responsibility
Robinson, 11, 1D, 240n
Ross, Sir David 228n
Taylor, Richard +A, 42, 4-50, quolet1
48, quoted 49, 156, 154, 226n,
227n, 229n, 238n
Thomas Aquinas, St, 258n
van Inwagen, Peter
Thomason, Richmond 11, 244n
Traditional Problem, The 2, 20-2,
scepticism 210-13 190-223
Schilop, P. A, 245n truth and falsity 314,53

An Essay on Free Will

"seieotilic world-view" 217, 245n at a time 34-43, 228n
scientism 215-18, 245n Twain, Mark. 65, 153
Seddon, George 232n
Smart, j. j, C, 225n, 237n universal causation: see Principle or
socio-biology 99-101 Universal Causation, The
'soft determinism' 13f universals 171,175-30,241n, 242n
Spinoza, Ranch 204
state of the world at an instant 59f, 84 van I n warn, Reiter 252n„ 235n, 287n,
state3 of attain, Individuation of 171, 2400
241n, 242n
"strongest motive" 225n
Sturgeon, Nicholu 242n
Walton, Douglas 236n
Wittgenstein, Ludwig 235n
Oxford: Clarendon Press, 983

Preface v-vi
I The Problems and How We Shall Approach Them 1
II Fatalism 23
III Three Arguments for Incompatibilism 55
IV Three Arguments for Compatibilism 106
V What Our Not Having Free Will Would Mean 153
VI The Traditional Problem 190
Notes 225
Index 246-8