Wayne State University

Law Faculty Research Publications Law School

1-1-1985

Practical Reasoning and Judicial Justification:

Toward an Adequate Theory
Vincent A. Wellman
Wayne State University, [email protected]

Recommended Citation
Vincent A. Wellman, Practical Reasoning and Judicial Justification: Toward an Adequate Theory, 57 U. Colo. L. Rev. 45 (1985).
Available at: https://digitalcommons.wayne.edu/lawfrp/163

This Article is brought to you for free and open access by the Law School at DigitalCommons@WayneState. It has been accepted for inclusion in Law
Faculty Research Publications by an authorized administrator of DigitalCommons@WayneState.
 PRACTICAL REASONING AND JUDICIAL
JUSTIFICATION: TOWARD AN ADEQUATE
THEORY
VINCENT A. WELLMAN*

I. INTRODUCTION ........................................ 45
II. A THEORY OF JUDICIAL JUSTIFICATION ................ 48
A. The Aims of a Theory of JudicialJustification....... 49
B. The Starting Points of a Theory of Judicial
Justification ....................................... 53
C Choosing Among Rival Theories .................... 59
III. Two DOMINANT VIEWS OF JUDICIAL JUSTIFICATION... 63
A. The Deductive Thesis .............................. 64
1. Deduction and Judicial Justification ............ 64
2. Evaluating a Deductive Theory of Judicial
Justification ................................... 68
B. The Analogical Thesis .............................. 80
1. Analogy and Judicial Justification .............. 80
2. Evaluating an Analogical Theory of Judicial
Justification ................................... 84
IV. PRACTICAL REASONING AND JUDICIAL JUSTIFICATION . 87
A. PracticalReasoning ................................ 88
B. Practical Reasoning and JudicialJustification ....... 92
C Evaluating a PracticalReasoning Theory ........... 106
V. PRACTICAL REASONING AND LEGAL THEORY .......... 109

I. INTRODUCTION

This article advances a theory of judicial justification: the argu-

mentation used by judges to justify their decisions.' The nature of
judicial justification and the appropriate criteria for evaluating judicial

* Assistant Professor of Law, Wayne State University Law School. Development of this paper
was funded in part by a Wayne State University Research Award. This article has benefited from the
comments and criticisms of Bruce Ackerman, John Stick, Michael McIntyre, and colleagues at Wayne
too numerous to list. Mistakes are, of course, the author's own responsibility.
1. I distinguish judicial justification from the more general category of legal reasoning. See infra
text accompanying notes 20-23. In particular, I argue that our concern over a court's justification for a
decision focuses on the validity of the arguments adduced in support of the decision. Judicial justifica-
tion is therefore distinct from the cognitive or psychological stages that an individual judge may go
through before reaching a conclusion in a particular case. See infro text accompanying notes 42-43.
46 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

arguments are topics of intense interest in jurisprudence. Interest in

these topics has spawned a variety of positions. Some commentators
have despaired of ever developing criteria of validity that could reflect
the full diversity of judicial arguments and decisions. This position is
commonly associated with the Legal Realists, 2 and especially with Je-
rome Frank, who expressed this attitude most forcefully.' Other writ-
ers have resisted despair and have argued for one or another particular
view of judicial justification. Two views have dominated the literature
of jurisprudence. Some contend that judicial argumentation is funda-
mentally deductive, involving the deduction of a decision from some
set of general legal standards. 4 Alternatively, others have argued that
judges reach and support their conclusions by a process of analogy to
other decided cases. 5
I contend that valid arguments in support of a judicial decision
are arguments of practical reasoning: the reasoning of ends to means.
Since Aristotle, practical reasoning has been suggested as an alterna-
tive to deduction as a form of reasoning. 6 Contemporary philosophi-
cal investigations have illuminated the nature of practical inferences
and the criteria of validity for practical arguments. 7 Drawing on these
investigations, I argue that practical reasoning provides a better theory
of judicial justification than either deduction or analogy.
Although scholars in law and philosophy have recognized the im-
portance of judicial justification, they have seldom examined judicial
reasoning as an independent question in jurisprudence.' Their differ-
ing views of judicial reasoning and their criteria of validity have been
largely adjuncts to their efforts to develop and defend some general
jurisprudential theory. Legal theorists have taken their respective po-
sitions on judicial argumentation in order to support or challenge the
broader theory of law in which their conception of judicial reasoning

2. For a useful account of the history of Legal Realism, see W. TWINING, KARL LLEWELLYN
AND THE REALIST MOVEMENT (1973). For a contemporary assessment of the Realistic American
Legal Culture, see B. ACKERMAN, RECONSTRUCTING AMERICAN LAW 6-22 (1983). It is a matter of
some debate - even among scholars involved in the movement -just who should be deemed a Realist.
Compare Pound, The Call for a Realist Jurisprudence, 44 HARV. L. REV. 697 (1931) with Llewellyn,
Some Realism about Realism - Responding to Dean Pound, 44 HARV. L. REV. 1222 (1931).
3. See J. FRANK, LAW AND THE MODERN MIND (1930).
4. See infra text accompanying notes 89-124.
5. See infra text accompanying notes 164-79.
6. See infra text accompanying notes 192-95.
7. 1 rely on the contributions of R.M. Hare and Anthony Kenny. See R.M. HARE, PRACTICAL
INFERENCES (1971); A. KENNY, WILL, FREEDOM AND POWER (1975). See infra text accompanying
notes 194-211.
8. There are a few notable exceptions. See Comment, Legal Reasoning: In Search of an Adequate
Theory of Argument, 59 CALIF. L. REV. 821 (1971); N. MACCORMICK, LEGAL REASONING AND
LEGAL THEORY (1978); M. GOLDING, LEGAL REASONING (1984).
1985] PRACTICAL REASONING AND JUSTIFICATION 47

figures. 9 They have not sought to uncover, for its own sake, the nature
and criteria of successful justification.
This failure to treat judicial reasoning as an independent issue has
had two unfortunate consequences. First, we suspect that theorists'
contentions about judicial justification are laden with the baggage of
their general jurisprudential theory. We are uncertain whether the
particular conception of judicial reasoning can stand on its own if we
choose to reject the accompanying theory of law. Second, those writ-
ers have failed to develop criteria by which we can evaluate different
accounts of judicial argumentation. We have, therefore, no clear sense
of how to choose the best theory of judicial justification from the avail-
able alternatives.
A central thesis of this article is that we cannot with any confi-
dence establish the nature of judicial reasoning and the appropriate
criteria of validity unless we first reflect on the nature of an adequate
theory of such reasoning. That is the topic of Section II of this arti-
cle.1 ° What should a theory of judicial justification explain? What
kind of theory will adequately explain the relevant phenomena? What
are the criteria for preferring one theory over another? Jurisprudential
scholarship has not heretofore focused on these questions. To develop
my answer requires that I refer to parallel investigations in philosophy
of science that bear on the nature of an adequate theory."t
Drawing on my conclusions about theory formation and prefera-
bility, I examine in Section III the two dominant views of judicial rea-
soning: the thesis that judicial reasoning is fundamentally deductive,
in Section IIIA,'2 and the contention that justification depends essen-
tially on analogical reasoning, in Section IIIB.' 3 I argue that neither
provides an adequate theory. In Section IV, I present my own view,
arguing that judicial justification is best understood as a form of prac-
tical reasoning.' 4 I build on the account of practical inferences devel-
oped by Anthony Kenny and R.M. Hare 5 and demonstrate that their
account can explain the important features of judicial argumentation.
The nature of judicial justification is an important topic in the
philosophy of law. The controversy surrounding that topic may be
compared to the debates in ethics about the nature of moral argument

9. See, e.g., Burton, Comment on "Empty Ideas": Logical Positivist Analyses of Equality and
Rules, 91 YALE L.J. 1136, 1140-47 (1982) (attacking deductive accounts of legal reasoning).
10. See infra text accompanying notes 20-87.
11. See infra text accompanying notes 79-86.
12. See infra text accompanying notes 89-163.
13. See infra text accompanying notes 164-89.
14. See infra text accompanying notes 192-259.
15. See infra text accompanying notes 194-211.
48 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

and the disputes in philosophy of science about scientific justification.

Questions about judicial reasoning and the validity of judicial argu-
ments, moreover, figure in a number of more general debates in juris-
prudence. It is beyond the scope of this article to consider the general
relationship, if there is one, between a theory of judicial reasoning and
a full-fledged theory of law.' 6 Some philosophers, at least, have been
inclined to advance their theory of law by urging a particular concep-
tion of judicial reasoning. Professor Dworkin's writings on the nature
of judicial decisions provide an example of just this inclination.' 7 In
challenging the view that judges in deciding cases have discretion to
rely on their political and moral beliefs, Dworkin has argued that le-
gitimate decision-making precludes any such discretion. We should
expect, he contends, a right answer to every legal controversy, and the
existence of a right answer means that there is no room in a valid
decision for the judge's personal beliefs."8 In the fifth and final section
of the article, 9 I consider Dworkin's contentions in light of my theory
of judicial justification, and I challenge his view that there must be a
right answer.

II. A THEORY OF JUDICIAL JUSTIFICATION

A theory is ordinarily advanced to explain and predict the rele-
vant phenomena, and an adequate theory will disclose the laws or
structures which underlie the phenomena and produce their observa-
ble regularities.2" Judicial justification poses some particular problems
for theory development, and evaluating the adequacy of a theory of
judicial justification requires some reflection on the nature and func-
tion of an adequate theory.
One problem for a theory of justification is identifying the phe-
nomena which the theory should explain. As the legal community
uses the term, "legal reasoning" may refer to wildly diverse instances
of cognitive activity: not only judicial argumentation, but also judicial

16. For an example of one theorist who has considered this relationship, see N. MACCORMICK,
supra note 8.
17. See especially R. DWORKIN, TAKING RIGHTS SERIOUSLY 1-130 (rev. ed. 1978).
18. Id.
19. See infra text accompanying notes 260-78.
20. "What is it to supply a theory? It is to offer an intelligible, systematic, conceptual pattern for
the observed data. The value of this pattern lies in its capacity to unite phenomena which, without the
theory, are either surprising, anomalous, or wholly unnoticed." N. HANSON, PATrTERNS OF DisCOV-
ERY 121 (1958). Legal theory has been generally unappreciative of the nature of an adequate theory
and how this issue might affect current debates. There are a few notable exceptions. See Hart, Defini-
tion and Theory in Jurisprudence,70 LAw Q. REV. 37 (1954); Schauer, An Essay on Constitutional
Language, 29 UCLA L. REV. 797, 814-28 (1982); Nance, Legal Theory and the Pivotal Role of the
Concept of Coercion, 57 U. COLO. L. REV. 1 (1985).
1985] PRACTICAL REASONING AND JUSTIFICATION 49

intuitions, lawyers' litigation strategies, and even scholarly research

paradigms. The myriad phenomena which one might plausibly in-
clude in the category of legal reasoning are so various that the cate-
gory's full range may exceed the explanatory capacity of any single
coherent set of theoretical principles.
It is simpler, and likely more fruitful, to theorize about a subtype
of these diverse phenomena. The arguments adduced by judges to jus-
tify their decisions are a useful focus for theoretical attention. We re-
gard judicial decisions as particularly significant in our legal system.
Their justifying arguments are thought to manifest important facets of
that system's nature and operation.2" Further, it is plausible that as-
pects of other types of legal reasoning can be traced to the nature and
form of the judicial decision. 22 Reflecting the importance of judicial
reasoning, this article advances a theory to account for the important
features of judicial justification. My approach differs, then, from that
of writers who have sought to explicate the nature of legal reasoning
more generally.23
In part A of this Section I consider the object of a theory of judi-
cial justification: what it is that the theory should explain. In develop-
ing a theory, some phenomena must be taken as the starting points for
theoretical attention. Else, the theorist could not begin to formulate
the laws and structures which could comprise the theory. I articulate,
in part B, those features which could provide the starting points for a
theory of judicial argumentation: what the legal community expects of
judges in their efforts to justify their decisions. Finally, in part C I
provide the bases for choosing among rival theories. To theorize
about judicial justification requires us to understand the appropriate
function of a theory of judicial justification. Is a theory of justification
descriptive, conceptual, or normative? I argue that an adequate the-
ory of judicial justification must play all three roles, and that this re-
quirement motivates the criteria for choosing one theory over another.

A. The Aims of a Theory of JudicialJustification

An adequate theory of judicial argumentation should ascribe a

logical structure to the arguments adduced by a judge. We should be
able to identify, in terms of the theory, premises and conclusions and
to describe the kind of inferences which are supposed to relate the
former to the latter.

21. See, e.g., H. HART & A. SACKS, THE LEGAL PROCESS: BASIC PROBLEMS IN THE MAKING
AND APPLICATION OF LAW iii-vi, 110-89 passim (tentative ed. 1958).
22. N. MACCORMICK, supra note 8, at 10-13.
23. See, e.g., J. STONE, LEGAL SYSTEM AND LAWYER'S REASONING (1964).
50 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

Even a well-developed theory of judicial reasoning would be use-

less, however, if it accounted indiscriminately for each judicial argu-
ment without regard to its logical form or its success in justifying its
conclusion. The legal community's critical responses to judges' argu-
ments presuppose that we may distinguish among attempts at justifica-
tion. Academics and practitioners acknowledge some arguments to be
well-reasoned, but treat others as flawed. Moreover, we argue with
one another about our different assessments of particular arguments.
These judgments and our attempts to dispute them with one another
presuppose that there are standards by which we can meaningfully
24
evaluate judges' reasonings and discriminate among their arguments.
A theory of judicial reasoning which accounted for every argu-
ment without discriminating the valid from the invalid would subsume
each inference employed by a judge, no matter how specious. A the-
ory so promiscuous would not aid us in understanding the legal sys-
tem. Some arguments, it is clear, are invalid, so the theory should
provide criteria which explain the legal community's judgments about
which justifications succeed and which fail.
An adequate theory of judicial justification is similar, in a number
of respects, to the theories advanced in linguistics to explain speech
patterns within a language.25 Speakers of a language are capable not
only of generating sentences in that language, but also of distinguish-
ing correct from incorrect sentences. To explain and illuminate a
speaker's underlying capacities, a linguist develops a grammar - a
general structure for producing acceptable sentences within the lan-
guage and for manipulating those sentences in acceptable ways.26 An
adequate grammar can explain both the actual and the expectable re-
sponses of speakers to the grammaticality of possible sentences. 27 Fo-
cusing on the speakers' judgments of grammaticality, linguists seek to
understand the language's full variety.
Similarly, a theory of legal reasoning should indicate the struc-
tures which underlie our evaluations of judicial argumentation. Its
data are not the surface features of judicial arguments, but rather the
critical reactions of judges and lawyers to particular arguments. It
should propound criteria for evaluating judicial argumentation: for
distinguishing good or acceptable judicial reasoning from bad or
inadequate.

24. See infra text accompanying notes 31-34.

25. The argument in this Section relies substantially on the notion of reflective equilibrium ad-
vanced by John Rawls for moral philosophy. See J. RAwLS, A THEORY OF JUSTICE 3-22, 46-53 (1971).
Rawls similarly invokes linguistic research as a useful comparison. Id. at 47.
26. See, e.g., J. GRINDER & S. ELGIN, GUIDE TO TRANSFORMATIONAL GRAMMAR 7-66 (1973).
27. Id. at 40-66.
1985] PRACTICAL REASONING AND JUSTIFICATION 51

This does not mean that judges necessarily craft their opinions in
terms of inference forms which a theory would recognize as forms of
valid reasoning, nor that the legal community's evaluations of judicial
justification will refer to the standards propounded by the theory. In
many cases, the conceptual apparatus proposed by the theory and its
normative criteria will be constructs only.28 An adequate theory of
judicial argumentation should explain our discriminations among the
argument forms employed. It need not explain those discriminations
in the same way, however, nor in the same terms as might figure in our
pre-theoretical discussions of judicial arguments.
It would be possible, of course, to construct a view of law and the
legal system in which our various evaluations of legal reasoning were
treated as otiose and inefficacious. For example, an extreme version of
Legal Realism might hold that the reasons provided by a judge in jus-
tification of a holding are insignificant, except as they figure in certain
mythic or political roles of the judiciary. 29 Any attention to criteria
for judicial justification would be useful only insofar as it revealed the
contours of that role. Similarly, a radical version of law and econom-
ics might contend that the only criterion for evaluating a judge's deci-
sions is the extent to which his decision advances the overall efficiency
of the legal and economic system. The judge's stated reasons would be
superfluous, on this view, except to the extent that his opinion leads
economic agents to efficient behavior.3 0 The legal community's com-
mitment to the notion that there are shared standards of justification is
thus not by any stretch of the imagination necessary, and theories
which discount the importance of these standards may well illuminate
features of the legal system which we might otherwise not notice. But
the fact remains that we rely on shared evaluations of judicial infer-
ences in our responses to legal decisions. We need an account of our
evaluations, if such can be managed.
It is arguable, however, that our evaluations of judicial arguments
are so divergent and inconsistent as to indicate that there are no

28. Cf. id. at 11-14 (linguists' aim to explain speaker's competence leads to posit of "deep" struc-
ture and of rules relating deep to surface structure).
29. 1 do not claim that any Realist actually held this extreme position. Some of Jerome Frank's
remarks could be read as advocating this view about judges' reasons, see J. FRANK, supra note 3, but
my argument depends on the possibility of this kind of extreme position and not on the correct interpre-
tation of Frank's views. For a thoughtful analysis of Frank's position, and the change in his thoughts
across his career, see J. GLENNON, THE ICONOCLAST AS REFORMER (1985).
30. Similarly, there may be no actual scholar of the law and economics analysis who has taken
this radical position. Ronald Dworkin has, on occasion, criticized Richard Posner for something like
this view, see, e.g., R. DWORKIN, supra note 17, at 97, but my point depends only on the possibility of
such a radical position, not on the correct interpretation of Posner's work.
52 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

shared criteria for justification. 3' We have all seen, it might be argued,
collections of cases where judges reach contrary holdings on seemingly
indistinguishable facts.32 Moreover, when evaluating the decision in a
particular case, different lawyers will often reach inconsistent judg-
ments; what seems artfully distinguished to one lawyer might be fatu-
ous error in the eyes of another. This morass of divergent, even
contradictory evaluations, it might be contended, indicates that at bot-
tom no meaningful criteria of validity could ever be discerned for judi-
cial argumentation.
Lawyers and judges sometimes differ in their evaluations of some
judicial arguments. What is more significant, however, is that they
dispute their conflicting evaluations. They will argue with one another
about the validity of a given decision, or about the existence and force
of a particular rule, notwithstanding their conflicting evaluations.
This reflects the legal community's assumption that criteria obtain for
recognizing good legal reasoning. Without common criteria for as-
sessing the validity of arguments, there would be no grounds from
which one could rationally establish the argument's invalidity.33
We are committed to the notion that there are standards of justifi-
cation which either are or are not satisfied by judges in their argu-
ments. An adequate theory of legal reasoning must take seriously the
various ascriptions that we are all trained to make about such reason-
ing - its instances, successes and failures 34 - and it must elucidate

31. It seems doubtful that anyone has asserted a nihilist position of such a radical form, although
the allegation of incorrigible inconsistency could be read into some of Jerome Frank's aspersions on
legal reasoning. See J. FRANK supra note 3. Two trends in The Critical Legal Studies tradition also
raise questions about the consistency of our evaluations of judicial arguments, although I do not mean
to attribute the position outlined in the text to any author. First, writers in the CLS tradition rely on
the presence of alleged "contradictions" in legal doctrines as the basis for criticizing those doctrines and
the purported structure of liberal idealogy which generates the doctrines. See, e.g., Feinman, Critical
Approaches to Contract Law, 30 UCLA L. REV. 829, 849 (1983). (I have doubts about the force of
these claims. See infra note 131.) Second, some writers in the CLS tradition have suggested that focus-
ing on legal reasoning as a distinct form of argument amounts to the vice of "formalism." See, e.g.,
Unger, The Critical Legal Studies Movement, 96 HARV. L. REV. 563, 565 (1983). The argument on this
score seems to be that, to the extent that our evaluations of judicial justification are held to depend on
controversial political or moral beliefs, our judgments of the validity of judicial arguments may be
incorrigibly at odds with one another. For an insightful discussion of formalism by a writer outside the
CLS tradition, see S. BURTON, AN INTRODUCTION TO LAW AND LEGAL REASONING 169-215 (1985).
32. Compare, e.g., Morsinkhoff v. DeLuxe Laundry & Dry Cleaning Co., 344 S.W.2d 639 (Mo.
App. 1961) with McIntosh v. Murphy, 52 Hawaii 29, 469 P.2d 177 (1970).
33. Cf. A.J. AYER, THE PROBLEM OF KNOWLEDGE 35-90 (1956) (discussing epistemological
scepticism and refutations thereof); G. RYLE, DILEMMAS 94 (1962) ("There can be false coins only
where there are coins made of the proper materials by the proper authorities.").
34. My claim that an adequate theory of judicial justification must "take seriously" the legal
community's various critical remarks about judicial arguments parallels the arguments made by Ronald
Dworkin that legal philosophy must take rights seriously as a part of the legal landscape. See R.
DWORKIN, supra note 17, at 14-15, 184-86, 204-05.
1985] PRACTICAL REASONING AND JUSTIFICATION 53

the common criteria of validity which explain our critical evalutions of

judges' arguments.

B. The Starting Points of a Theory of JudicialJustification

Our evaluations of judicial justification presuppose certain fea-
tures of the legal process which bear on the role of judges and the
judicial decision. These presuppositions are revealed in certain com-
mon expectations we bring to bear on judges' attempts to justify their
decisions. Many of these expectations are unexceptional, even obvi-
ous, and articulating the obvious may seem pointless. Our shared ex-
pectations, however, can help provide the basis of our theory of
judicial justification.

1. Justification and Reasons

We expect judges to try to justify their decisions.3 5 We require
that they state - generally in a written opinion - the warrant for
their conclusions. By way of comparison, it is clear that judges are not
regarded as oracles; we would not tolerate their pronouncing decisions
on the basis of visions or visitations. Nor are judges logical black
boxes whose outputs are expected to follow mechanically from some
set of inputs.3 6 Judges issue decisions, to be sure, but our legal system
places certain constraints on their decision-making, most notably that
their results be supportable according to certain standards of justifica-
tion. In reaching their conclusions they are expected to reflect on the
various dimensions of the case at hand and respond in sometimes sub-
tle ways.
Our expectation that judges justify their decisions is only part of
our more elaborate theory of the judicial process. We expect judges to
justify their decisions but recognize also that the judicial decision is
significant for more than just the arguments adduced. We distinguish
between the holding of the case and the explanation for the holding,
37
and the holding itself is significant apart from the explanation.
Whether the deciding judge's rationale is silly or cogent, the decision
may serve as precedent for later judges.38

35. See generally H. HART & A. SACKS, supra note 21. By itself, this requirement does not mean
that the judge's justification must take the form of substantive reasons. It is conceivable, of course, that
an adequate justification would consist in the judge's citing his institutional authority to decide the case
before him, and then giving his decision, without references to substantive grounds for his decision.
36. Cf. Dworkin, Is Wealth a Value?, 9 J. LEGAL STUD. 191, 221-23 (1980) (arguing that judicial
decisions are not properly explained by an algorithm).
37. For a classic review of the distinction between holding and dicta and of competing views of
how to tell the one from the other, see R. CROSS, PRECEDENT IN ENGLISH LAW 35-40 (2d ed. 1968).
38. See N. MACCORMICK, supra note 8, at 84-85; R. DWORKIN, supra note 17, at 110-23.
54 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

Judicial justification may take different forms, but it is central to

our expectations of judges that they give reasons for their decisions -
grounds for concluding as they have and not in some alternative man-
ner. 39 Just what counts as a reason is a frustratingly flexible and con-
text-dependent notion. On the one hand, we do not expect every
opinion to set forth a complete justification of the conclusion, reaching
back in every case to legal first principles. Our doctrine of stare decisis
means that a judge may, at least on occasion, take some legal questions
as effectively decided.4' Moreover, in weighing the applicability and
strength of authority, the judge need not have considered every possi-
ble distinction in otherwise unproblematic doctrines. Some counter-
arguments may be dismissed cursorily, and some proffered distinctions
may be rejected without ado. On the other hand, the reason-giving
requirement is not trivial. We may distinguish between true reasons
and statements which are reasons in only a Pickwickian sense, and we
expect the decision to rest on the former sort of grounds for warrant.4 1
The judge may not legitimately ignore worthy legal arguments, nor
may he rely on obvious howlers as the "grounds" of his decisions.
Given the flexibility of our notion of a reason in our expectation
of judges, it is not surprising that accounts will differ as to whether a
particular argument has fulfilled this requirement. These divergences
notwithstanding, we may note, in broad generality, certain distinct
features of our standards on this score. For example, we do not nor-
mally demand a psychological account of how judges actually reach
their conclusions. We distinguish between the reasons which might
warrant a given result and a history of the cognitive stages undergone
in deciding that the result in question was warranted.42 However the
judge may have come to the opinion, our concern with judicial reason-
ing focuses instead on the arguments he adduces to justify the deci-
sions he makes.4 3

2. Like Treatment

Foremost among our standards for this reason-giving task is that

39. See, e.g., Sartorius, The Justification of the JudicialDecision, 78 ETHICS 171, 172-75 (1968).
40. See R. CROSS, supra note 37.
41. See N. MACCORMICK, supra note 8, at 14-16.
42. This distinction is commonly expressed in terms of a difference between the context of discov-
ery and the context of justification. See R. WASSERSTROM, THE JUDICIAL DECISION 25-30 (1961).
The importance of this distinction should not tempt us to discount the context of discovery as an
interesting focus of investigation, for judicial hunches and intentions may well be interesting and even
revealing. But our demand for justification does not entail a demand that the judge have employed the
correct hunch or intention.
43. Id. at 27.
1985] PRACTICAL REASONING AND JUSTIFICATION 55

judges should treat like cases alike." This requirement has been re-
lated to a hope in the legal system for substantive justice,4 5 although it
is more likely that treating like cases alike is necessary but not suffi-
cient for substantive justice.4 6 The requirement of like treatment has
also been related to our demand that judges approach cases without
prejudice toward particular parties and without interest in the person-
ality or identity of the parties.4 7 Whatever its virtues, our expectation
that judges treat like cases alike constrains the kind of reasons which
could be adduced in favor of a legal conclusion. It requires that a
decision for one party cannot be warranted unless a decision for an-
other party whose legal position is entirely identical with the first's
would also be warranted.48
Formally, this means that the judge's warrant must be general.
He must be able to refer to a statement, universal in scope, which
identifies a class of situations or parties and urges, if it does not com-
pel, a decision in favor of the victorious side. The nature or source of
this universal statement may vary. It might describe a large or small
class of cases or litigants; it might be about principles or rights, or
about technical doctrines instead; it might itself be justified by legisla-
tion or from past legal decisions. But the proposition's universal form
guarantees that cases which are alike in the respect described will all
be treated alike.
However important it may be that like cases be treated alike, we
have only a flexible and context-dependent criterion by which to assess
likeness. The use by judges of spurious distinctions among similar
cases would render this requirement impotent without other, addi-
tional constraints on the reason-giving task. In general, it is easier to
note what is inappropriate than it is to define when an argument is
relevant or appropriate. It is not enough for a decision to be justified
in terms of any universal proposition without regard to the content of
the proposition. It would not be enough, for instance, that the win-
ning party have red hair, or blue eyes, even if the judge seems to accept
the implication that all similarly situated redheads would also have
triumphed. It is also clear that the justification must reflect, to some
degree, the particularities of the case at hand. To cite an example of
Ronald Dworkin, we would be troubled if the judge decides cases
merely by following a logical algorithm based on nothing more than
the order in which the cases were heard, even if that algorithm pro-

44. See, e.g., Winston, Treating Like Cases Alike, 62 CALIF. L. REV. 1 (1974).
45. Id.; N. MACCORMICK, supra note 8, at 73-99.
46. Winston, supra note 44.
47. N. MACCORMICK, supra note 8, at 72-99.
48. Id.
56 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

vided a logically effective account of the correct legal decision.4 9

3. Rules of Law
We expect that in giving reasons judges will appeal to certain spe-
cial propositions styled "rules of law." 5 The notion of a rule of law
has been a hotly contested issue in twentieth century jurisprudence.
Debates flourish about various features of rules - about, for example,
their force in judicial justification and their logical structure. 5 None-
theless, our practice acknowledges that general propositions of law
will figure significantly in judicial justification.
Many of the contentions about rules do not bear on the nature of
an adequate theory of legal reasoning. For example, some of the con-
troversy surrounding rules relates to past theories about legal process
and political disputes about the role of the judges. 52 These disputes
remind us, if we needed reminding, that judges are important for more
than just their opinion-writing skills, and that judicial decisions are
significant beyond the arguments adduced to justify the result. There
is an obvious political dimension to the judicial role,5 3 and we may
wish to evaluate a judge's decisions in terms of the political cast of his
holdings. Unless we suppose that the only standard of judicial justifi-
cation is political, however, we need an account of judicial argumenta-
tion and its validity. This requires an account of judges' use of legal
rules.
Legal scholarship of this century has been drawn to consider the
nature and significance of rules by the challenges of the Legal Realists.
They disputed what they thought to be the received view of legal rea-
soning, which, as the Realists understood it, ascribed to rules a distinct
status and function within the legal system. They hoped to undermine
the received view, and they challenged its picture of rules and
reasoning.
Many of Realism's objections to the received view are better seen
as disputes over which features may properly be ascribed to legal rules
than as a complete denial of their existence in the legal system. Some
of the Realists, for example, complained that judges feel free to in-
dulge hunches and intuitions in coming to a decision about the proper

49. Dworkin, supra note 36.

50. See Sartorius, supra note 39, at 176-77.
51. Compare, e.g., Raz, Legal Principles and the Limits of Law, 81 YALE L.J. 823 (1972) with R.
DWORKIN, supra note 17, at 14-80.
52. See T. BENDITr, LAW As RULE AND PRINCIPLE 22-25 (1978).
53. See R. DWORKIN, supra note 17, at 81-130; M. HORWITZ, THE TRANSFORMATION OF
AMERICAN LAW, 1780-860passim (1977).
1985] PRACTICAL REASONING AND JUSTIFICATION 57

disposition of a controversy.54 From that they would infer that legal

rules are impotent, for rules don't seem to constrain judges from in-
dulging their hunches. That inference could follow, however, only if a
rule's essential function is to limit the judge's cognitive processes. But
we have already noted that a judicial decision may be justified even if
the judge's cognitive paths to the decision are circuitous.5 5 If legal
rules are significant for their role in judicial justification, then it is
hardly dispositive to document their lack of influence on a judge's
hunches or intuitions.
Similarly, some Realists asserted that one cannot predict a given
decision with any certainty, no matter how well one may know the
relevant rules. 56 It may be that we want judicial decisions to engender
stability in the social order,57 and that predictability of results is useful
in promoting stability. But, from our lack of certainty about the appli-
cation of some rules it hardly follows that rules play no role in judicial
justification. A better conclusion may be that rules do not necessarily
apply in a predictable fashion. In short, we may debate whether legal
rules have certain particular features, and, as a result, we may con-
clude that rules do not have all the features we might wish. From
such disappointment, however, we need not conclude that rules are
fictitious.5 8
The Realists are an interesting chapter in American legal scholar-
ship, and it is worth speculating on the sources and motives for their
challenges to legal rules.59 But in pursuing a theory of legal reasoning
we must reject their contentions about the fictitiousness of rules as
being, for the most part, rhetorical overkill. Legal practice is striking
in converging on the notion that there are rules of law and that they
figure significantly in the justificatory task. As Wasserstrom has ob-
served, whichever analysis of rules we might choose to embrace and
however much we might doubt that legal rules determine legal results,
it remains a striking feature of our legal system that when a lawyer
refers to the "rule of Hadley v. Baxendale" her colleagues will under-
stand her reference.' Her colleagues also understand the significance
of the rule regarding consequential damages for breach of contract
and, except in the rare case, will agree on its application.

54. See J. FRANK, supra note 3, at 3-13, 46-52.

55. See supra text accompanying notes 42-43.
56. See, e.g., J. FRANK, supra note 3, at 3-13.
57. See H. HART & A. SACKS, supra note 21.
58. See T. BENDITT, supra note 52, at 22-42; Dickinson, Legal Rules: Their Function in the Pro-
cess ofDecision, 79 U. PA. L. REV. 833-38 (1936).
59. See Gilmore, Legal Realism: Its Cause and Cure, 70 YALE L.J. 1037 (1961); G. GILMORE,
THE AGES OF AMERICAN LAW 77-91 (1977).
60. R. WASSERSTROM, supra note 42, at 35-36.
58 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

4. Other Concerns

An adequate theory of judicial justification must reflect the fact

that in addition to reasoning from rules judges must, on occasion, rea-
son about rules. A theory should be rich enough to explain why some
decisions about rules are justified and others are not. In some cases,
the court must reason to a new rule or a new application or formula-
tion of an existing rule.61 Some controversies are such that more than
one rule may apply, and the judge must decide, among the various
applicable rules, which rule's application would be warranted.6 2
These forms of reasoning about rules indicate that an adequate theory
of justification cannot be limited to just the application of rules to par-
ticular cases.
Along with rules, judges characteristically appeal to principles
and policies of law. On at least one view, it is arguments of principle
and policy which provide the justification for the extension, reformula-
tion or revision of legal rules. 63 It is a matter of substantial contro-
versy in the literature just how properly to distinguish principles and
policies from rules, and, for that matter, how to distinguish principles
from policies.6 Like rules, principles and policies are usually held to
be general, and together they are sometimes referred to as "standards"
of the law. 65 An adequate theory of justification should account for
legal standards, of whatever kind.
Another salient form of reason offered by judges to justify their
conclusions is the analogy. 66 Generally speaking, an analogy is an ar-
gument from demonstrated or assumed resemblances between cases to
an inferred resemblance that is relevant to the dispute.67 It is clear
that judges will, in their justifying arguments, rely on the claim that
the controversy at hand is similar in certain vital respects to another
dispute. The demonstrated similarity therefore provides some reason
for treating the case at hand in a fashion comparable to the previous
decision. The logical structure of analogical arguments is a matter of
some controversy - for example, it has been disputed whether analog- 68
ical arguments rely in any important way on rule-like statements -

61. See H. HART & A. SACKS, supra note 21.

62. See Winston, supra note 44, at 17-18, 22-23.
63. See H. HART & A. SACKS, supra note 21.
64. Compare, e.g., Raz, supra note 51, with R. DWORKIN, supra note 17, at 22-28, 71-80, 82-89.
See also id. at 254-301 (discussing criticisms of Dworkin's distinction).
65. See R. DWORKIN, supra note 17, at 22.
66. See, e.g., S. BURTON, supra note 31, at 25-40; M. GOLDING, supra note 8, at 44-48.
67. See, e.g., M. GOLDING, supra note 8, at 44-48.
68. See infra text accompanying notes 169-75.
1985] PRACTICAL REASONING AND JUSTIFICATION 59

but it seems unproblematic that judges do at times resort to analogical

inferences to justify their decisions.
There are other features of judicial justification which have been
bruited about in the literature: whether, for example, judicial justifica-
tion is inherently and correctly political or ethical in character, 69 or
whether to be legitimate, judicial decisions must rely on "neutral"
principles.7 ° Many of the divergent claims about judicial argumenta-
tion depend on problematic contentions about the proper role of the
judiciary in our system of government. These problematic contentions
cannot be included in the list of features which an adequate theory
must accommodate, since their inclusion would unwarrantedly deter-
mine the shape of the theory.7 1 We can hope that an adequate theory
of judicial justification will aid us in the evaluation of these related
debates, but we cannot require of a theory that it agree with one side
or the other. What remains is to articulate just how we should choose
one theory of justification over another, for then we might properly
commit ourselves to one or another of these problematic visions of the
law.

C. Choosing Among Rival Theories

In discussing theories of legal reasoning, it is common to distin-
guish three putatively different kinds of theories: descriptive, concep-
tual and normative.7 2 This beguiling division should be carefully
limited. To be sure, we may distinguish descriptive from normative
remarks and note that they have different logical features. The logical
differences mean that we may describe a justifying argument as ana-
logical in nature and, without pain of contradiction, also argue that
the analogy is ill-drawn or inappropriate. But the distinction may not
apply in any useful fashion to theories. Any unidimensional theory of
judicial justification - one that was merely descriptive, solely concep-
tual, or purely normative - would be inadequate.7 3 A useful theory

69. See R. DWORKIN, supra note 17, at 115-18, 123-30.

70. See, e.g., Wechsler, Toward Neutral Principles of ConstitutionalLaw, 73 HARV. L. REV. I
(1959).
71. Rejecting these contentions about judicial argumentation is part of the process of arriving at a
reflective equilibrium concerning judicial justification. See J. R.AwLs, supra note 25.
72. In rough terms this classification would amount to the following differences. A descriptive
theory of legal reasoning would describe the various instances of legal reasoning; a conceptual theory
would identify certain concepts and notions at work among the various instances of legal reasoning; and
a normative theory would posit a standard for the evaluation and perhaps improvement of the practice
of such reasoning, or perhaps an ideal against which various instances could be measured. See, e.g.,
Soper, Legal Theory and the Obligation of a Judge: The HartlDworkinDispute, 75 MIcH. L. REv. 473
(1977). Professor Dworkin has responded to Soper's criticisms along the same lines as I develop here.
See R. DWORKIN, supra note 17, at 350-52.
73. Suppose, for example, that we could formulate a theory of judicial justification which was
60 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

of legal reasoning will, instead, integrate all three dimensions.

To illustrate this point, reconsider the question raised by the
Legal Realists of rule nihilism." The true rule nihilist (as opposed to
someone who merely doubts the particular attributes of legal rules)
would contend that the very notion of a legal rule is without founda-
tion, and that any theory of justification predicated on such entities is
therefore untenable.7 5 It can be seen that if rule nihilism is supposed
to be supported by observable legal practice, then it is conceptually
incoherent.
To assert meaningfully that there are no such things as legal rules
at work in our legal system requires some meaningful criteria in terms
of which we could review legal practice to see whether rules play a role
in the legal system. Else, the nihilist would have no intelligible basis
for denying that rules obtain and that they figure in legal reasoning.
But this means that the nihilist's argument presupposes a coherent
conception of rules that a judge could use in justificatory arguments.7 6
So, the nihilist cannot sensibly maintain on the basis of observable
legal practice that rules are nonsense and that judicial reasoning can-
not involve them.7 7

descriptive only - that is, without any conceptual or normative commitments. Such a theory would
provide us with nothing more useful than a laundry list of the observable instances of judges' reasoning.
Without some conceptual apparatus there would be no basis on which to categorize instances of such
reasoning into useful types or subtypes; there would be no warrant for comparing these various infer-
ences; and there would be no foundation for ascribing an argumentative structure to any of the various
instances. Even a theory which was descriptively and conceptually powerful enough to explain the
observable examples of judicial argumentation would likely be inadequate without some normative
component. For, unless we could distinguish valid from invalid inferences, the theory would subsume
all instances of such argumentation, no matter how specious.
By the same token, no adequate account of legal reasoning could be normative or conceptual only,
without a substantial descriptive commitment. If the proposed theory is to count as a theory of judicial
justification, then its evaluative standards and conceptual structures must be such that instances of
those norms or structures would be recognized by the legal community as judicial arguments. Else, the
theory would vitiate the common criteria of judicial argumentation that, we have seen, our practice
presupposes. The conceptual structures suggested by a theory must be discernable in acknowledged
instances of judicial justification if the theory is to count as a theory of legal reasoning. Similarly, we
would reject a theory whose norms were satisfied by no recognizable instance of judicial argument. If
the theory's commitments could not be predicated over observable instances of legal reasoning it would
simply fail to be a theory of legal reasoning as we understand the notion.
74. Cf. T. BENDITr, supra note 52, at 10-11 (distinguishing rule nihilism); T. MORAWETZ, THE
PHILOSOPHY OF LAW 75-81 (1980) (discussing rule nihilism).
75. Similarly, one could distinguish rule scepticism - the- claim that there are good reasons to
doubt whether anything like a rule enters into judicial justification. See H.L.A. HART, THE CONCEPT
OF LAW 132-37 (1961) (discussing rule scepticism).
76. Accord A. J. AYER, supra note 33, at 36-41 (experience cannot justify epistemological
scepticism).
77. To demonstrate rigorously that rules are incoherent would require something like an impossi-
bility proof: a formal demonstration that various proposed conditions for the existence of a legal rule
cannot be satisfied simultaneously. For example, Arrow's startling Impossibility Theorem for social
1985] PRACTICAL REASONING AND JUSTIFICATION 61

The point here is that nihilism of this genre goes awry at least in
part because it fails to integrate its conceptual and normative commit-
ments with its descriptive basis: the claim that the idea of a legal rule
is nonsense cannot be squared with the appeal to evidence about the
role a rule does or does not actually play. The nihilist, in sum, would
be more charitably understood as asserting a far less aggressive thesis:
generalizing from judges' observable use, rules do not have certain at-
tributes they are thought to have.78
That an adequate theory of judicial justification must integrate
descriptive, conceptual and normative commitments means that we
need sophisticated and sensitive criteria for evaluating a theory's ac-
ceptability. At a minimum, of course, a theory of justification must be
descriptively adequate. The structures of inference and the standards
of validity advanced by the theory must be such that instances of those
structures and norms would be recognized by the legal community as
actual justifications.
Descriptive adequacy alone is not enough, however, for a theory
to be acceptable. Philosophers of science have noted that theories of
any sort are underdetermined with respect to their data.7 9 That is, for
any set of data - no matter how rich and consistent with some partic-
ular theory - there are indeterminately many theories which are con-
sistent with the same phenomena.8 0 Since any theory could be revised
to accommodate discordant data, the fact that a theory is consistent

decision making begins with a set of plausible requirements for any method of making collective choices
- requirements that seem intuitively to lie at the heart of our appreciation of democratic theory - and
then demonstrates that no rule for collective choice can satisfy all those requirements. See K. ARROW,
SOCIAL CHOICE AND INDIVIDUAL VALUES (1951). For a useful explanation and discussion of Arrow's
Theorem, see A. FELDMAN, WELFARE ECONOMICS AND SOCIAL CHOICE THEORY 178-95 (1980). To
my knowledge, none of the critics of legal rules has attempted anything like an impossibility proof.
78. A grudging recognition of this point can be observed in recent writings of the Critical Legal
Studies movement. Some writers in this tradition have questioned the intelligibility of coherence of
legal reasoning based on rules. See, e.g., Tushnet, Post-Realist Legal Scholarship, 1980 Wis. L. REV.
1383, 1384-86. More recent CLS scholarship has focused on the project of what members of the move-
ment call an "internal" critique of the vision of law and legal reasoning that they attribute to liberal
political theory. That is, they attempt to identify certain central tenets of liberal theory as regards law
and then show that all these tenets cannot be achieved together. See, e.g., Singer, The Player and The
Cards: Nihilism and Legal Theory, 94 YALE L.J. 1, 10-14 (1984). The internal critique is limited,
therefore, in that it does not attempt to dispute the notion of rules vel non, but only alms at showing
that liberal theory cannot meet its own criteria for success. See id.; Dalton, An Essay in the Deconstruc-
tion of Contract Doctrine, 94 YALE L.J. 997, 1006 (1985): "The critique seeks to demonstrate how the
various guidelines proposed by liberal legalism lack the clarity on which the liberal order presents itself
as depending."
79. The point about the underdetermination of theories is most commonly attributed to Pierre
Duhem. See P. DUHEM, THE AIM AND STRUCTURE OF PHYSICAL THEORY (1975). For a discussion
of underdetermination and Duhem's contributions, see C. GLYMOUR, THEORY AND EVIDENCE 30-31
(1980).
80. C. GLYMOUR, supra note 79, at 31.
62 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

with our observations does not compel that theory's adoption. 8'
Rather, the preferable theory is one that can explain and predict the
relevant phenomena more simply, more powerfully, and more ele-
gantly. Just how these attributes of theories lead to our accepting one
candidate and rejecting its rivals is beyond the scope of this article, but
82
some general guidelines for theory adoption can be noted.
First, between two theories of judicial justification as to which we
otherwise find no important distinction, we should prefer the simpler.
A theory's simplicity consists in its capacity to explain the relevant
data with as few theoretical primitives as possible.83 The theory which
can explain our judgments about the validity of arguments with fewer
logical structures, fewer distinct kinds of inferences, and fewer criteria
for validity is to be preferred over a rival which proliferates its theoret-
ical commitments.
Next, other things being equal, a theory of judicial justification
should be powerful. This involves two things. First, where we con-
front judicial arguments about whose validity we are uncertain, a pow-
erful theory will provide us with a basis on which to judge them valid
or invalid. This means that the theory must advance criteria of valid-
ity which support evaluations of controversial as well as un-
problematic instances of judicial reasoning. Second, a powerful theory
of judicial justification should help us illuminate other issues of legal
philosophy. We understand that justifying a legal decision involves
several important facets of the judicial process and, hence, of the law.
A better theory of justification will lead us to further insights about the
legal system.8 4
In addition to simplicity and power, we should favor theories
which are elegant. As applied to theories of judicial justification, ele-
gance bears on how well the theory of justification fits our other juris-
prudential commitments. Where we have well settled views about
other facets of the judicial process, the better theory's conceptual and
normative commitments are generally compatible with those well set-
tled views. We would avoid, for example, a theory that conflicted rad-
ically with our unproblematic assumptions about stare decisis and

81. See generally T. KUHN, THE STRUCTURE OF SCIENTIFIC REVOLUTIONS (1962). For a de-
tailed account of how the possibility of theory modification bears on the choice of physical theories, see
L. SKLAR, SPACE, TIME, AND SPACETIME 242-72 (1974).
82. These guidelines are drawn mostly from C. HEMPEL, PHILOSOPHY OF NATURAL SCIENCE 33-
46 (1966). Cf. C. GLYMOUR, supra note 79, at 110-75 (criteria for comparing theories); see W. QUINE
& J. ULLIAN, THE WEB OF BELIEF 42-53 (1970) (virtues of hypotheses).
83. This expresses the core idea behind the notion of theoretical parsimony, commonly associated
with the work of William of Ockham and his famous razor. See Moody, William of Ockham, in 8 THE
ENCYCLOPEDIA OF PHILOSOPHY 306, 307 (P. Edwards ed. 1967).
84. Accord C. GLYMOUR, supra note 79, at 110-75.
1985] PRACTICAL REASONING AND JUSTIFICATION 63

prefer, instead, the theory whose picture of justification accords with

those assumptions.8 5
Further, elegance requires that, where our views about the law
and legal decisions are not well settled, we should avoid the aggressive
theory. If there are rival conceptions about the nature of law and of
legal systems, then, other things being equal, we should prefer the the-
ory that is compatible with both conceptions. Adopting a theory of
judicial justification should not, by itself, foreclose our pursuit of a
viable claim about a related topic in legal theory. A theory should
illuminate other jurisprudential issues, not dispose of them
surreptitiously.86
The proposals which will be examined in Sections III and IV ex-
emplify these qualities to some extent. I will argue that a theory of
judicial justification in terms of practical reasoning is preferable in the
respects just reviewed to a theory in terms of deduction or analogy. It
is important to note that in one sense each of these theories is plainly
normative. They are theories of justification: of successful arguments
which validate their conclusions. It follows that they offer criteria of
argumentation under which some inferences will be found unaccept-
able. While the respective theories all invoke norms for the evaluation
ofjustificatory arguments, they are not prescriptive in the sense that is
normally attached to that term. That is, they do not contend that
judicial justification should differ significantly from existing legal prac-
tice. They propose no new ideal for justification - no new warrant
for deciding cases. They do not aim to reform existing practice, except
to the extent that they provide a basis on which unacceptable argu-
ments can be distinguished and criticized. They are theories of judi-
cial justification: they attempt to explain our commonly shared
evaluations of how judges do actually justify their decisions.8 7 In the
next two sections, it will be demonstrated that they are not equally
successful in meeting this aim.

III. Two DOMINANT VIEWS OF JUDICIAL JUSTIFICATION

In the literature of jurisprudence two approaches to the problem

of legal reasoning have been dominant: one view which regards legal

85. Accord W. QUINE & J. ULLIAN, supra note 82.

86. Accord C. GLYMOUR, supra note 79, at 110-75.
87. Cf Ryan, Introduction, THE PHILOSOPHY OF SOCIAL EXPLANATION I (A. Ryan ed. 1973):
"[lI]t would be difficult to understand as a philosophy of science one which did not recognize as scientific
achievements at least most of what scientists and laymen accept as such .. . . [Similarly], [t]he moral
philosopher wishes neither to legitimate every moral position whatever nor to give an account of moral-
ity which makes what we ordinarily take to be moral positions simply unrecognizable as such." (Em-
phasis in original.)
64 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

reasoning as a species of reasoning by analogy (the "analogical the-

sis"), and another which regards it as a species of deduction (the "de-
ductive thesis"). While other approaches have been proposed,8 8 these
two approaches have seen the greatest development. In this section, I
examine these conceptions of legal reasoning in light of the conclu-
sions reached in the previous section about the nature and function of
a theory of justification. I present these views, beginning with the de-
ductive thesis, and evaluate them in light of the criteria articulated in
the previous section.

A. The Deductive Thesis

1. Deduction and Judicial Justification
The deductive thesis contends that deduction provides the appro-
priate model for legal reasoning in general and judicial justification in
particular. As it has been developed for judicial justification, the de-
ductive thesis centers around two claims. First, the thesis claims that
judicial argumentation involves rule-like general propositions that sub-
sume a class of litigants or factual situations and specify legal conse-
quences for the members of the class.89 Second, and most
distinctively, the deductive thesis contends that on at least some occa-
sions it is a valid legal inference to deduce a legal conclusion from
premises consisting of a rule and the claim that a specific case is of the
sort covered by the rule.9 °
In varying guises, this approach has enjoyed recurring but contro-
versial favor in the jurisprudential literature.9" An unsophisticated
version of this thesis is frequently caricatured as the dreaded "mechan-
ical jurisprudence." Whether or not anyone has actually advanced
mechanical jurisprudence as a viable theory of legal reasoning,92 it has
been a frequent whipping boy. Holmes initiated the modern abuse,
pronouncing that the life of the law has been not logic, but experi-
88. See, e.g., T. PERRY, MORAL REASONING AND TRUTH, 75-92, 196-215 (1976); J. HOROVITZ,
LAW AND LOGIC: A CRITICAL ACCOUNT OF LEGAL ARGUMENT (1972). See also infra note 149 (dis-
cussing deontic logic).
89. N. MACCORMICK, supra note 8, at 18-32, 73-86.
90. Id. at 53-62.
91. In favor of the deductive theory: N. MACCORMICK, supra note 8; Guest, The Logic in Law, in
OXFORD ESSAYS IN JURISPRUDENCE (Ist ser. 1961); Hart, Philosophy of Law, in 6 THE ENCYCLOPE-
DIA OF PHILOSOPHY (P. Edwards ed. 1967); Comment, supra note 8.
Against deduction: Lloyd, Reason and Logic in The Common Law, 64 LAW Q. REV. 468 (1948);
G. GOTrLIEB, THE LOGIC OF CHOICE (1968); Burton, supra note 9; Wilson, The Nature of Legal
Reasoning.-a commentary with specialreference to ProfessorMacCormick's Theory, 2 LEGAL STUD. 269
(1982).
92. "So far [nominalists] have had little luck in caging and exhibiting mechanical jurisprudents
(all specimens captured - even Blackstone and Joseph Beale - have had to be released after careful
reading of their texts)." R. DWORKIN, supra note 17, at 15-16.
1985] PRACTICAL REASONING AND JUSTIFICATION 65

ence, 93 and Pound added the derogatory label "mechanical." 94

Since these worthies set the tone, mechanical jurisprudence has
been pilloried for a variety of indiscretions." Some of the criticisms of
deduction charge, in effect, that it does not reflect the actual cognitive
processes of the deciding court.96 Whatever the truth of this charge, it
is largely irrelevant for developing a theory of judicial justification;
justification, as I have noted, is different from the psychological
processes of the judge.97 Deduction has also been scorned as inade-
quate to explain the myriad ways in which judges use legal rules to
rationalize their decisions. 98 And, it has been criticized as incapable of
accounting for the reasoning advanced by judges to justify their con-
clusions in novel situations, when they must justify their choices
among rules or formulate the rules they use. 99
Despite the contumely heaped upon mechanical jurisprudence,
the deductive thesis has generated adherents. Two theorists, in partic-
ular, have carried its banner. Richard Wasserstrom, in his pioneering
and insightful work, The JudicialDecision, argued forcefully that the
critics of mechanical jurisprudence had failed to disprove the deduc-
tive thesis, understood more broadly." ° Neil MacCormick, in his re-
cent monograph, Legal Reasoning and Legal Theory, has expanded
Wasserstrom's picture to accommodate our sense of "hard cases."1' '
Together, they have advanced two lines of argument in favor of the
deductive thesis.
First, by invoking more subtle conceptions of rules, they have
blunted the caricature of mechanical jurisprudence - the blind and
inflexible application of simple rules to complex situations.' 0 2 Rules
can be of different sorts and play different roles in judicial justifica-
tion. 10 3 It is not, for example, essential to the nature of rules that it be
easy to ascertain whether some particular rule is significant in the legal
system. Nor is it crucial that it be a straightforward matter to state
the rule's correct formulation. Scientific laws, for example, are fre-
quently hard to decipher, and ascertaining the correct physical law

93. 0. W. HOLMES, THE COMMON LAW 1 (1881).

94. Pound, Mechanical Jurisprudence, 8 COLUM. L. REV. 605 (1908).
95. For a thoughtful recent discussion of the significance and failings of any mechanical concep-
tion of judging, see Gordley, Legal Reasoning: An Introduction, 72 CALIF. L. REV. 138, 140-46 (1984).
96. See, e.g., J. FRANK, supra note 3, at 31-34.
97. See supra text accompanying notes 42-43.
98. See Burton, supra note 9; G. GOT-rLIEB, supra note 91.
99. See supra note 98.
100. R. WASSERSTROM, supra note 42.
101. N. MACCORMICK, supra note 8, at 195-221.
102. See R. WASSERSTROM, supra note 42, at 138-72; N. MACCORMICK, supra note 8, at 66-72.
103. See T. BENDITr, supra note 52, at 30-41.
66 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

may well require creativity and imagination. 0 4 But these difficulties

do not, by themselves, convince us that scientific laws do not obtain or
that they mask the underlying processes. It may well be that the scope
of a given rule is uncertain.10 5 Least of all need it be simple to predict
the result of the rule's application. 0 6 What is crucial for the deductive
thesis is the claim that a rule, when applicable, counts as some reason
for a decision. This claim is consistent with a variety of different con-
ceptions of rules. Given a different conception of the possible roles of
rules in judicial justification than was assumed (and then attacked) by
the Realists, many of the horrors of "mechanical jurisprudence" sim-
ply do not bear on the adequacy of the deductive thesis.' 0 7
Second, the deductive thesis need not deny that other, non-deduc-
tive forms of reasoning sometimes figure in the judicial decision. Was-
serstrom coined the notion of a "two-level procedure of
justification."' '° At one level the inference of a legal conclusion from
a legal rule may be deductive in nature. But at another level non-
deductive reasoning may be used to warrant the rule or to define its
scope. MacCormick has elaborated on this notion, propounding two
non-deductive criteria of inference - consistency and coherence - to
account for inferences at the non-deductive level.' 09 What is essential
to the deductive thesis, its adherents have argued, is the claim that it is
at least a justified form of argument to deduce a legal decision from a
rule and the factual specification of a particular case. "1 Legal conclu-
sions inferred from such premises by a valid deductive inference are
warranted if the rule was appropriate and the factual characterization
correct.
The case for the deductive thesis goes beyond merely refuting the
objections raised against it. Its proponents have laid the foundation
for the claim that it provides an adequate theory of judicial justifica-
tion. In particular, proponents have taken great pains to establish the
view's descriptive adequacy. Acknowledging that other forms of rea-
soning may also be employed by lawyers and jurists, advocates of the
deductive thesis contend that at least some opinions are, in fact, de-
ductive in form. To prove their point they have rewritten actual opin-

104. See generally T. KUHN, THE STRUCTURE OF SCIENTIFIC REVOLUTIONS (1962).

105. Rules may be formulated in terms of vague or open-textured words. See H.L.A. HART,
supra note 75, at 121-32.
106. See T. BENDirr, supra note 52, at 31-34.
107. Id. at 30-41.
108. R. WASSERSTROM, supra note 42, at 138-72.
109. N. MACCORMICK, supra note 8, at 100-28.
110. Id. at 52.
1985] PRACTICAL REASONING AND JUSTIFICATION 67

ions as sequences of deductively valid arguments."' This, they argue,

demonstrates the "possibility" of a purely deductive justification. 1 2
Beyond its descriptive accuracy, deduction's normative ramifica-
tions are attractive. To the extent that legal arguments either take the
form of deductive arguments or can cogently be rewritten to take that
form, and to the extent that such arguments are unproblematically
regarded as having supported their conclusions, deductive inference
seems to constitute one form of justification for decisions. The hall-
mark of deduction is its formal criterion for the validity of a given
argument form.' 13 An argument is deductively valid if and only if
there is no possibility, in an argument of that same form, of inferring a
false conclusion from true premises.114 Deductively valid arguments
are truth-preserving arguments, and truth-preservingness is a ready
and reliable test for the acceptability of inferences." 5 If truth-preserv-
ingness were also a standard by which lawyers and jurists could evalu-
ate the justification of a given decision, then the deductive thesis would
be normatively useful as well as descriptively adequate.
Furthermore, the proposal that legal reasoning involves deductive
inferences is appealing on conceptual grounds. A common feature of a
wide range of theories of law has been the focus on legal rules." 6 The
deductive thesis conforms to this common expectation of judging. The
"major" premise of the form of deductive inference with which we are
most familiar' resembles those abstract propositions that we associ-
ate with legal justification: "All men are mortal"; "A contract not to
be performed within a year's time of its making is unenforceable unless ' 18
it is evidenced by a writing, signed by the party to be charged." "
Seen along these lines, a legal rule is construed as a general proposition
that identifies a class of actual situations and prescribes certain legal
consequences for cases of that class. If the case at hand is subsumed
under a particular rule, and if the rule prescribes particular conse-

11. See, e.g., N. MACCORMICK, supra note 8, at 19-34. But see Wilson, supra note 91 (challeng-
ing MacCormick's rewriting).
112. N. MACCORMICK, supra note 8, at 19.
113. Id. at 21. For a general discussion of the nature of deductive validity, see W. SALMON,
LOGIC (2d ed. 1973); 1. Copi, INTRODUCTION TO LOGIC (4th ed. 1972).
114. N. MACCORMICK, supra note 8, at 21.
115. There are a variety of procedures for testing the truth-preservingness of an argument: truth
tables, see, e.g., I. Copi, supra note 113; semantic trees, see, e.g., R. JEFFREY, FORMAL LOGIC, ITS
SCOPE AND LIMITS (1967); and axiomatic methods, see, e.g., B. MATES, ELEMENTARY LOGIC (1965).
116. See supra text accompanying notes 50-62.
117. As an example of this form of inference, consider the classic syllogism: "All men are mortal.
Socrates is a man. Therefore, Socrates is mortal." The first sentence, asserting the universal mortality
of humankind, is commonly referred to as the major premise, and the second sentence, stating Socrates'
humanity, the minor premise.
118. See N. MACCORMICK, supra note 8, at 72-86.
68 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

quences for subsumed cases, then it follows in a deductively valid in-

ference that the prescribed consequences are true of the given case.
Along these same lines, developments in logic appear useful in
relating the issues of legal reasoning to other jurisprudencial concerns.
It has already been noted that judges are expected to treat like cases
alike, and that to meet this expectation they are bound to frame their
justifications in terms of universal statements of law." 9 The ability of
deductive logic to capture valid inferences involving propositions of
universal scope suggests the appropriateness of deductive logic as the
fundamental form of legal inferences. 20
Deduction looks like a particularly fruitful model for judicial rea-
soning because it offers a rich and already well-developed theory of
argument.' 2 ' It is a reliable and powerful form of reasoning. In many
contexts valid deductive inferences are regarded as the paradigm form
of justification.122 Given our general respect for deduction as a justify-
ing form of argument, legal conclusions would seem to deserve a simi-
lar respect when they are warranted by deductive logic. Moreover,
logicians have constructed an elaborate formal apparatus in which for-
mal conclusions about deduction itself can be proven in the most rig-
orous fashion. 2 ' Powerful metatheorems - theorems about the
capability of various formal systems of deductive logic - have been
established using this formal theory of deduction. 24 Deductive logic,
in short, is an attractive foundation for a theory of legal reasoning
because it appears intellectually elegant and well-established.

2. Evaluating a Deductive Theory of Judicial Justification

Despite the apparent virtues of deduction it is my contention that
any deductive theory of judicial justification suffers from a number of
conceptual and normative maladies. These infirmities result from the
deductive theory's commitment to the truth-preserving criterion of va-
lidity and its supposition that the statements of a judicial argument

119. See supra text accompanying notes 44-48.

120. See I. TAMMELO, MODERN LOGIC IN THE SERVICE OF LAW 79-93 (1978). Logicians
standardly distinguish between propositional and quantified logic. Use, in the latter logic, of a quanti-
fier provides a simple mechanism for representing in symbolic form the difference between singular and
universal propositions. See, e.g., W. SALMON, supra note 113, at 74-79. At least one commentator has
criticized MacCormick for failing to appreciate the significance of quantified logic and its relation to
our expectation that like cases be treated alike. See White, Philosophy and Law: Some Observations on
MacCormick's LEGAL REASONING AND LEGAL THEORY, 78 MICH. L. REV. 737, 742 (1979).
121. See, e.g., I. TAMMELO, supra note 120, at 7-70.
122. See, e.g., C. HEMPEL, supra note 82, at 3-18; W. SALMON, supra note 113, at 13-17.
123. For an elaboration of the formal theory of deductive propositional and quantified logic, see
G. HUNTER, METALOGIC (1971).
124. Among them: consistency, completeness, and decidability. See id. at 78-79 (consistency), 92-
96 (completeness), 118-20 (decidability).
1985] PRACTICAL REASONING AND JUSTIFICATION 69

have a certain logical form. Construing judicial decisions so that the

decisions could properly be said to be deductively valid attributes to
the judicial role various features which conflict with important and
well-settled expectations of judging and judicial argument, 25 or else
requires untenably2 6aggressive assumptions about the nature of law and
the legal system. 1

Deductive Validity
If an inference is deductively valid then the conclusion follows
from the premises in a truth-preserving inference form. Suppose that
a case is correctly subsumed under a particular legal rule which for a
given jurisdiction accurately states the legal consequences for cases of
that class. It follows that the prescribed legal consequences are true of
the subsumed case. Whether or not we wish to apply that particular
rule to that case, if the rule subsumes the case then the prescribed
consequences follow.' 2 7
In deduction, the truth-preserving nature of the inference form is
the only warrant for drawing the conclusion. Moreover, it is logically
sufficient. Any applicable deductively valid inference is ipsofacto war-
ranted. 128 And, since there is no other warrant for the conclusion
there is no other basis on which to challenge the argument's
validity. 129
The situation is different in the law. We may distinguish between
the applicability of a legal rule, on the one hand, and the warranted
application of the rule on the other. 3 0 For example, suppose there is a

125. Therefore, a deductive theory's commitment to the deductive criterion of validity conflicts
with the third of my criteria for choosing among rival theories articulated in Section II. See supra text
accompanying notes 83-86.
126. The commitment to truth-preservingness, therefore, runs afoul of the elegance criterion. See
supra text accompanying notes 85-86.
127. Philosophers and logicians commonly express this feature by saying that the premises of a
valid deductive argument necessitate the conclusion. See, e.g., W. SALMON, supra note 113, at 15. Of
course, the validity of an argument does not necessitate the truth of its premises or conclusion. Philoso-
phers therefore distinguish the validity of an argument from the argument's soundness: a sound argu-
ment is a valid argument with true premises (and, therefore, a true conclusion). See, e.g., I. Con, supra
note 113, at 33.
128. W. SALMON, supra note 113, at 15.
129. Cf. id. at 82:
Given a valid deductive argument, we may add as many premises as we wish without de-
stroying its validity. This fact is obvious. The original argument is such that, if its premises
are true, its conclusion must be true; this characteristic remains no matter how many prem-
ises are added as long as the original premises are not taken away. By contrast, the degree
of support of the conclusion by the premises of an inductive argument can be decreased by
additional evidence in the form of additional premises.
130. The phrasing of this distinction is Kenneth Winston's, from his discussion of arguments by
H.L.A. Hart and Chaim Perelman that the generality of legal rules leads to a "minimum" fairness of
the legal system. See Winston, supra note 44, at 17-18, 22-23. Winston points out that a rule's descrip-
70 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

legal controversy that appears to be subsumed under conflicting rules

of law. That is, suppose that the case at hand is simultaneously a
member of two classes governed by different rules, but that the rules
prescribe mutually incompatible legal consequences for cases they sub-
sume.1 3 ' By hypothesis, each rule applies to the case at hand, and to
warrant his decision the judge must justify the application of one rule
and the rejection of the other. On a deductive theory of judicial justifi-

tion of the class of cases which are subsumed by the rule is "necessarily incomplete" - i.e. that the
description does not preclude a case's being simultaneously described as a case of another sort. Id. at
16. He then argues that the generality of a rule does not entail that the rule's consistent application will
be fair:
If the descriptions of cases embodied in rules of law are necessarily incomplete, then there is
no guarantee that the application of a particular rule to a case will take account of all the
features of the case relevant to determining fairness of the application. In different words,
the description in the rule is sufficient for determining the applicability,but not the war-
ranted application, of the rule to the case. Cases always overflow the boundaries within
which rules attempt to confine them. As a consequence, the identification of a case as a
member of a class (i.e. as subsumable under a rule) does not commit one, on moral grounds,
to the same identification for any other case satisfying an identical description.
Id. at 17-18. Cf. also Note, Understandingthe Model of Rules: Toward a Reconciliationof Dworkin and
Positivism, 81 YALE L.J. 912, 917-21 (1972) (distinguishing a decision that a rule is applicable from a
decision that the rule should be applied).
131. Some terminological precision is required on this point. As I will use the term, a conflict
obtains between rules if the rules subsume the same case and prescribe for that case incompatible re-
sults. For example, the Statute of Frauds requires a writing if certain kinds of contracts are to be
enforced. Among the kinds of contracts "within" the Statute are those not to be performed within a
year's time from their making, and those involving an interest in land. Courts early on developed a part
performance exception in the latter category. Part performance by one party could make the contract
enforceable notwithstanding the fact that there was no writing to evidence the contract. Yet, courts
steadfastly resisted a part performance exception as regards contracts not to be performed within a
year. See, e.g., Morsinkhoff v. DeLuxe Laundry & Dry Cleaning Co., 344 S.W.2d 639 (Mo. App.
1961). Suppose a particular contract falls under both parts of the Statute, and one of the parties has
rendered part performance. Should the court enforce the contract? Judicial doctrines would be said to
conflict on the question of enforcing that contract. For an exposition of the part performance doctrine
and the conflicting judicial approaches to the exception, see 3A CORBIN ON CONTRACTS § 459 (1960).
Grant Gilmore has described a similar situation posed by the RESTATEMENT OF CONTRACT'S twin
bases for the enforcement of promises - the Holmsian "bargained-for" consideration, articulated in
§ 75, and the emerging promissory estoppel which was embodied in § 90. See G. GILMORE, THE
DEATH OF CONTRAT 60-72 (1974). "The extent to which the new § 90 was to be allowed to undercut
the underlying principle of § 75 was left entirely unresolved." Id. at 64. Promissory estoppel and
Holmsian consideration conflict, on this usage, as regards a particular contract where promissory estop-
pel would enforce the promise and bargained-for consideration would deny enforcement.
However, there will be myriad instances where no conflict results from these two rules: where, for
example, there is neither consideration nor detrimental reliance. Since in such a situation it is possible
to satisfy both rules simultaneously, it is misleading and inappropriate to say that the rules are inconsis-
tent. Rather, rules are inconsistent if they conflict on every case. See Marcus, Moral Dilemmas and
Consistency, 73 J. PHIL. 121, 128-31 (1980). Afortiori, it is a mistake to assert that the rules contradict
one another. Since it is common for lawyers and legal academics to refer to situations of conflicting
legal rules as situations of legal "inconsistency," I will say that applying conflicting rules may lead to
inconsistent results, and that the legal system is, in a derivative sense, inconsistent if it is possible to
generate incompatible results by applying valid but conflicting rules. For further analysis of the nature
of legal conflicts and their possible significance for legal theory, see Munzer, Validity and Legal Con-
flicts, 82 YALE L.J. 1140 (1973).
1985] PRACTICAL REASONING AND JUSTIFICATION 71

cation two conflicting legal conclusions would each be warranted.

However, the judge must decide the case. Although both rules apply,
he must decide which conclusion to impose and, implicitly, which rule
is appropriate.
Proponents of the deductive thesis could respond to this challenge
by trying to refine their picture of judicial justification so as to account
for the distinction between an applicable rule and a rule whose appli-
cability is warranted. 132 More specifically, proponents of the "two-
level" version of the thesis could argue that at the non-deductive level
of justification the judge can evaluate the various rules that seem to be
applicable and conclude which rule's application would be warranted.
MacCormick, in particular, has argued that justification at times in-
volves a different form of argument - what he labels "second-order
justification" - which establishes the preferability in the legal system
of one possible rule over another. 13 3 Proponents of the two-level ver-
sion could try to extend MacCormick's idea of second-order justifica-
tion. The court could, on this argument, rely on second-order
justifications to determine which rule's application would be
warranted.
This is, no doubt, an appealing argument and, if it could be sus-
tained, would seem to accommodate the distinction between a rule's
applicability and its warranted application while preserving the es-
sence of the deductive theory. The argument's thrust is that however
the judge derives the appropriate rule, applying it to the given case
involves a deductive inference. Some reflection shows that this re-
sponse is ultimately unsatisfying. In deduction, the conclusion of a
34
truth-preserving inference follows, whether or not we approve of it.,
So, regardless of the judge's second-order ruminations about which
rule is appropriate to the case at hand, if both rules are applicable and

132. It is a consequence of the underdetermination of theories of judicial justification that a de-

ductive theorist could revise his theory in a way to accommodate this apparent difficulty for the theory,
see supra notes 79-82 and accompanying text, either by adding "auxiliary hypotheses" to his theory or
by redescribing the distinction between a rule's applicability and the rule's warranted application. Mac-
Cormick, at least, takes the latter tack, arguing that cases where more than one rule seems to apply are
actually cases where further refinement of the competing rules is required. See N. MACCORMICK, supra
note 8, at 101-19.
The moral of Section II, supra, is that theoretical emendations, such as MacCormick's, are possi-
ble. However, they do not necessarily keep the revised theory above reproach. MacCormick's argu-
ment, in particular, avoids conflict only by supposing that his criteria for choice among competing rules
- consistency and coherence - will yield a unique answer in every case of competing rules. Else, the
situation would still obtain where deducing the consequences of each rule would be warranted. There-
fore, his revision succeeds only at the cost of overly aggressive theoretical commitments. Cf. Winston,
supra note 44 (arguing that it is implausible that only warranted rules are applicable).
133. N. MACCORMICK, supra note 8, at 100-28.
134. See supra text accompanying notes 127-29.
72 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

the consequences of each follow in a deductively valid inference, then

both consequences are warranted. In short, even for the two-level ver-
sions of the deductive theory, the deductive theorist must either toler-
ate the existence of inconsistencies in the legal system, or else posit
that no such inconsistencies obtain. 13 5 It can be seen that neither horn
of the dilemma is acceptable to the deductive theorist.
Consider, first, the problem that inconsistency poses for a deduc-
tive theory. In formal logic, a system of logic is inconsistent when a
36
proposition p and its negation -p can both be proven in the system. 1
Inconsistency would be a deeply troubling result for a deductive sys-
tem; from an inconsistent premise set any conclusion whatsoever can
be justified. 137 If inconsistency were tolerated, a deductive system
would be an utterly unreliable means of justifying claims.138 For that
reason, it is crucial in developing a formal system of deductive logic to
establish that using valid inferences in that system does not lead to
39
inconsistent conclusions. 1
While the deductive proponent cannot accept inconsistency, he
cannot preclude its existence in the legal system.140 Indeed, our expec-
tations about the role of judges in the legal system suggest that com-
plete consistency throughout the system is implausible. The
consistency of a deductive system of logic implies that we may rely on
indirect or reductio ad absurdum arguments. 4 ' Yet, reductio argu-
ments are unreliable in the law. When a judge decides a particular
case he might be taken to have denied the validity of one party's argu-
ments, but he is not necessarily understood to have decided that any
other legal propositions are false. Consider, for example, the situation
where there are two applicable but conflicting legal rules,142 but where

135. See supra note 131 for an explanation of this notion of inconsistency in the legal system.
136. See G. HUNTER, supra note 123, at 78-79.
137. Suppose both p and -p. From p, one may validly deduce p or q, for any q whatsoever, by
the rule of inference known as addition. But, from p or q, and -p, one may validly deduce q. So, q can
be justified in deduction, regardless of its content. Id.
138. Cf also J. POLLOCK, KNOWLEDGE AND JUSTIFICATION 39-46 (1974) (consistency as a re-
quirement for epistimological justification).
139. See, e.g., G. HUNTER, supra note 123, at 79-83.
140. While consistency can be proven formally as a metatheorem about a system of logic, there is
no reason to suppose that it could be proven of a legal system.
141. See, e.g., P. SUPPES, INTRODUCTION TO LOGIC, 36-41 (1957). Reductio arguments take the
following form. Given premises which we know are true, if a contradiction is validly deducible from
those premises in conjunction with some additional premise or premises, then it follows that the addi-
tional premise (or one of the additional premises) is false. Put differently, affirming the truth of the
given premises entails the denial of the added (false) proposition. There is no reason to suppose that
consistency, or any comparable metatheorem, could be proven formally of the set of legally valid infer-
ences. Nor is there any good reason independent of the metatheory, to suppose that such a set is
consistent.
142. See supra note 131.
1985] PRACTICAL REASONING AND JUSTIFICATION 73

the second is neither argued by any of the parties nor considered by

the bench. The judge's use of the first does not entail the falsity of the
second. Indeed, his decision does not even mean that the second rule
was inapplicable.
We distinguish in our legal system between the holding of a case
and its dicta, and the two are valued quite differently in our system of
precedent. This distinction means that even when a given holding ap-
pears entirely inconsistent with some legal rule not at issue in the deci-
sion, the holding, no matter how valid, does not entail the falsity of the
apparently inconsistent rule. Of course, the deciding judge might at-
tempt a more aggressive characterization of his decision - suggesting,
perhaps, that his holding meant that the inconsistent rule was false.
But his suggestions could be understood as dicta only. In any subse-
quent case in which the inconsistency of the first decision with the
other rule itself became an issue, the judge deciding the second case
would have to evaluate the problematic rule and reach his own deci-
sion about which authority to apply and which to reject.
It is, finally, an embarrassment to the deductive theorist that he
cannot provide an affirmative argument in support of the legal sys-
tem's consistency. The system's inconsistency undercuts the reliability
of deductive justification, and given consistency's significance it would
be heroic, if not quixotic, simply to assume that no inconsistencies in
fact obtain. The aggressiveness of such an assumption must surely
count against the deductive theory's plausibility. Moreover, many ju-
risdictions are rife with instances of apparent inconsistency; the deduc-
tive theorist would have to suppose that each of these apparent
inconsistencies could be explained away as apparent only.
At most, a deductive proponent could argue that judges and law-
yers work to rid the legal system of its inconsistencies. MacCormick,
for one, posits consistency of the legal system as one of the norms of
second-order justification. 4 3 But, it is hard to see how the proponent
could demonstrate that consistency is an unequivocal norm of the
legal system, and hence, that inconsistency would inevitably be

143. N. MACCORMICK, supra note 8, at 106. It seems entirely plausible that judges do indeed
work to rid the legal system of inconsistencies. So, for example, while the final chapters have yet to be
written for the interplay of bargained-for consideration and promissory estoppel, see supra note 131, it
now appears the latter is ascendant. Indeed, the detrimental reliance idea seems so powerful that in
many jurisdictions it has prevailed over the Statute of Frauds itself. See, e.g., McIntosh v. Murphy, 52
Hawaii 29, 469 P.2d 177 (1970); RESTATEMENT (SECOND) OF CONTRACTS § 139 (1982).
It also is beyond dispute that, when confronted with inconsistencies, courts frequently reinterpret
or reformulate one or both of the conflicting rules so as to eliminate the grounds for inconsistency.
What is at issue, however, is whether every instance of incompatible rules can cogently be supposed to
be just an instance of incomplete analysis or infelicitous phrasing on the part of past jurists.
74 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

weeded out of the legal system. 144 Even if he could advance a norm of
consistency, that norm does not preclude the existence of current in-
consistencies in the system, and those existing inconsistencies indicate
deduction's current unreliability.

Logical Form

The deductive thesis' reliance on the truth-preserving criterion of

validity commits the theory's proponents to another unsatisfactory po-
sition regarding what philosophers call the "logical form" of legal
statements: the kind of sentence a legal statement should be construed
to be, and the proper logical parts of a legal statement. 45 Specifying a
statement's logical form will indicate how that statement can validly
figure in arguments of various sorts, and asserting that a particular
statement figures validly in some particular argument implicitly com-
mits one to claims about that statement's logical form. 146 The deduc-
tive criterion of validity - truth-preservingness - can apply only to

144. MacCormick, for example, simply asserts consistency is a norm, and, apparently, an une-
quivocal one:
The idea of a 'consistent' body of norms I use in a strict sense: however desirable on conse-
quentialist grounds a given ruling might be, it may not be adopted if it is contradictory of
some valid and binding rule of the system. Of course, an ostensibly contradictory precedent
may be 'explained' and 'distinguished' to avoid such a contradiction. . . . But if such de-
vices for reconciliation fail, the requirement of consistency would require rejection of an
otherwise attractive ruling on the ground of its irresoluble conflict with (contradiction of)
established valid rules.
N. MACCORMICK, supra note 8, at 106. Thus, he assumes not only that the norm of consistency must
be fulfilled, whatever else, but further that there are at present no inconsistencies among the set of valid
and binding rules of the system. Else, there would be no resolution of the situation which is considered
in the text, where two incompatible rules apply to the same case: for applying either would result in a
ruling that conflicted with some valid rule.
It is beyond this essay to establish that inconsistencies will always be with us, but support for that
idea may be found in a number of reflections on the process of common law adjudication. Cf. G.
CALABRESI, A COMMON LAW FOR THE AGE OF STATUTES 82-90, 96-109, 112-14 (1980) (arguing that
common law judges can accommodate and have, in the past, accommodated into the legal topography
statutory excursions, many of which are plausibly inconsistent with settled law); Kress, Legal Reason-
ing and Coherence Theories: Dworkin's Rights Thesis, Retroactivity and the Linear Order of Decisions, 72
CALIF. L. REV. 369, 380-83, 395-402 (1984) (arguing that in any legal theory in which the truth of a
disputed proposition is determined by its coherence with other settled claims, the set of true proposi-
tions is subject to change as new judicial decisions are made).
145. [G]rammar is supposed to tell us that "Socrates" and "The tallest man in the world"
are phrases of the same sort, while "Socrates" and "George thinks Mabel" are of different
sorts. The classification is logical in character, in the sense that prhases are classified to-
gether (in our example) if systematic substitution of one for the other does not affect the
validity of arguments. It is convenient therefore to say that characterizations of phrases
which are relevant to matters of validity are characterizations of the logical form of those
phrases.
Introduction, THE LoGic OF GRAMMAR (D. Davidson & G. Harman ed. 1975) (emphasis in original).
146. Id. at 2-3, 8-14.
1985] PRACTICAL REASONING AND JUSTIFICATION

inferences between statements which can be true or false. 47 Legal

statements, then, must be propositions,if the deductive thesis is to have
any cogency. This commitment, in turn, precludes any legal theory
which holds that legal statements are not assertions of fact but have,
instead, the form of commands or norms.
In the literature of jurisprudence, there is a long and venerable
tradition of legal theories which treat at least some of the operative
14 8
statements of a legal system as commands, directives, or norms -
the kind of statements which do not bear truth values. 4 9 Of course, it
may be true (or false) that someone has commanded an inferior to do
something or other, and it may be true that some norm is obligatory in
the legal system. But we would not standardly regard the command
"Do such-and-such" as either true or false. If neither true nor false,
then commands and norms can be neither premises nor conclusions in
arguments which look to establish the truth of some statement. 5 °
More importantly, accepting truth-preservingness as the sole cri-
terion of validity means that commands and norms cannot figure at all
in judicial argumentation. Even if one were to suppose that com-
mands could somehow figure in deductive arguments, it can be seen
that inference forms which are deductively valid may be invalid
among commands and that acceptable inferences among commands
are deductively invalid. In deduction, from the proposition p one may
validly infer p or q for any q whatsoever.' 5 ' On the definition of the

147. See, e.g., I. Copi, supra note 113, at 5.

148. See, e.g., J. AUSTIN, THE PROVINCE OF JURISPRUDENCE DETERMINED (1832) (law as com-
mands); H. KELSEN, THE PURE THEORY OF LAW (1967) (law as norms); J. RAz, PRACTICAL REASON
AND NORMS (1975) (law as norms).
149. See I. COII, supra note 113, at 5. There is some debate within the philosophical community
about whether norms can bear truth values. It is arguable that moral norms, at least, bear truth values,
inasmuch as we might wish to talk about some moral claims being true and others false.
The development of deontic logic - a logic for obligations and permissions - suggests that at
least some norms may be construed as involving truth values. Space does not permit a full discussion of
the thesis that we can understand judicial justification as a version of deontic logic. See C. ALCHUOR-
RON & E. BULYGIN, NORMATIVE SYSTEMS 65-94, 144-65; Kanger, Law and Logic, 3 THEORIA 105
(1972). Research in deontic logic is exceedingly technical and defies even a short description. See
generally B. CHELAS, MODAL LOGIC: AN INTRODUCTION 190-203 (1980). Its technical nature means
that it may be possible for a deontic thesis to explain the important features of judicial justification
which are noted in Section II, supra. However, it is important to note that most of the standard ac-
counts of deontic logic treat it as an extension of deductive logic. In particular, in the most common
elaboration of deontic logic - what is commonly termed a semantics of possible worlds - norms are
treated as valid if they are true in every possible world. Thus, in the absence of further elaboration by
proponents of the deontic thesis, we would expect that many of the objections to the deductive thesis -
the problems of the deductive criterion of validity and the difficulties with logical form - would also
afflict the deontic thesis.
150. This will be true for both deductive and inductive arguments. See, e.g., W. SALMON, supra
note 113, at 13-15.
151. See supra note 137.
76 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

connective "or" in deduction, there is no possibility of p or q being

false when p is true. Inferring the disjunction is always valid. Thus,
"The letter is posted or the letter is burned" is a valid inference in
deduction from "The letter is posted." But, from the command "Post
the letter," it would be patently unacceptable to infer "Post the letter
or burn it," no matter how valid the inference might have been among
truth-bearing propositions with the same descriptive content. No
commander who wants the letter posted would accept fulfillment of
the inferred disjunction as fulfillment of the original order. 52 On the
other hand, while inferring "Bring me a cloak" from "Bring me some-
thing to keep me warm" may be an acceptable inference among com-
mands, the corresponding inference among the related propositions
would be invalid in deductive logic.
Some interesting and historically important theories of law hold
that legal statements are commands or else are essentially like com-
mands in certain respects. Legal theorists from Austin to Kelsen, and
since, have propounded the view that legal orders and rules are com-
mands, directives, or norms, and not propositions capable of bearing
truth values. 15 3 Since valid inferences among commands or directives
are not valid in deduction, construing judicial justification as essen-
tially deductive would preclude any such command-based theory of
law. We reach the same impasse when we consider more recent theo-
ries of law that conceive of legal statements as norms, rather than as
154
propositions.

152. See A. KENNY, supra note 7, at 73. This example of Kenny's reveals, in miniature, many of
the issues pertinent to debates over logical form. The descriptive sentences "The letter is posted," and
"The letter is burned" can be represented by p and q respectively. The fiat "Post the letter" would be
symbolized by the addition of a marker of imperative mood: F(p). See infra text accompanying notes
196-99. An issue of logical form is posed by the question of how properly to symbolize and, ultimately,
how to understand the statement "Post the letter or bum it." One understanding would be represented
by the expression F(p) or F(q) - a disjunction of two imperatives. On that understanding, if commands
could bear truth values the inference from F(p) to F(p) or F(q) would be truth preserving and valid in
deduction. A more natural understanding of the sentence, however, would lead to the symbolization
F(p or q) - which might be expressed as "Let it be the case that the letter is posted or burned."
Inferring F(p or q) from F(p), however, can be seen as invalid in the logic of commands. Another
example of the same sort would be the inference from "Vote for Mayor Daly" to "Vote for Mayor Daly
or somebody." No party boss would accept the chain of reasoning.
For a variety of reasons Kenny and Hare have argued that reasoning from imperatives and norms
requires that the markers of mood should remain outside the descriptive contents, thereby preserving as
invalid what is intuitively troubling about these examples. See A. KENNY, supra note 7, at 73-75.
153. See J. AUSTIN supra note 148; H. KELSEN, supra note 148.
154. See J. RAZ supra note 148. But see Hart, Problems of Philosophy of Law, in 6 THE ENCY-
CLOPEDIA OF PHILOSOPHY 268-69 (P. Edwards ed. 1967):
It has been contended that the application of legal rules to particular cases cannot be re-
garded as a syllogism or any other kind of deductive inference, on the grounds that neither
general rules nor particular statements of law (such as those ascribing rights or duties to
individuals) can be characterized as either true or false and thus cannot be logically related
1985] PRACTICAL REASONING AND JUSTIFICATION 77

It is beyond the scope of this essay to consider the many intricate

arguments that bear on this question of the logical form of legal state-
ments. Many remarks by judges and lawyers have the surface form of
a proposition, purporting to describe some state of affairs, for example:
"[A] moral obligation is a sufficient consideration to support a subse-
quent promise to pay where the promisor has received a material bene-
fit."' 5 5 But the apparent propositional form of a statement is not
conclusive. Research in both linguistics and philosophy of language
has led to the acknowledgment that the true form of a statement may
be other than its apparent form.' 5 6 Even if the statement quoted above
is a valid rule of law, we are not thereby committed to the truth of the
apparent assertion about moral obligation, nor even to the existence of
a thing called "consideration," which could be either sufficient or not.
We may, instead, construe legal rules as having the form of norms, or
as having the form of a directive to the deciding judge.
Whether or not one accepts a theory that construes legal state-
ments as having a non-propositional form, such theories have been
important in jurisprudence. But they are incompatible with the de-
ductive thesis and its commitment to the truth-preserving criterion of
validity. Other things being equal, we should prefer a theory of legal
reasoning that does not force us to accept an aggressive position on
logical form. We might, for independent reasons, determine that the
best view on this subject is as the deductive thesis supposes, but we
should be reluctant to view legal statements as propositional in form
just because we accept a deductive theory of legal reasoning.
There is a further discrepancy posed by the deductive thesis' re-
quirement that legal statements be either true or false. To suppose
that legal statements are propositions capable of bearing a truth value

either among themselves or to statements of fact; hence, they cannot figure as premises or
conclusions of a deductive argument. This view depends on a restrictive definition, in terms
of truth and falsehood, of the notion of a valid deductive inference and of logical relations
such as consistency and contradiction. This would exclude from the scope of deductive
inference not only legal rules or statements of law but also commands and many other
sentential forms which are commonly regarded as susceptible of logical relations and as
constituents of valid deductive arguments.
In Section IV it will be argued that the better account of both commands and directives and also of
legal inferences requires just what Hart hopes to avoid: a non-deductive understanding of the state-
ments involved.
155. Webb v. McGowin, 27 Ala. App. 82, 85, 168 So. 196, 198 (1936). The stylistic habit of
writing about rules and principles in the declarative is not limited to judges. Reformers, as well, frame
their pronouncements as if they were matters of fact even though they may be urging changes in ex-
isting law. See. e.g., RESTATEMENT (SECOND) OF CONTRACTS § 86(1) (1981): "A promise made in
recognition of a benefit previously received by the promisor from the promisee is binding to the extent
necessary to prevent injustice." It is arguable that this stylistic device has served political or jurispru-
dential aims. See G. GILMORE, supra note 59, at 12-18, 108-10; M. HORWITZ, supra note 53, at 253-66.
156. See supra note 146.
78 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

suggests that the judge's conclusion is a description of the world, simi-

lar to the hoary "All men are mortal" and "Socrates is mortal." But
this seems unnatural, to say the least. It seems more plausible, in-
stead, to regard the judge's holding as a decision, and not as a
7 5
description.1
Indeed, construing legal statements as descriptions of the world is
rife with conceptual difficulties. A rule, on this argument, should be
construed as a general proposition. Suppose a judge decides a case by
subsuming it under a rule. If the rule is true, then all subsumed cases
have the prescribed legal consequences. Indeed, those consequences
are true whether or not the judge actually decides the case in accord-
ance with that rule; that the judge actually cited that rule is unimpor-
tant. But now suppose that the judge had decided differently than he
in fact decided. How should we regard this other judicial conclusion?
As the falsification of the rule? A counter-instance? It would seem
that the "truth" of the rule depends on the actual decision, but, the
rule can be no more true than the extent to which judges actually fol-
58
low the rule.'
Defenders of the deductive thesis have sought to support it by
avoiding the many criticisms raised against it. In particular, propo-
nents of the "two-level" version of the thesis have sought to
recharacterize the deductive inference as the last step of a chain of
judicial argument, the initial steps of which may not be deductive.' 5 9
It might be thought that such a recharacterization avoids the undesir-
ably aggressive quality of the deductive thesis. It can be seen, how-

157. See, e.g., N. MACCORMICK, supra note 8, at 103-04:

Legal rulings are normative - they do not report, they set patterns of behaviour; they do
not discover the consequences of given conditions, they ordain what consequences are to
follow upon given conditions. They do not present a model of the world, they present a
model for it. (Emphasis in original.)
That MacCormick recognizes the problems consequent to a deductive theory of judicial justification
does not stay him from advocating the theory nonetheless. His response to this particular problem is an
argument that propositions of law are "relatively" true or false - they are true or false relative to a
given legal system at a given time. Id. at 271. But conjuring relativized systems of truth and falsity -
required, on MacCormick's account, to accommodate both the apparent normativity of law and his
commitment to a truth-preserving criterion of validity - surely runs counter to the criteria of theory
preferability outlined above. See supra text accompanying notes 83-86. With this move, MacCormick's
theory loses simplicity, and it makes a series of undesirably aggressive assumptions.
158. Else, the deductive theorist must provide an account of what it means for a legal rule to be
true, if truth does not mean that the rule is operative in the legal system as a premise for deductive
reasoning. Thus, the deductive theory loses simplicity, and risks overly aggressive commitments. For
example, a natural law theorist might accommodate this problem about the truth of a disregarded legal
rule by positing that laws are "true" in the same way that moral laws are "true." But, I take for
granted that recourse to a natural law theory in order to save a deductive account is a paradigm case of
an overly aggressive commitment required by a theory of judicial justification.
159. See N. MACCORMICK, supra note 8, at 100-19.
1985] PRACTICAL REASONING AND JUSTIFICATION 79

ever, that the deductive theorist's commitment to the truth-bearing

nature of legal statements is thoroughgoing and incorrigible, and that
this objectionable consequence of the deductive thesis is unavoidable.
For example, in dealing with the difficulty about the logical form
of legal statements, it might be supposed that commands and direc-
tives could somehow figure in the non-deductive "level" of judicial
reasoning without creating logical difficulties for the deductive tier:
the concluding syllogism. This approach would be misconceived. In
the two-level version of the deductive thesis the non-deductive infer-
ences lead to the adoption of the appropriate legal rule which, in the
concluding syllogism, deductively implies the judicial result. Even if
the reasoning process used to generate that rule is non-deductive in
nature, the rule itself must be a proposition with a truth value. Other-
wise, it could not figure in the concluding deductive syllogism. But,
commands and directives, not bearing truth values, can be neither
premises nor conclusions in arguments that establish the truth of some
proposition. Therefore, commands and norms cannot figure in the in-
ferences of the non-deductive level, inasmuch as the conclusion of that
stage must be a rule, bearing a truth value. So, a two-level version of a
deductive theory cannot integrate commands and directives into the
justification procedure if that justification is fundamentally deductive.
But statements of obligation are commonly the last step of a legal
argument. Some commands are issued directly to the parties, as in the
case of injunctions. Others are less direct but nonetheless authorita-
tive - as in the standard conclusion of an opinion, "It is so ordered."
The deductive theorist must provide some account of how a command
could be said to follow from the conclusion of a deductive syllogism,
in order to provide a deductive account of judicial argumentation. A
long and contentious history adorns the philosophical issue of whether
any descriptive statement could validly imply a statement of obliga-
tion. t " Unless the deductive theorists can resolve this controversy,
deduction cannot account for this common feature of judicial
opinions.
Finally, we may observe that many of the deductive theory's ad-
herents argue for the theory on the ground that we lack an attractive
alternative. "If the value of [legal] logic is totally denied, a reaction to
this denial is to ask what its reasonable alternative could possibly be.
So far there appears to be no sensible reply to this question."' 6 1 Phi-

160. For a collection of a number of insightful essays see THE IS-OUGHT QUESTION (W. Hudson
ed. 1969). This issue was also raised in a legal context by Professor Northrop. See Northrop, Law,
Language and Morals, 71 YALE L.J. 1017, 1031-32 (1962).
161. I. TAMMELLO, supra note 120, at vii. Hart's claim, see supra note 154, seems to involve the
80 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

losophers and logicians have been slow to develop alternatives to de-

duction as a logical account of commands and directives, and lawyers
and jurisprudents have trailed behind the others. This argument
might carry some weight if there really were no viable alternative, for
we need some account of judicial justification in order to pursue re-
lated controversies in philosophy of law. But the force of this claim is
undercut when we consider, in Section IV, recent work providing a
logical account of practical reasoning.
In sum, however attractive the deductive thesis might appear at
first glance, closer examination reveals various consequences of the
thesis that conflict with our understanding of the role of judges and
the legal process. These conflicts pose a challenge to the adequacy of
the thesis: we would prefer a theory which did not raise these conflicts,
1 62
or which provided a more insightful account of the legal process.
Further, the deductive thesis may also be seen to require a series of
aggressive theoretical claims about the nature of law. The aggressive-
ness of these claims should make us reluctant to embrace the deduc-
tive thesis even if it did not commit us163to positions inconsistent with
our understanding of the judicial role.

B. The Analogical Thesis

1. Analogy and Judicial Justification
The second dominant view of judicial justification describes
judges' arguments as depending on analogical inferences, or "reason-
ing by example." "Analogical reasoning involves the passage from as-
sumed or given resemblances to an inferred resemblance."'t64 The
analogical thesis received an extended treatment in E. Levi's An Intro-
duction to Legal Reasoning,165 which remains the locus classicus of the
view. The argumentation of judges and lawyers, Levi contended:
.. .is reasoning from case to case. It is a three step process de-
scribed by the doctrine of precedent in which a proposition de-
scriptive of the first case is made into a rule of law and then
applied to the next similar situation. The steps are these: similarity
is seen between cases; next the rule of law inherent in the first case
is announced; then the rule of law is made applicable to the second
case. 166

same kind of argument: inferences among commands must be deductive, else they would not be suscep-
tible of logical relations.
162. See supra text accompanying notes 83-86.
163. Id.
164. M. GOLDING, supra note 8,at 102.
165. E. LEVI, AN INTRODUCTION TO LEGAL REASONING (1948).
166. Id. at 1-2.
1985] PRACTICAL REASONING AND JUSTIFICATION 81

A quick canvas of the literature of jurisprudence indicates that

this view of judicial justification has enjoyed great popularity and is
widely regarded as an accurate description of a fundamental process
by which judges reach their conclusions. 6 7 Those who advocate anal-
ogy as a model for judicial reasoning, however, have failed to develop
an adequate theory which can account for the significant features of
judicial justification which were noted in Section IH.168
In the first place, the advocates of analogy as the foundation of
judicial argument have failed to illuminate the logical structure of an
analogical argument. What are its premises and its conclusions?
What is the inference that relates the former to the latter? Advocates
of analogy have yet to extend their claim about judicial reasoning in a
way that illuminates these important conceptual issues.
We may nevertheless distinguish two different versions of analogi-
cal inferences appropriate to judicial justification. The first of these
involves an inference to a rule, and then an inference from a rule to a
specific case. The second version posits an inference directly from one
case to the next, without the intermediating rule.
Levi's characterization of analogy is an example of the first ver-

167. "Reasoning by analogy is the most prevalent form of legal reasoning." Murray, The Role of
Analogy in Legal Reasoning, 29 UCLA L. REV. 833, 847 (1982). See also Burton, supra note 9, at 1145
(common law reasoning "works primarily by analogical reasoning"); Soper, supra note 72, at 487
("Legal reasoning, deductive only in a trivial sense, is primarily a matter of determining relevant simi-
larities and differences among fact situations that distinguish cases covered by a standard from cases
that are not.").
The role of reasoning by analogy in the law is discussed at some length in M. GOLDING, supra note
8, at 44-49, 102-12. While Golding argues that judges will commonly employ analogical arguments, his
discussion indicates that, at bottom, he views judicial argumentation as distinct from analogical reason-
ing as ordinarily understood. Indeed, he appears to view judicial reasoning as fundamentally practical
reasoning. He argues, most notably, that questions of judicial classification have the form: "Should X
be treated as a Y for certain legal purposes?" This way of putting the question, he argues, "has the
advantage of revealing that the judge's affirmative answer [on the question of a steamboat owner's
liability in Adams v. New Jersey Steamboat Co., 151 N.Y. 163, 45 N.E. 369 (1896)] is based on the
claim that the same practical legal argument for imposing a stringent responsibility on innkeepers is
also applicable to steamboat proprietors, because of the similarity between the two cases." M. GOLD-
ING, supra note 8, at 106 (emphasis in original).
However, the proper reading of Golding's view is uncertain. For, he understands "practical rea-
soning" as a description of the kind of reason advanced for an action, and contrasts it with deduction,
induction, and analogy, which he describes as kinds of arguments. In other words, he does not appear
to consider the possibility that practical reasoning could be understood as a distinct form of argument
- different from but comparable to deduction or analogy. In this respect, he seems unaware of the
work by Kenny and Hare on which Section IV of this paper relies in arguing that practical reasoning
represents a distinct form of reasoning. See infra text accompanying notes 192-211. Nonetheless,
Golding's characterization of legal reasoning draws several distinctions between judicial argumentation
and analogical reasoning as the analogical theorists would describe it. M. GOLDING, supra note 8, at
102-12.
168. See supra text accompanying notes 35-71.
82 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

sion. He describes analogy as a three step process.' 69 First, similarity

is perceived between two cases, and, second, the rule of law "inherent"
in the first case is announced. Then, third, the announced rule is ap-
plied to the second case.
Nothing in Levi's discussion indicates the nature of the first or
second steps of the process. t70 The third and last step is described in a
fashion that suggests a deductive inference: a rule of law is applied to a
particular case. So far as he has described this last step, Levi has failed
to indicate the nature of the inference that relates the applicable rule
to the case.
It is at least arguable that, so far as Levi has characterized them,
judicial arguments by analogy are just one form of reasoning by de-
duction. t7 t We noted in Section IIIA that deductive theorists have
propounded a two-level procedure in which only the last step of the
justification is a deductive argument.17 2 If, in an analogical justifica-
tion, the "application" of a rule to a case is a deductive inference, then
Levi's description of analogy would be consistent with MacCormick's
two-level deductive theory of judicial justification. The first and sec-
ond steps of the analogical process might be understood as one kind of
argument at the non-deductive level of justification. The third step
would be an ordinary instance of deduction, deriving the resolution of
the case at hand from a general standard. In short, Levi's description
of the analogical argument is so underdeveloped that we cannot be
sure that it differs in any important way from the two-tier version of
the deductive thesis we observed in Section IIIA.
Recent investigations of legal arguments by analogy have sug-
gested a second conception of analogical arguments which minimizes
the significance of the legal rule. 17 3 The premises of an analogical ar-

169. E. LEVI, supra note 165, at 1-2. Precisely how Professor Levi understood analogical reason-
ing is a complicated matter, see, e.g., id. at 4 n.8, and this article will not undertake the exegisis neces-
sary to attribute to him any particular view.
170. Indeed, it is possible that the first two steps are meant by Levi as involving questions of the
context of discovery rather than the context of justification. See supra note 42. In these remarks I
assume Levi meant the latter.
171. Cf, Soper, supra note 72, at 487 (legal reasoning is deductive in a "trivial sense"). While this
is an attractive reading of Levi, since it would clarify other difficulties with his view, his book provides
too little detail on which to base an interpretive claim. See also M. GOLDING, supra note 8, at 103 n.l 1.
172. See supra text accompanying notes 108-10.
173. See M. GOLDING, supra note 8, at 102-11. In Golding's view, arguments by analogy are best
understood not as inferences to a rule which then may be applied to the next case, but rather as ques-
tions of classification. Id. at 103. Golding argues, in analogizing the case at hand to some line of
precedent, that the judge is attempting to answer the question: Should this case be placed in the same
category with those cases? More precisely, he views legal analogies as having the following structure:
(i) Case x has characteristics F, G, . . . [.]
(ii) Case y has characteristics F, G, ... [.]
(iii) x also has characteristics H.
1985] PRACTICAL REASONING AND JUSTIFICATION 83

gument, on this conception, are various statements of similarities or

differences between the two cases. The argument's conclusion is the
resolution of the second case in a fashion similar to the decision in the
first case.' 74 Whatever the premises and conclusion, on this concep-
tion, it is difficult to specify the nature of the inference connecting the
1 75
former to the latter.
Proponents of the analogical thesis have also failed to illuminate
the normative features of judicial justification. What makes an analog-
ical argument valid? What are the criteria by which we could evaluate
any given use of analogy by a judge? Consider the situation in which a
problematic legal dispute is similar to two (or more) different prece-
dents. There is some analogy to be drawn between the case at hand
and each of the precedents. Which analogy is preferable?
Under Levi's conception of analogical arguments, discussed first
above, each competing precedent might give rise to a rule of law. If
the competing precedents are sufficiently dissimilar, then the rules "in-
herent" in the different cases could be different. These different rules
could yield different and possibly inconsistent legal consequences for
the case at bar. 176 In this situation, it would be possible to analogize
the dispute in two or more conflicting ways. So far as Levi's view has
been developed, no comparison is a priori illegitimate if some similar-
ity obtains. Hence, the choice of analogy - and, through the choice
of analogy, the choice of legal rule - can only be evaluated in terms of
the comparative similarity of the disputed situation to one or another
of the available precedents. There is no difference between acceptable

(iv) F, G, . . . are H-relevant characteristics.

(v) Therefore, unless there are countervailing characteristics, y has characteristic H.
Id. at 107. Characteristic H, on this view, is some legally relevant attribute. Although on Golding's
model no rule of law is derived from the past cases, there is a rule of decision: unless there are counter-
vailing characteristics the similarities between x and y warrant classifying both in category H. Joseph
Raz has offered an account for analogical arguments which parallels Golding's at least for cases where
there is no binding precedent. See J. RAZ, THE AUTHORITY OF LAW 201-06 (1979).
Murray's account, supra note 167, also seems to deny the relevance of rules to analogical argu-
ments. He relies on an Aristotelian account of reasoning by example as reasoning "from part to part"
and contrasts that reasoning with reasoning from part to whole or from whole to part. Id. at 847 (citing
Aristotle, PriorAnalytics, in THE BASIC WORKS OF ARISTOTLE 62, 103 (R. McKeon ed. 1941). How-
ever, at other places Murray suggests that analogical arguments may be used to establish new rules or
reform old ones. See Murray, supra note 167, at 851.
174. M. GOLDING, supra note 8, at 107; Murray, supra note 167, at 841 n.30.
175. Golding's discussion indicates that he is uncertain about some parts of the inference pattern
he has outlined. See M. GOLDING supra note 8, at 107-12. In particular, it is not clear whether the
analogical inference is a matter of determining the truth of a classification or the usefulness of classify-
ing a case in one category rather than another. See id. at 106-07. See also supra note 167. Murray
discusses the structure of an analogical inference at only one point. See Murray, supra, note 167, at 841
n.30.
176. See supra note 131 and accompanying text.
84 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

and unacceptable analogies in terms of the form of the inference in-

volved, and each rule is warranted by its precedents. The competing
authorities can be differentiated only in terms of the resemblances of
the case at hand to the various precedents.
The same point can be made with respect to the second version of
analogical reasoning. The premises of the analogical inference, on this
conception, are the similarities between the two cases. By hypothesis,
however, there will be similarities between the case in dispute and each
of the competing precedents. If the prior cases were decided in incon-
sistent ways, then the correct disposition of the case in dispute remains
unsettled; nothing about the form of the analogical inferences gives us
reason to prefer one analogy over another.
A theory of judicial justification, if it is to be useful, must provide
us with a criterion or set of criteria by which we can discriminate valid
from invalid arguments. In the situation where competing analogies
suggest different dispositions of a disputed case, neither version of the
analogical thesis has provided a basis on which we might discriminate
good analogies from bad. 177
The choice of legal authority is sometimes said by advocates of
the analogical thesis to depend on the relative importance or signifi-
cance of competing analogies,17 8 or, more often, on the relevance of the
various similarities. 79 Whichever description they might prefer, pro-
ponents have failed to provide useful criteria by which we might evalu-
ate judicial argumentation and assess the respective analogies. Neither
"significance" nor "relevance" have been given any independent con-
tent. What makes one similarity relevant, but another not? We are
left, it seems, with nothing more than our intuitive sense of the attrac-
tiveness of different possible analogies. Whether we regard a given
opinion as having justified its conclusion depends on whether or not
we concur with the deciding judge's aesthetics regarding similarities.
The analogical thesis, in short, has not provided us with any useful
normative insights.

2. Evaluating an Analogical Theory of Judicial Justification

The analogical thesis has been so little developed that we can
barely deem it a true theory of judicial justification. Undeveloped as it
is, however, it can also be seen to lead to an unnecessarily problematic
view about the logical nature of legal statements.

177. Many of the theorists who have considered argument by analogy have made this very point.
See. e.g., Murray supra note 167, at 851-52; M. GOLDING, supra note 8, at 45, 111.
178. See, e.g., Murray, supra note 167, at 851.
179. See, e.g., M. GOLDING, supra note 8, at 45; J. RAz, supra note 173, at 202-04.
1985] PRACTICAL REASONING AND JUSTIFICATION 85

Philosophers and logicians standardly describe reasoning by anal-

ogy as a species of inductive reasoning. 8 ° On this characterization,
the conclusion of a judicial analogy is a remark about the likelihood
that the case at bar is a member of the same legal category as the
selected body of precedent. 18 ' Proponents of the analogical thesis
have generally failed to rebut this characterization.' 8 2
However descriptively suggestive the analogical thesis may ap-
pear at first blush, it seems clear that we would not regard the judicial
holding as merely an observation that some case is likely to be of such-
and-such a kind. Nor would we suggest that a lawyer's brief, employ-
ing a legal analogy to urge a particular decision - is arguing that one
probabilistic conclusion is more accurate than another. Instead, a
judge's conclusion is a decision - a pronouncement that the law re-
quires what he holds - and the lawyer's argument urges just such a
decision for his client.
Moreover, since induction is a form of argument to conclusions
regarding the likely truth or falsity of various propositions, legal anal-
ogies seem to yield conclusions about the likely truth or falsity of vari-
ous legal holdings. Adopting the analogical thesis seems to commit us
to the view that statements of law are true or false. As we saw in
Section IIIA, this commitment precludes us from pursuing any other
view of the logical nature of legal statements. This thesis, then, begs
an important question about legal theory and is, for that reason, unat-
tractive.' 8 3 Again, we might decide to adopt a position in jurispru-
dence which holds that legal statements bear truth values. But we
should accept that position because we think that it is the better view
of law and legal reasoning, and not just because we believe that judi-
cial argumentation is frequently analogical in nature.
We may also observe that different inductive arguments, all valid,
may lead to inconsistent conclusions. This means that conceiving of
analogical arguments as inductive in nature has one possible advan-

180. See, e.g., W. SALMON, supra note 113, at 97-100.

181. Id.
182. Indeed, proponents of analogy as an essential form of judicial justification have generally
failed even to try to rebut this characterization. One exception is Murray, supra note 167. He distin-
guishes analogy from induction on the ground that induction arrives at a conclusion or rule, but does
not apply that conclusion to a new case; analogy, however, applies the conclusion to a new case. Id. at
847. Most discussions of induction in logic and philosophy, however, would contradict Murray's asser-
tion that induction does not apply its conclusions to new particulars. See, e.g., W. SALMON, supra note
113, at 97-100. Following the standard description of inductive reasoning, we may note two different
forms of inductive inference: one form, induction by enumeration, leads to a conclusion in the form of a
statistical generalization. Id. at 83-84. A second form, statistical syllogism, applies the statistical gen-
eralization to a particular situation. Id. at 87-91.
183. See supra text accompanying notes 83-86.
86 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

tage for the development of a theory of judicial justification. We

noted, in Section IIIA, the distinction between a rule's applicability
and its warranted application. 184 This distinction, we saw, entailed a
difficulty for the deductive thesis in a situation where two or more
conflicting rules applied to the same dispute. Inductive arguments,
however, are somewhat more amenable to this feature of judicial
justification.
Unlike deductive arguments, inductive arguments do not necessi-
tate their conclusions.'8 5 To the contrary, adding additional relevant
data can alter the truth of an inductive conclusion. 86 For example,
compare these two inductive inferences:
A. (1) The vast majority of 35-year old American men will sur-
vive for three more years.
(2) Henry Smith is a 35-year old American man.
(3) Therefore, Henry Smith will survive for three more years.
B. (1') The vast majority of men with advanced lung cancer will
not survive for three more years.
(2') Henry Smith has advanced lung cancer.
(3') Therefore, Henry Smith will not survive for three more
87
years. 1
Each inference appears acceptable and each relies on true premises.
But, these arguments nonetheless lead to incompatible conclusions.
If we conceive of competing rules as different pieces of relevant
evidence, then we can explain the conflict between competing rules:
they are different factual characterizations of the controversial situa-
tion, each leading to a different inductive conclusion. Conceiving of
analogical arguments as inductive does not advance us far enough,
however, for it still does not help us distinguish which of the applica-
88
ble characterizations would be warranted.
The analogical thesis seems to offer one significant insight - its
perspicacity in acknowledging that judges sometimes justify a decision
by comparing the case before them with other decisions. Beyond this
insight, the chief virtue of the thesis for many of its defenders seems to
be that it is not a deductive account.' 89 These are weak reeds for the
selection of a theory, and it is at least arguable that analogy does not
even provide a full-fledged theory of judicial justification to compare

184. See supra text accompanying note 130.

185. W. SALMON, supra note 113, at 13-15.
186. Id. at 89.
187. Id.
188. Id. at 90-91.
189. See, e.g., Burton, supra note 9, at 1142-44.
1985] PRACTICAL REASONING AND JUSTIFICATION 87

with deduction, already considered, or practical reasoning, to be con-

sidered next.

IV. PRACTICAL REASONING AND JUDICIAL JUSTIFICATION

While the deductive and analogical theses have dominated the

literature concerning legal reasoning, there are other accounts of legal
argumentation.19 In this Section, I advance an alternative theory of
judicial justification which conceives of judicial arguments as examples
of practical reasoning.191 In practical reasoning we reason from ends
to means - from our aims and needs to conclusions about what to do.
Practical reasoning, on the view I present here, is a distinct form

190. See, e.g., T. PERRY, supra note 88; J. HOROVITZ, supra note 88. See also supra note 149
(commenting on deontic logic).
191. rhe view that judicial justification involves practical reasoning in some sense or other has
been suggested by a number of writers. See, e.g., M. GOLDING, supra note 8, at 106; Ladd, The Place of
PracticalReason in JudicialDecision, in VII NoMos: RATIONAL DECISION 126 (1964); Visser't Hooft,
On Legal Reasoning and the Concept of PracticalReasoning, 13 RECHTSTHEORIE 269 (1982). Even
proponents of other views have made remarks suggesting that they think that legal reasoning is some-
how "practical" in nature. See, e.g., Murray, supra note 167, at 848 n.45; N. MACCORMICK, supra note
8, at 108, 266-74.
Few of these writers, however, have articulated an effective criterion of validity by which valid
practical judicial arguments could be distinguished from invalid. As a result, the reader is frequently
unable to ascertain just what it is that supposedly makes practical reasoning distinct from any other
kind of reasoning. My contention, following Kenny and Hare, is that we may distinguish practical
reasoning in terms of the logics of satisfactoriness and satisfaction, and valid practical inferences in
terms of the respective criteria of validity of these logics.
The significance of distinguishing practical reasoning from deduction in terms of the different crite-
ria of validity is highlighted by the schizoid position taken on legal reasoning by Neil MacCormick. In
a variety of places, MacCormick has suggested that legal reasoning is a form of practical reasoning. See
N. MACCORMICK, supra note 8, at 1-8, 166-74; N. MACCORMICK, H.L.A. HART: PROFILES IN JURIS-
PRUDENCE (1981); MacCormick, Legal Reasoning and PracticalReason, in VII MIDWEST STUDIES IN
PHILOSOPHY 271-86 (P. French, et al. ed. 1982). But, MacCormick has been a persistent and forceful
exponent of the deductive theory, and shows no sign of recanting on this point. See MacCormick, The
Nature of Legal Reasoning: A Brief Reply to Dr. Wilson, 2 LEGAL STUD. 286 (1982). It appears, then,
that MacCormick feels that deduction and practical reasoning are fundamentally compatible. How-
ever, the examples in Section IIIA of this paper show how inferences among commands are not valid in
deduction, while, as this section will indicate, they may nonetheless be valid in the logic of satisfactori-
ness. Thus, if practical reasoning is captured by the logics of satisfactoriness and satisfaction, practical
reasoning and deduction are not compatible.
Of the welter who have in some sense advocated practical reasoning as a model for judicial argu-
mentation, only two have arguably anticipated my position, relying on the logics of satisfactoriness and
satisfaction. Joseph Raz has argued at some length that legal statements are norms and that reasoning
among norms is practical reasoning. See J. RAz, supra note 148. In a footnote, Raz asserts that the
reasoning involved in practical inferences "conform[s] to the logic of satisfactoriness." Id. at 181 n.7.
He does not, however, develop this aspect of his thesis, nor does he draw out the differences between the
logics of satisfactoriness and satisfaction on the one hand, and deduction or analogy on the other.
Martin Golding has also suggested that judicial reasoning is practical reasoning. See M. GOLD-
ING, supra note 8, at 55-60, 102-12. He does not develop, however, what he means by practical reason-
ing other than to cite J. RAZ, PRACTICAL REASON AND NORMS (1975) and PRACTICAL REASONING (J.
Raz ed. 1978).
88 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

of reasoning, with its own logical structure and criteria of validity. In

part A of this Section, I outline the nature of practical inferences and
the criteria of valid practical arguments. In part B, I suggest how
practical reasoning can explain the important features of judicial justi-
fication. Finally, I argue in part C that practical reasoning provides a
better theory of judicial justification than either the deductive or ana-
logical theses.

A. PracticalReasoning
As an introduction to practical reasoning, consider the following
example, adapted from Aristotle. 9 2
I need a covering to keep me warm.
A cloak is a covering.
Therefore, I need a cloak.
I must make what I need.
I must make a cloak.
Two things are significant about the inferences involved in this
chain of reasoning. First, there is a sense in which we would under-
stand the reasoning to be valid. From a practical problem the speaker
has reasoned to a plan of action which, so far as we can tell, will satisfy
his needs. Second, while the reasoning is valid, its validity is not de-
ductive in nature. From the premises "I need a covering to keep me
warm" and "A cloak is a covering" the conclusion "Therefore, I need
a cloak" hardly follows in a truth-preserving form of argument. To
19 3
the contrary, the argument is an instance of a classic fallacy.
Deduction is not the only form of reasoning on which we can
rely. Practical reasoning, as distinguished from deduction, is both
ubiquitous and unproblematic. We reason from ends to means in all
manner of ordinary situations without undue concern: whether to take
the bus rather than drive; whether to eat at home or go out; whether to
invest in money funds or the stock market. And, we accept the con-
clusions of our practical deliberations without inquiring into their de-
ductive validity. R.M. Hare' 94 and Anthony Kenny' 95 have
illuminated the nature of practical inferences and have provided an
account of their validity. Kenny's contributions, in particular, show
that practical inferences have a distinct logical structure in terms of

192. See A. KENNY, supra note 7, at 72 (citing ARISTOTLE, DE MOTU ANIMALIUM (701a18)).
193. Logicians commonly refer to this as the fallacy of affirming the consequent. See, e.g., I.
Copi, supra note 113, at 233, 271.
194. R.M. HARE, supra note 7; R. M. HARE, THE LANGUAGE OF MORALS (1952).
195. A. KENNY, supra note 7. Parts of his book have been reprinted in Kenny, PracticalReason-
ing and Rational Appetite, in PRACTICAL REASONING, supra note 191, at 63-80.
1985] PRACTICAL REASONING AND JUSTIFICATION 89

which deciding to make a cloak can be a valid practical conclusion

even if the inference is not deductively valid.
At the heart of these analyses of practical reasoning lies a distinc-
tion between the descriptive content of a sentence like "I need a
cloak,"196
'
and its mood. 19 7 The command "Close all the windows in
the school" has the same descriptive content as the assertion "All the
windows of the school are closed," but a different mood. Expressions
of desire and intention, such as "I need a covering to keep me warm"
and "I must make what I need," share, on this analysis, the same
mood as commands. Together, sentences with this mood are desig-
nated as fiats.198 "Fiats contain descriptions of possible states of af-
fairs whose actualization satisfies the desires expressed by them."' 99
Where p is some proposition which describes a state of affairs, then
F(p) stands for the fiat which we could express by the statement "Let
it be the case that p."
Practical reasoning is reasoning that involves fiats, and valid
practical inferences are valid inferences among fiats. Deductive logic
may be distinguished by its criterion of inferential validity: the truth-
preservingness of the form of the argument at issue." ° But, the valid-
ity of inferences among fiats cannot be captured by this criterion.
Valid inferences among commands, for example, are not valid in de-

196. Termed by Hare a "phrastic." See THE LANGUAGE OF MORALS, supra note 194, at 18.
197. Hare originally termed the marker of mood - imperative, asertoric, or the like - as a
"neustic." Id. He apparently now terms it a "tropic" instead. See A. KENNY, supra note 7, at 39 n. 1.
198. Kenny acknowledges Hofstadter & McKinsey, The Logic of Imperatives, in PHILOSOPHY OF
SCIENCE, 446 (1939) as the origin of the term "fiat." See A. KENNY, supra note 7 at 39. The class of
fiats includes commands, requests, intentions and desires. Commands and requests together are termed
directives, which are distinguished by their being uttered to an agent. For a fiat to be satisfied, the state
of affairs described in it must obtain. For a directive, the state of affairs must be brought about through
the agency of the person to whom the directive is given, and because of the utterance of the directive.
199. A. KENNY, supra note 7, at 79.
200. As Kenny has argued, it is useful to distinguish among different logics in terms of the rules
of inference which they deem valid:
[I]t is possible to think of logic as primarily the study of patterns of inference rather than
the formalization of logical truths. Historically, both approaches have been made domi-
nant in the work of different authors. . . . Most contemporary logical systems employ
both logical truths and rules of inference: in axiomatic systems such as Frege's the logical
truths which are the axioms and theorems are dominant, and rules of inference within the
system are applied only to logical truths; in natural deduction systems such as Gentzen's it
is the rules of inference which are basic; they are applied to non-logical propositions and the
logical truths are yielded only as the results of particular applications of the rules.
In assertoric logic, to every valid inference schema there corresponds a logically true
conditional statement, and to every logically true conditional statement there corresponds a
valid inference schema; so that it may seem to be a matter merely of combinatorial ele-
gance, not of philosophical importance which approach is taken. But of course when we
turn to study the logic of imperatives or fiats, the matter is quite different. For here, it
seems, we can take only the rules-of-inference approach. For commands, requests, and
wishes, not having truth-values at all, a fortiori cannot be logical truths.
Id. at 85-86.
90 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

duction, and inferences which are valid when truth-bearing proposi-

tions are involved are unacceptable when commands are at issue.2 ° t
To express adequately the inferential relations among fiats we need
two other logics - the logic of satisfactorinessand the logic of satisfac-
tion - and criteria of validity appropriate to each of these logics.
For an account of judicial justification, the logic of satisfactori-
ness is the more significant of these logics. The Aristotelian arguments
leading to the need for a cloak - from the need for a covering to a
need for a cloak, and from a need for a cloak to the necessity of mak-
ing a cloak - involve inferences in the logic of satisfactoriness.2 °2 The
logic of satisfaction, on the other hand, concerns the relationship be-
tween a fiat and the actual state of affairs, namely, whether the fiat is
fulfilled. 2 3 Let me explain these two logics, and their respective crite-
ria of validity, in turn.
(a) Satisfactoriness. In practical reasoning we consider the merits
of decisions and plans of action. We are not concerned whether some
state of affairs is true or false, but whether instead the plan or decision
will serve our purposes and gratify our desires. In Kenny's terminol-
ogy, we wish to know whether the plan is satisfactory.
A plan is satisfactory or not with respect to a certain set of wants
and desires. Satisfactoriness, unlike truth, is a relative notion. More
precisely, a fiat is satisfactory relative to a set of wants and desires if
and only if whenever the fiat is satisfied every want and desire in the
set is satisfied. 2°
In deduction, validity is defined in terms of an argument's truth-
preservingness: if the argument form is deductively valid then there is
no possibility of inferring a false conclusion from true premises. Va-
lidity in the logic of satisfactoriness prevents our inferring an unsatis-
factory plan of action from a satisfactory goal. Valid forms of
argument in this logic are satisfactoriness-preserving. If the goal is
satisfactory and the inferences valid then the plan chosen to imple-
ment the goal will be satisfactory as well. More precisely, a fiat F(b)
may be validly inferred from another fiat F(a) in the logic of satisfacto-
riness if and only if "necessarily whenever [F(a)] is satisfactory [rela-
20 5
tive] to a certain set of wants then [F(b)] is satisfactory to that set."

201. Id. at 73.

202. The inferences involved in the example of the cloak are interesting in that the premises
include both fiats and propositions. Handling such mixed inferences requires a few special rules. See
id. at 83-84 n.10.
203. Id. at 81.
204. Id. at 80-81.
205. Id. at 81. In describing deductive inferences, one normally says that one proposition implies
another, but the sense of implication at work in deduction is a technical one, defined truth-functionally
1985] PRACTICAL REASONING AND JUSTIFICATION 91

Relative to that set of wants we will have reasoned from our ends to
good means for achieving those ends.
Two features, in particular, are notable about the logic of satisfac-
toriness. First, it is the mirror image of deductive logic. Suppose that
p and q are respectively the descriptive contents of the fiats F(p) and
F(q). If it is valid in deduction to infer q from p, then F(p) may be
validly inferred from F(q) in the logic of satisfactoriness.20 6 Second,
the relativity of satisfactoriness means that the same plan may be satis-
factory relative to one set of desires but not satisfactory to another set
with different constituents. Even if the desires and wants which popu-
late the two sets are largely the same, the plan's satisfactoriness rela-
tive to one set does not guarantee its satisfactoriness relative to the
other.
(b) Satisfaction. An adequate account of practical reasoning also
requires the logic of satisfaction. With the logic of satisfaction we may
conclude whether a fiat has been satisfied - whether a command has
been obeyed or a project has been carried out. The logic of satisfaction
expresses the inferences from fiats to their fulfillments. This inference
is straightforward: a fiat F(p) is satisfied only when the assertion p -
the fiat's descriptive content - is true.2 °7 Valid inferences in the logic
of satisfaction are satisfaction-preserving. "[T]hey are designed to 2pre-
08
vent one from passing from a satisfied fiat to an unsatisfied fiat."
The logic of satisfaction is, as Kenny has put it, "an exact and
uninteresting parallel" of deductive logic. 209 "[W]henever we can in-
fer [the truth of q] from [the truth of p] we can infer the satisfaction of
[F(q)] from the satisfaction of [F(p)]. '' 21° The logic of satisfaction is
also the mirror image of the logic of satisfactoriness. If it is the case

in terms of the truth of the propositions involved. See, e.g., W. SALMON, supra note 113, at 37-38. To
avoid confusion with this technical sense, I described the relations among fiats in terms of the passive
"infers."
206. A. KENNY, supra note 7, at 81-82. In deduction, an argument is valid if and only if the
hypothetical sentence whose antecedent consists of the argument's premise or premises and whose con-
sequent is the conclusion is a tautology. Thus, in the immortal example of logicians everywhere, the
argument:
All men are mortal.
Socrates is a man.
Therefore, Socrates is mortal.
is valid because the complex hypothetical sentence "If all men are mortal and Socrates is a man, then
Socrates is mortal" is a tautology. Tautology may be tested in a number of ways. See supra note 115.
Since the logic of satisfactoriness is the mirror image of deduction, the fiat F(q) may be validly inferred
from the fiat F(p) in the logic of satisfactoriness if and only if the sentence "Ifp then q" is a tautology.
207. A. KENNY, supra note 7, at 82.
208. Id. at 81.
209. Id.
210. Id.
92 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

that when F(p) is satisfied F(q) is also satisfied, then if F(q) is satisfac-
tory F(p) is also satisfactory.
Following Kenny and Hare, we may explain practical inferences
as non-deductive inferences among fiats, including commands and in-
tentions.211 We may assess the validity of particular practical infer-
ences in terms of the logic of satisfactoriness - whether the decision is
a good one relative to a set of wants and desires - and in terms of the
logic of satisfaction - whether the decision has been carried out. By
using valid practical arguments we provide good reasons for our ac-
tions; the conclusions are good means to given ends.

B. PracticalReasoning and JudicialJustification

It is at least initially plausible that judicial argumentation should
involve practical reasoning. Judges' arguments are commonly practi-
cal in force, and their conclusions are decisions about what to do.
Thus, we would expect that judicial reasoning could be understood as
reasoning to the fulfillment of purposes and values in the legal system.
What remains is to demonstrate that practical reasoning, under-
stood in terms of the logics of satisfactoriness and satisfaction, can
adequately account for the special features of judicial justification that
I noted in Section II, supra.212 Practical reasoning is a distinct form of
reasoning, different from deduction and from analogy (if analogical
arguments are understood as probabilistic in nature). We can identify
the premises and conclusions which constitute a practical inference,
and we can evaluate the inference as valid or invalid in the logics of
satisfactoriness and satisfaction. In short, valid practical arguments
provide reasons for action, and, if the practical arguments are valid,
justify the decisions about what to do. The logics of satisfaction and
satisfactoriness provide effective criteria for distinguishing valid judi-
cial arguments from invalid.2 13 In these respects, at least, practical
reasoning is an alternative to deduction or analogy as a form of rea-
soning which might explain judicial argumentation.2 14 This account

211. According to Joseph Raz, inferences among norms also conform to the logic of satisfactori-
ness. See J. RAz, supra note 148, at 181 n.7.
212. Supra text accompanying notes 35-71.
213. The logic of satisfaction parallels deductive logic, and the logic of satisfactoriness is a mirror-
image of deductive logic. See supra text accompanying notes 204-10.
214. In other respects, it might be argued that practical reasoning is not as conceptually attractive
as is, say, deduction. In contrast to the rich and complex metatheory which logicians and mathemati-
cians have developed for deductive and modal logics, there has been a paucity of formal work done with
the logics of satisfactoriness and satisfaction. For two formulations of these logics, see Levinson &
Atlas, A Version of A.J. Kenny's Logic of Practical Inference, Xerox TS, Mathematical and Social
Sciences Board Workshop of the Formal Pragmatics of Natural Language (1973); Atlas, More on A.J.
Kenny's Logic of Practical Inference (unpublished, 1975). These papers are developed only for the
1985] PRACTICAL REASONING AND JUSTIFICATION 93

of practical reasoning also explains other salient features of judicial

justification.

1. Rules
Judicial justification involves rules. 21 5 By subsuming a particular
case under a statement of universal form a judge may justify his deci-
sion by applying a legal rule. And, in problematic cases - where no
rule clearly subsumes the case, or where two or more conflicting rules
may apply - the judge may reason to the extension of a rule to an
unprecedented case, or, perhaps, to the formulation of a new rule.
Some non-legal examples will provide a context for demonstrating
how practical reasoning can illuminate these features of the justifica-
tory task.
Consider, for example, the simple fiat expressed by a general in-
tention, "Let it be the case that I close all the windows in the
house."2 6 In the logic of satisfactoriness, it would be valid to infer the
particular intention "Let me close this window" from the general in-
tention together with the additional premise "This is the only window
in the house." If it was satisfactory that I close all the windows, then
it would be satisfactory that I close this, the only window. 2 17 The final
step of the argument involves the logic of satisfaction: the fiat "Let it

propositional, not the quantified, logics of satisfaction and satisfactoriness, and they regard the logics of
practical inference as extensions of standard formal sentential logic. I am unaware of any further for-
mal results; in particular, I have not uncovered any metatheoretical results comparable to complete-
ness, or decidability, which are routine for first order deductive logic. See, e.g., G. HUNTER, supra note
123.
215. See supra text accompanying notes 50-60.
216. It would be more natural to express the command in another fashion - perhaps, "I need to
close all the windows in the house," or "I must close all the windows in the house." The phrasing in
the text is more perspicuous with respect to the reasoning involved.
217. The reasoning involved would be more complicated if there were multiple windows. Then,
from the fiat "Let me close all the windows," the appropriate inference in the logic of satisfactoriness
would be "Let me close the first window, and the second, and the third .... " for as many windows as
there are in the house. To conclude, in a case where there are multiple windows, just "Let me close this
window" would not, in fact, be satisfactoriness-preserving, for the general fiat would not be satisfied if
only one of the windows were closed. See A. KENNY, supra note 7, at 83. For ease of exposition, I
have assumed in this discussion that there is only one window.
This feature of the logic of satisfactoriness is potentially troublesome, for it would appear that we
could never infer a singular fiat from a general fiat in the logic of satisfactoriness. Instead, the only
valid conclusion from a general fiat in the logic of satisfactoriness would seem to be the conjunction of a
string of singular fiats, from which no one of the conjuncts could be validly inferred in the logic of
satisfactoriness. But this is obviously not a barrier in ordinary practical reasoning, for we are perfectly
capable of concluding that we should close this window first and then go on to all the rest of the
windows. So, as part of practical reasoning, we must have some inference procedure that allows us to
focus on one of the conjuncts first, complete that part of our obligation under the fiat, and then proceed
to the rest. I develop later the notion of second-order rules of practical reasoning. See infra notes 227-
29 and accompanying text.
94 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

be the case that I close this window" is satisfied just in case the win-
dow is closed by me.
Legal standards should be general, and I noted, in Section lIA,
how the deductive thesis could account for this expectation of judicial
argumentation: applying a general rule to a particular case can be un-
derstood as a valid deductive inference from a universal proposition.
Fiats, too, can be universal: "Let it be the case that I close all the
windows in the house." A rule's generality can also be understood in
terms of a fiat's universality. Applying the rule to a particular situa-
tion in the logic of satisfactoriness would lead to the conclusion of a
particular fiat: "Let me close this window." The particular command
or norm is a valid conclusion from the general rule if the inference is
satisfactoriness-preserving. In short, applying a rule to a situation is
valid in the logic of satisfactoriness if it is satisfactory that the situa-
tion be resolved in accordance with the rule.2 8
On this account of practical reasoning, a judge's inference from a
legal rule to a decision in a case is an inference from a general to a
particular fiat. Judicial justification is, as the deductive theorists have
claimed, sometimes syllogistic in nature, but the syllogisms need not
be construed as deductive in form. They are better understood as
practical in nature, involving the logics of satisfactoriness and
satisfaction.
To understand judicial argumentation as fundamentally practical
reasoning requires that we posit a variety of goals and purposes to the
legal system, similar to the desires and purposes of an individual.
Legal statements - both rules and specific conclusions - are fiats
which are satisfactory or not relative to various goals and purposes of
the legal system. A rule or decision which serves some set of goals and
purposes is satisfactory relative to that set. Valid applications of a
legal rule are valid inferences in the logic of satisfactoriness. If the
legal rule is satisfactory relative to some set of goals and purposes and
applies to the case at hand, then a valid inference will yield a satisfac-
tory conclusion relative to those goals and purposes.
Let us suppose, for example, that the fiat expressed by the Statute
of Frauds is satisfactory relative to the goals and purposes of contract
law.2" 9 The Statute is a universal fiat and may be applied to a particu-

218. For the same reasons as were indicated in note 217, supra, the situation is more complicated
when the fiat is general. Since the appropriate inference from the general fiat would be a fiat urging the
action be done for each of the particulars in sequence, closing one of the windows and then stopping
would not satisfy the general fiat. Rather, the general fiat would be satisfied only when every one of the
windows was closed.
219. It is perhaps arguable that I have begged a number of important issues regarding judicial
justification by building my account around the arguably special circumstances in which justification
1985] PRACTICAL REASONING AND JUSTIFICATION 95

lar situation - say, an oral contract of employment which cannot be

performed fully within a year of its making. If a fiat calling for non-
enforcement of that unwritten contract is a valid inference from the
Statute in the logic of satisfactoriness, then it would be satisfactory
relative to the purposes of contract law to apply the rule and deny
enforcement of that contract.22 °
Practical reasoning also involves reasoning to rules which are sat-
isfactory relative to certain goals, and it can explain judicial reasoning
to the acceptance of legal rules. Consider first an example from Jo-
seph Raz which exhibits practical reasoning to a rule in an everyday
22
non-legal situation. '
Raz discovers something wrong with his car, and decides to take
it to be repaired the next day. He needs to go to a meeting today, so
he decides to go ahead and drive the car to the meeting, knowing that
it is defective and possibly dangerous. It occurs to him that friends at
the meeting might ask for a ride home, but he doesn't want to take the
risk of driving them in the car until it has been fixed. He knows that if
his friends ask for a ride and he hasn't resolved before they ask to turn
them down, he'll find it difficult to refuse their requests. He decides, in
advance of the meeting, not to give any rides. Raz continues:
I may go further. I may, reflecting on the matter, decide on this
occasion to make it a rule never to take anyone in my car when I
suspect it has some mechanical fault. If so I am simply making a
general decision. Of course even if I adopt the rule now I may
have to decide again in the future what to do in a particular case,
but my problem then will be different from what it would have
been had I not adopted the rule. Having adopted the rule what I
have to decide is whether to act on it in this particular case. What
I am not doing is assessing the merits of the case taking all relevant
facts into consideration.222
As Raz' example indicates, we reason practically to the adoption
of rules in everyday situations. Our premises are desires and beliefs: a
desire to use the car before taking it to the garage and a disinclination
to assume the risk of transporting a friend in the flawed auto, together
with the belief that it will be difficult to refuse a request for a ride

involves a statute as a dispositive norm. But, the Statute of Frauds is extraordinary in that, while
statutory in nature and origin, its place in the legal landscape is more plausibly comparable to a com-
mon law rule. As Grant Gilmore has put it, the Statute of Frauds has become "judicialized" - "the
courts borrow a principle, initially statutory, erect a common law structure around it and forget the
statute." Gilmore, Putting Sen. Davies in Context, 4 VT. L. REV. 233, 234 (1979).
220. See supra note 217.
221. Raz, Reasonsfor Actions, Decisions and Norms, in PRACTICAL REASONING, supra note 191,
at 128-39.
222. Id. at 139-40.
96 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

when it is made. Practical reasoning establishes a plan of action rela-

tive to these premises. That is, following the rule to refuse requests
will be satisfactory relative to those intentions and desires. Further,
we may compare rules in terms of their satisfactoriness. The troubled
owner might have adopted a different rule. He might have decided to
explain the risks to his colleagues who sought a ride and to let each
make an informed choice about whether to ride with him. Or, he
might have required his colleagues to sign a waiver before letting them
ride. Depending on the full range of the owner's desires, these alterna-
tive plans might be satisfactory. In general, we may note that more
than one rule may be satisfactory relative to a given set of desires and
purposes.
Judicial reasoning that leads to the adoption of a legal rule exhib-
its the same structure and can be evaluated by the same criteria as in
the example above. Courts are led to accept a rule if following it will
be satisfactory to the relevant goals and purposes of the legal system.
Consider, for example, the rule in contract law regarding the revoca-
bility of bids submitted to a general contractor. Ordinarily, the offeror
is master of his offer and may withdraw it without liability at any time
prior to its acceptance. 2231oWhen, however, the offer is a bid which the
general contractor may use in preparing his own bid, the general con-
tractor's reliance binds the bidding subcontractor to an "implied sub-
sidiary promise" not to withdraw his offer for a reasonable time.224
This rule at least arguably advances several of contract law's goals. It
protects the general contractor's justifiable and foreseeable reliance on
the subcontrator's offer; this enhances the ease and reliability of the
construction industry's practices regarding bidding. In addition, it
protects the offeror by requiring the general contractor to accept the
bid within a reasonable time and by barring the general contractor
from shopping for a better offer. 225 The rule is valuable, if it is, be-
cause it serves various goals of contract law that underlie the doctrines
of offer and acceptance.
Moreover, we may assess this rule against its rivals in terms of
how well each alternative serves the various goals of contract law. For
example, it might be doctrinally simpler to dispense with the implica-
tion of a subsidiary promise and hold that the general contractor's
reliance on the bid makes the subcontractor's offer enforceable. In
effect, this alternative rule would hold that a general's reliance gives

223. See, e.g., James Baird Co. v. Gimbel Bros., Inc., 64 F.2d 344 (2d Cir. 1933).
224. See, e.g., Drennan v. Star Paving Co., 51 Cal. 2d 409, 333 P.2d 757 (1958).
225. Id. But see Schultz, The Firm Offer Puzzle: A Study ofBusiness Practice in the Construction
Industry, 19 U. CHI. L. REV. 237, 283-85 (1952) (arguing that construction industry does not actually
need bid irrevocability).
1985] PRACTICAL REASONING AND JUSTIFICATION 97

him an enforceable option to contract on the terms submitted in the

bid. In contrast with the implied subsidiary promise rule, making the
subcontractor's offer enforceable as an option, would allow the general
contractor to shop for a cheaper bid: the subcontractor would be
barred from withdrawing his offer and the general contractor would
have an unwarranted advantage in his negotiations with other possible
subcontractors.
While fundamentally similar to an individual's practical reason-
ing to the acceptance of a rule, judicial reasoning is complicated by
special considerations that bear on the selection of one possible legal
rule over its rivals. In reasoning about legal rules, courts must con-
sider that legal rules figure in a system of rules and in a system of
adjudication. Some rules, if adopted, might call into question other
settled areas of the law or, conversely, could trigger the resolution of
parts of the law which are in turmoil. Some rules will be more easily
understood by the affected parties and their counsel, and some will be
easier for courts to follow. These considerations can influence judicial
deliberation about rules to an extent not found in non-legal reasoning
to rules. 226 But judicial deliberation is nonetheless practical in form,
and these considerations are best understood as among the desiderata
to be weighed by the court in deciding whether to adopt a particular
rule.
Even in Raz's simple example of the troubled car owner, the
speaker's evaluations of the satisfactoriness of possible rules plausibly
involved some assessment of what could be called administrative con-
venience. Raz knew that it would be difficult to refuse his friends'
requests at the time the requests would be made, so he was led to
adopt a rule against giving rides. The rule was accepted with an eye to
the likely difficulty of turning down those requests. It is plausible that
he might contemplate a different approach if he thought it would be
hard to stick with the blanket rule against rides. For example, he
might decide on a rule to give rides if asked, but not to volunteer. Or,
if he were steeped in the common law he might search for a fiction by
which he could decline his friends' requests without seeming churlish:
he might exaggerate the danger of riding in his car, or he might claim
that a clause in his insurance policy prevented him from giving rides.
In some situations we will adhere to a previously formulated rule
even though we suspect that had we reassessed de novo the rule's satis-
factoriness in the situation at hand we might have drawn a different
conclusion. We understand that a rule's satisfactoriness may depend
on how, as a general matter, it helps us to dispose of cases. We reason

226. See H. HART & A. SACKS, supra note 21, at 131-86passim.

98 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

practically to the rule's adoption and to its proper application, but we

need not review the rule's satisfactoriness relative to the purposes and
goals of the legal system to justify the proper resolution of each case.
So long as considerations of the ease and prudence of applying and
justifying a rule are among the accepted purposes of the legal system
relative to which a rule must be satisfactory, our evaluations of various
rules in the light of those concerns will be a proper part of practical
reasoning.
More generally, we can distinguish certain kinds of reasons which
elaborate satisfactoriness conditions for our practical reasoning. Fol-
lowing Raz, I call them second order reasons.2 27 Second order reasons
tell us how to decide cases as part of a system of law. Judicial argu-
mentation is rich with standards about how judges should reason in a
particular situation. It is clear that among the standards for satisfac-
toriness in judicial justification are those associated with the idea of
stare decisis. On occasions, applying a rule to a new case will be dis-
tasteful, if not disgraceful. But the legal system recognizes a value in
fidelity to precedent. It is related to the law's ambition to treat like
cases alike. It may also promote stability in and respect for the legal
system. 22 ' Therefore, we expect that judges will generally follow pre-
cedent rather than reassess the satisfactoriness of every problematic
rule on each occasion of its application. And, where there was no rule,
we would expect it to be satisfactoriness-preserving to develop a rule,
for that might ease future case disposition, and provide a system of
rules on which people could rely in ordering their affairs. Judicial rea-
soning may invoke a number of criteria for judging the appropriate-
ness of inferences.22 9

2. Like Treatment
The thesis that judicial justification is fundamentally practical
reasoning explains the judicial justification of legal rules and their ap-
plications to particular cases. The practical reasoning thesis is also
consistent with our expectation that courts treat like cases alike. If a
general fiat is satisfactory, then reasoning from that fiat in satisfactori-
ness-preserving inferences will ensure that every case to which the fiat
is applied will be treated in accordance with the fiat's prescriptions. In

227. Raz has developed this aspect of practical reasoning in PRACTICAL REASON AND NORMS,
supra note 148, at 35-48. Full elaboration of the variety of norms involved in legal reasoning is beyond
the scope of this essay, but it seems unproblematic to suppose that such criteria as the need for "bright"
lines, the hope to curtail further litigation, or the desire to reduce transaction costs may all be under-
stood as second order reasons for deciding a case in a particular way.
228. See H. HART & A. SACKS, supra note 21, at 587-95.
229. J. RAZ, supra note 148, at 35-48.
1985] PRACTICAL REASONING AND JUSTIFICATION 99

short, if the rules are validly applied then like cases will be treated
alike.
The practical reasoning thesis is also consistent with the legal
community's observations about the vicissitudes of judicial justifica-
tion. General legal fiats would be heir to the same misfortunes as the
general legal propositions posited by the deductive theory. On the
practical reasoning thesis, any statement of a rule may be as suscepti-
ble to ambiguity, vagueness, and misunderstanding as would be its
counterparts in deductive or analogical reasoning. Expressions of a
legal rule may also, on the practical reasoning thesis, need interpreta-
tion and construction in order to be applied to a specific case, and may
be vulnerable to manipulation by willful judges. Thus, nothing about
the observable vagaries of legal language commits us to a deductive as
opposed to a practical reasoning theory of judicial justification.

3. Analogies
It is clear that judges sometimes employ analogies as part of the
process by which they reason to satisfactory legal decisions. Since an-
alogical inferences lead to a decision by the judge about how to resolve
a particular controversy, it is at least initially plausible that judicial
analogies are fundamentally practical in nature.
In Section IIIB, I distinguished two different descriptions of the
logical structure of judicial analogies.2 3 ° On one version, an analogy
to past cases leads to the extraction of a rule from those cases, and that
rule is applied to the case at hand. On the other version, the analogy
to past cases justifies the decision without the extraction of a rule.
Each can be explained in terms of practical reasoning.
On Levi's view, the court perceives a rule "inherent" in the de-
cided cases.2 3 ' Just how a judge comes to conceive of the "inherent"
rule is properly understood as a question of the context of judicial
discovery and not of judicial justification.23 2 As a theory of judicial
justification, the practical reasoning thesis has therefore as little to say
about the judicial perception of analogies as it does about the hunches
which may encourage judges to decide a case one way rather than
another. The issue for a theory of judicial justification is what war-
rants one analogy over the host of other comparisons which could
have been drawn.
On my view, the satisfactoriness of a comparison of cases relative
to the goals and purposes of the law warrants the court's reliance on

230. See supra text accompanying notes 169-75.

231. E. LEVI, supra note 165, at 1-2.
232. See supra text accompanying notes 42-43.
100 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

an analogy to infer a legal rule. Cases may be compared and con-

trasted in a number of ways, but advocates of the analogical thesis
hold that the court's choice of analogy should turn on the relevance or
the relative importance of the various similarities.23 3 In other words,
what distinguishes an acceptable from an unacceptable analogy is how
well a distinction in terms of one similarity or another would serve the
goals and purposes of the law, including the goals and purposes of a
precedential system. On this view, what justifies an analogy and the
extraction of a rule of decision will be the same considerations and the
same forms of inference that justify the derivation of a legal rule. The
rule of the analogy, once extracted, is applied to the case at hand, in a
manner that parallels the application of a rule of law to a particular
case.
On the second version of judicial analogies, comparing the pres-
ent controversy to past cases warrants some particular resolution of
the case at bar, without focusing on the intermediate inference of a
legal rule.2 34 On this second version, practical legal analogies could be
understood along the following lines:
A, B, and C all have characteristics x, y, and z.
D also has characteristics x, y, and z.
A, B, and C were (satisfactorily) resolved in manner S.
Therefore, let D be resolved in manner S.
In short, D should be treated in the same fashion as were A, B,
and C because D resembles those other cases in ways (x, y, and z) that
bear on the purposes of the rule resolving those cases.23 5 As in the
first version, the proper analogy is drawn in terms of the goals and
purposes of the law or the legal system. But, if judicial analogies are
drawn in order to serve the purposes of the legal system, then analo-
gies are essentially practical in nature; if it is satisfactory that cases A,
B, and C are resolved in manner S, and D is relevantly similar to the
other cases, then it would be satisfactory that D be handled in the
same manner.

233. See supra text accompanying notes 178-79.

234. See supra note 173 and accompanying text.
235. The structure in the text is a revision of the structure proposed in M. GOLDING, supra note
8, at 107. Although Golding indicates that he regards judicial analogies as practical arguments, in
some sense, his structure is cast in the form of propositions with truth-values. I have rewritten Gold-
ing's structure to cast it in the form that a modified account of analogies might take in practical reason-
ing. In the situation of competing analogies, Golding suggests that the choice between a premise "(iv)
y, z, . . . are q-relevant characteristics" and an alternative premise "(iv') s, t, . . . are non-q character-
istics" depends on considerations that sound eminently practical. See id. at Ill: "If, as usually will be
the case, premise (iv) rests on a different goal or right than premise (iv'), the judge should estimate
which goal or right is the more important goal or right: the more important one is the weightier one."
1985] PRACTICAL REASONING AND JUSTIFICATION 101

4. Counter-Instances and Defeasibility

Construing the application of a rule to a case as an inference in
the logic of satisfactoriness can also account for aspects of judicial ar-
gumentation that were problematic for the deductive theorist. The de-
ductive thesis, we saw, is committed to the position that legal rules
and legal decisions are true, and this creates certain problems for the
deductive theory.2 36 The rule "Contracts not to be performed within
one year are unenforceable unless in writing and signed by the party to
be charged" is not easily understood as a true assertion about the law.
Unless the Statute of Frauds is in fact followed in every case, the rule
would not be universally true. The practical reasoning thesis provides
a more natural understanding of the significance of legal rules than
does the deductive theory. If judicial justification is essentially practi-
cal, then a legal rule indicates what is satisfactory relative to certain
goals and purposes of the legal system. The judge's failure to instanti-
ate the Statute is not a real counter-instance, on this thesis. Rather
than falsifying the Statute, as is entailed by the deductive theory, the
judge's failure simply means that the Statute has not been applied. On
that occasion, the judge would fail to satisfy the goals and purposes of
contract law. But those goals and purposes would still obtain, and the
Statute would still be a satisfactory norm for the legal system.
Practical reasoning has yet another advantage over the deductive
theory. We noted, in Section IIIA, that a deductive theory could not
explain the distinction between the applicability of a rule and its war-
ranted application. 237 An important feature of the logic of satisfactori-
ness - its defeasibility - provides a useful explanation for this and
other aspects of judicial justification.
To illuminate this feature of the logic of satisfactoriness, recall the
example from Raz about the troubled car owner. We observed that he
could have chosen any one of several different possible rules as his
plan of action. So far as he had articulated his desires and goals in
that situation, various different rules might have been satisfactory.
Relative to other unstated desires, however, some of the possibilities
might well be unattractive to a driver in Raz's position. For example,
asking his friends to sign a release before letting them ride in the car
might have disserved another of Raz's aims: it might have offended his
friends. Because the rule about signing releases might have conflicted
with other purposes, that rule might not have been satisfactory relative
to an expanded set of goals which included some of the driver's other
goals and desires.

236. See supra text accompanying notes 157-58.

237. See supra text accompanying notes 131-44.
102 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

Whether a rule is satisfactory depends on the set of desires and

goals which one hopes to satisfy by following the rule. A rule, we have
noted, is not satisfactory simpliciter;rather, it is satisfactory relative to
a set of aims. Satisfactoriness' inherent relativity means that practical
reasoning is defeasible: if we add a further satisfactory premise to our
argument in the logic of satisfactoriness, we can no longer be sure that
the conclusion of the inference will remain satisfactory.2 3 Including
the further premise may defeat the satisfactoriness-preserving quality
of the inference, 23 9 even though it would not defeat the validity of a
deductive argument.
Ordinary and quite obvious instances of practical reasoning ex-
hibit defeasibility. Let us return to Raz's example again. The troubled
car owner anticipated that his colleagues might ask for a ride. His rule
against giving rides was formulated to enable him to deal adequately
with such requests relative to a given but plausibly incomplete set of
aims. If a colleague were to encounter an emergency and, in dire
straits, request a ride, Raz might well reconsider his rule in light of the
new situation. His colleague's emergency raises not just new facts, but
also calls into play desires not previously considered when formulating
the rule. The rule he originally formulated might no longer be satis-
factory, relative to the new set of desires. The rule might be defeated
because a new satisfactory premise had been added to his practical
argument.
In the ordinary non-legal case, our grasp of the desires and pur-
poses we seek to satisfy will be incomplete. As the situation is elabo-
rated, we may become cognizant of additional desires which we wish
to satisfy as well. In that case, we may discriminate among the various
rules, not by reference to the initial set, but by adding other satisfac-
tory premises. With respect to this revised set of aims some of the
candidate rules may be satisfactory, but perhaps not all.
To decide a case a judge must, in certain situations, choose
among competing and conflicting rules, each of which appears sup-
ported by the relevant authority, or among rival formulations of a
rule, each of which seems consistent with the decided cases. 2" If one
rule were acceptable and the other not, then the judge could reject one
and apply the other. But if, in the judge's assessment of the immedi-
ately applicable precedent and the goals of that area of the law, the
rival rules each appear tenable, then to justify his decision the judge
must articulate a basis for accepting one and rejecting the other. Prac-

238. See A. KENNY, supra note 7, at 92.

239. Id.
240. See, e.g., McIntosh v. Murphy, 52 Hawaii 29, 469 P.2d 177 (1970).
1985] PRACTICAL REASONING AND JUSTIFICATION 103

tical reasoning's defeasibility provides a useful explanation for the ar-

guments used by judges to differentiate among the competing rules or
rival formulations.
Where two rules are well supported by the relevant authority, the
apparent force of each rule indicates, on my view, that each alternative
is satisfactory relative to the goals and purposes of that area of the law.
Therefore, to differentiate among the rival alternatives, the judge may
point to other goals of the law or the legal system, or to other prece-
dent in a related but distant area of the law. When other satisfactory
fiats are added to the premises of the practical afgument it may be that
one of the alternatives is no longer satisfactory relative to the new,
expanded set. By appealing to further goals and values of the legal
system the judge can justify the choice of one rule over the other.
We should expect that reasoning to rules which are satisfactory to
the legal system's goals and purposes will be a complicated matter, for
it is doubtful that we have a firm grasp of all of the system's various
purposes. Consider a simple rule like the Statute of Frauds. The Stat-
ute is a valid rule in the law, on my view, if it is satisfactory relative to
the various purposes of the law of contracts and the legal system in
general. Relative to those purposes, the Statute's satisfactoriness is not
a prioriabsurd. The Statute arguably serves several aims: our desire to
channel enforceable promises into certain forms; our interest in secur-
ing appropriate evidence of the promise's existence and nature; and
our hope of impressing on the parties to the contract the importance of
the enforceable obligations they are assuming so that we can be sure
that they entered into the contract willingly.2 4 ' The Statute may also
make it easier for courts to enforce the regime of contract law by pro-
viding a simple precondition for enforceability of certain kinds of con-
tracts. Given the rule's possible usefulness within the law of
obligations, it may be satisfactory, relative to the goals and purposes of
contract law, to deny enforcement of a contract which is within the
Statute but unwritten.
But the Statute is not an unequivocal contribution to the law of
contracts. It contemplates that some freely made agreements will not
be enforced just because they lack a particular formality. The Statute,
therefore, sometimes seems like the vehicle for fraud rather than its
prophylactic.24 2 The Statute of Frauds might not be a satisfactory rule
if we add to our set of satisfactory premises the notion that we want to
prevent the avoidance of honest bargains by the mere manipulation of
a formal requirement. In that case, since the Statute contemplates

241. See, e.g., Llewellyn, What Price Contract?, 40 YALE L.J. 704, 747-48 (1931).
242. McIntosh v. Murphy, 52 Hawaii 29, 469 P.2d 177 (1970).
104 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

that sincerely made agreements will be unenforceable in the absence of

a writing, following the Statute may no longer be satisfactory. In
short, the Statute's satisfactoriness is problematic. It is plausible that
there are divergent and countervailing considerations which, if added
to our enumeration, might defeat the rule's validity.
We recognize, in fact, a long and complicated series of values at
work in the law of contracts, 24 3 and in the legal system more gener-
ally. 2"4 It is beyond the scope of this article to provide an account of
how each of these various expressions of the goals of the common law
might be explained by the practical reasoning thesis. 245 But we would
expect that a fully developed theory of legal reasoning will explain the
variety of standards that might be said to justify a rule and the differ-
ing ways in which they figure in judicial argumentation.2 4 6 With re-
spect to the problems of judicial justification, what is signal about
these various standards is the role they might play in establishing a
legal decision's satisfactoriness. By referring to some principle of the
law or to some rough assessment of public welfare we may expand the
set of goals and purposes which we hope to satisfy in the case deci-

243. See, e.g., Macaulay, Justice Traynorand the Law of Contracts, 13 STAN. L. REV. 812, 813-16
(1961).
244. See, e.g., R. DWORKIN, supra note 17, at 22-39, 82-130, 297-311.
245. Within current scholarship, one finds a plethora of accounts of the differences between prin-
ciples and policies. Compare H. HART & A. SACKS, supra note 21, at 151-65, with R. DWORKIN, supra
note 17, at 22-39. Compare also id. with id., at 82-130. Debates also flourish over the differences be-
tween rules and principles. Compare, e.g., R. DWORKIN, supra note 17, at 22-39, with Wellington,
Common Law Rules and ConstitutionalDouble Standards: Some Notes on Adjudication, 83 YALE L.J.
221 (1973). Since jurisprudence lacks agreement on the nature of these different norms and on the
differences among them, to show how the practical reasoning thesis could account for the principles
and policies would require, in effect, an account for each different version of these differences.
246. While jurisprudence lacks an agreed-upon explanation of these norms, it is clear that the
views of Ronald Dworkin on the nature of principles and policies, see supra note 244, are as central as
any. As I indicated in note 245, supra, Dworkin's views, while central, are not necessarily consistent on
many details. However, it is possible to describe, in very broad strokes, how the practical reasoning
thesis would account for certain salient features of Dworkin's view. In his first statement of the distinc-
tion between principles and policies, R. DWORKIN, supra note 17, at 22-39, Dworkin asserts that a
policy sets out "a goal to be reached, generally an improvement in some economic, political, or social
feature of the community .. " Id. at 22. A principle, he states, is "a standard that is to be observed,
not because it will advance or secure an economic, political, or social situation deemed desirable, but
because it is a requirement of justice or fairness or some other dimension of morality." Id.
Inasmuch as a policy states a goal to be achieved, it seems unproblematic to regard a policy as a
statement of a value in the law that is satisfactory to further that goal. A principle also states a value in
the law, and it seems similarly unproblematic to regard a principle as the expression of a value that can
serve as a premise in judicial justification through the logic of satisfactoriness. The difference between
principles and policies inheres in the way that they constrain judicial justification. Any decision that
furthers that policy would seem to be satisfactoriness-preserving, although that inference could be de-
feated by including some principle, or perhaps another policy of greater importance, in the premises for
judicial justification. A decision that fails to observe a principle cannot be satisfactoriness-preserving,
except in circumstances where some other principle of greater weight is also at issue, or circumstances
of great social danger. See id. at 95-96.
1985] PRACTICAL REASONING AND JUSTIFICATION 105

sion. 24 Including these legal standards can allow us to differentiate

among competing legal resolutions in the logic of satisfactoriness.
We recognize that not every applicable rule will be applied to
each case; the application of some rules might be unwarranted, their
applicability notwithstanding. This feature of judicial reasoning poses
difficulties for the deductive thesis,24 8 but it is understandable within
practical reasoning, for practical reasoning, unlike deduction, is
defeasible. 24 9
The defeasibility of practical inferences explains the distinction
between the applicability and warranted application of a rule. We
may say that a rule is applicable in the strict sense to a situation if the
situation exhibits those features that, in the formulation of the rule,
are the appropriate predicates for its application. In this strict sense,
then, any rule is applicable that correctly subsumes the situation; more
than one rule may be applicable. But, whether a rule's application is
warranted depends on the rule's satisfactoriness relative to some set of
goals and purposes of the law. Any situation will exhibit some fea-
tures in addition to those which make a particular rule applicable to it.
If these additional features call into question other goals and purposes
of the law, then applying that particular rule to the situation at hand
may frustrate those other purposes and goals. In that circumstance
the particular rule would be applicable, in the strict sense, but its ap-
plication would not be warranted.
When different rules, each applicable to the same situation, lead
to incompatible legal decisions, the court must apply one and reject
the other.25 ° Rejection of a rule does not, on my view, mean that the
rule is without vitality in the law. So far as it had been developed, the
rejected rule was satisfactory relative to some particular set of goals
and purposes of the law; it might still be satisfactory to apply that rule
to cases different from the one at hand. It may well be that the re-
jected rule, suitably qualified, continues to bind courts in other
situations.2"

247. This function of principles and policies is common to the accounts of Dworkin, supra note
17, and H. HART & A. SACKS, supra note 21.
248. See supra text accompanying notes 131-44.
249. See A. KENNY, supra note 7, at 92: "Theoretical deductive reasoning is not defeasible in the
sense that the addition of a premise cannot invalidate a previously valid inference: if a conclusion
follows from a given set of premises it can be drawn from any larger set containing those premises no
matter how many are added to the set." See also supra note 129.
250. It might be said, on such an occasion, that the rejected rule was not applicable. It would be
wrong, however, to conclude that the rejected rule was necessarily inapplicable in the strict sense. For,
the rule was applicable, but the court concluded that its application to this case was unwarranted
relative to the goals and purposes of the legal system.
251. When the judge distinguishes the case at hand from those situations to which applying the
 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

C. Evaluating a PracticalReasoning Theory

Practical reasoning, understood in terms of the logics of satisfac-
toriness and satisfaction, can explain the significant features of judicial
justification while satisfying the criteria for theory acceptability articu-
lated in Section II. The thesis that judicial argumentation is funda-
mentally practical reasoning provides criteria for recognizing valid
judicial inferences without committing us to unnecessarily aggressive
or unappealing positions about the judicial process. The practical rea-
soning thesis is therefore preferable to either of the two dominant
views. However, it must be acknowledged that the practical reasoning
thesis is not an unambiguous improvement over its deductive counter-
part. In particular, there are two aspects of this thesis which may be
thought troublesome.
Doubts may be felt, first, about the consequences of practical rea-
soning's defeasibility for a theory of law. In particular, one might
wonder whether any argument in the logic of satisfactoriness could
ever be conclusive or final. No matter how sure one might be of a
conclusion's satisfactoriness relative to a given set of goals, adding
some further desideratum could negate that conclusion's satisfactori-
ness relative to the revised set.252 It might therefore be argued that
unless it were possible to articulate a complete set of legal goals and
purposes, legal conclusions could be established only provisionally -
they seem liable to being overturned upon the introduction of some
further premise. Now, to defeat any particular conclusion would re-
quire both adducing some further premise and also demonstrating that
upon the addition of that further premise the conclusion in question
was not satisfactory relative to the revised set of goals and purposes.
But the possibility of defeasance seems to cast doubt on the acceptabil-
ity of every conclusion in the logic of satisfactoriness. Without a com-
plete set of legal goals and purposes, in short, it seems that no judge's
decision could be said to be justified once and for all.
It is, of course, arguable that the set of goals and purposes for
judicial justification is in fact complete. Indeed, completeness of legal
premises would be the consequence of an act-utilitarian theory of law:
the set of goals and purposes would consist only of the utilitarian

rejected rule would be warranted, he suggests a reformulation of the rejected rule. That is, he notes
those features of the case at hand which, relative to various goals and purposes of the legal system,
warrant recourse to the other rule. If the distinction which the judge draws is satisfactory relative to
the goals and purposes of the legal system, then we should expect that judges and lawyers will come to
employ the rejected rule, suitably reformulated so as to distinguish it from the type of situation to
which its application would be unwarranted.
252. Cf. W. SALMON, supra note 113, at 1 1-13 (noting the inconclusiveness of inductive reason-
ing). See also supra text accompanying notes 236-40.
1985] PRACTICAL REASONING AND JUSTIFICATION 107

maxim, and the satisfactoriness of any legal decision would be assessed

relative to that single concern. 253 However, this contention is unat-
tractive. Utilitarian theories of law are problematic, to say the very
least.254 In addition, while the issue seems beyond rigorous demon-
stration one way or the other, I share Kenny's doubts that the prem-
ises of practical reasoning could reasonably be supposed to be
complete.255
This objection to the practical reasoning thesis rests, to some ex-
tent, on an equivocation. Practical reasoning is conclusive. Valid con-
clusions follow from their premises in accord with the logics of
satisfactoriness and satisfaction. It is significant, however, that practi-
cal reasoning, unlike deduction, does not necessitate its conclusions.2 56
The conclusion of an argument in the logic of satisfactoriness is not
necessarily satisfactory just because the given premises are satisfac-
tory. But, the fact that deduction necessitates its conclusions gives rise
to an important objection to the deductive thesis - namely, that de-
duction cannot accomodate the distinction between the applicability of
a rule and its warranted application. That practical reasoning does
not necessitate its conclusions is, if anything, a virtue when we try to
explain the important features of judicial justification.
Moreover, while it is doubtful that a complete set of goals and
purposes for the law could ever be articulated, the lack of a complete
premise set for practical reasoning is of less significance for the decid-
ing judge than it might initially seem. The possibility that a further
desideratum might undercut the satisfactoriness of the judge's conclu-
sion is less troubling when we recall the legal community's expecta-
tions of judges. I noted in Section II that judges are not expected to

253. This is, as Ruth Marcus has noted, an "unregenerate" version of act-utilitarianism. See
Marcus, supra note 131, at 128.
254. See, e.g., R. DWORKIN, supra note 17, at 94-100, 232-38.
255. See A. KENNY, supra note 7, at 94:
The notion of a premise which is complete enough to prevent defeasibility while specific
enough to entail a practical conclusion is surely chimerical. Only in restricted contexts can
we even approach completeness in the specification of practical premises: we insist, for
instance, that the listing of contra-indications on a marketed drug should be not only accu-
rate but, within limits, complete . . . . It is a presupposition of truth-functional logic that
any proposition which has the value 'true' does not also have the value 'false'; but it is not
the case that a proposal for action which has the value 'good' may not also have the value
'bad'. . . . An argument which shows something to be a good thing to do in no way shows
that something incompatible with the conclusion it reaches may not also be a good thing to
do.
Cf H.L.A. HART, supra note 75, at 136: "It does not follow from that fact that ... rules have
exceptions incapable of exhaustive statement, that in every situation we are left to our discretion and
never bound to keep a promise. A rule that ends with the word 'unless ... ' is still a rule."
256. See supra note 127.
108 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

justify every legal decision from first principles.2 5 7 At least within the
constraints of a system of stare decisis, judges are permitted if not ex-
pected to take some issues for granted. Indeed, I also argued that if a
judge did try to decide every possible legal ramification of the contro-
versy at hand, his resolutions of those more remote questions would
likely be regarded as mere dicta. In sum, the objection to practical
reasoning's defeasibility seems largely beside the point: the decisions of
an individual judge are not treated as justified once and for all but,
rather, are open to review and reconsideration depending on future
developments of the law.
Second, the practical reasoning thesis commits us to a position on
logical form that some will find problematic: it denies that legal state-
ments are true or false. The extent to which this consequence of the
practical reasoning thesis counts against the thesis will depend on the
reasons why a theorist might hold that legal statements bear truth val-
ues. It is possible, for instance, 'that the principal attraction to the
notion that statements of law are true or false has been jurisprudence's
interest in the deductive thesis. In that case, the superiority of practi-
cal reasoning as an account for judicial reasoning should obviate the
commitment to truth values.
Beyond our interest in the deductive thesis, it may be that legal
theory has an independent stake in the metaphysical status of legal
statements. If so, then that fact would count against the view of judi-
cial argumentation here presented. We must be careful, however, to
make sure that our theory of law really commits us to the truth or
falsity of legal statements. For, it is plausible that some positions in
legal theory have characteristically been framed in terms of the truth
of legal statements even though the position is compatible with other
claims about the nature of statements of law.
Consider, for example, the prospect of a natural law theory. One
might advance a natural law theory which simultaneously holds that a
legal proposition is law properly so called if and only if it is consistent
with the dictates of morality, and that moral norms are necessarily
true or false. It would follow that legal propositions are true or false.
On this version of a natural law theory, one could not easily adopt the
view that legal reasoning is practical reasoning. But other formula-
tions of a natural law theory are available which do not commit us to
the position that legal statements are true or false. We might hold,
instead, that statements of legal norms are law, properly so called,
only if those statements accord with certain moral norms, but avoid
any position on the status of moral norms. On this formulation the

257. See supra text accompanying note 40.

1985] PRACTICAL REASONING AND JUSTIFICATION 109

practical reasoning thesis would require no substantial reconsideration

of our commitment to natural law. Moral norms may be binding even
if one does not also view them as true or false.
The significance of the metaphysical claim about legal statements,
in other words, should itself be regarded as an unsettled issue. If we
decide that our best theory of law requires a commitment to legal
statements bearing truth values, then we should rethink our under-
standing of judicial argumentation, but we should be sure that our best
theory of law does indeed require that commitment.
That master of the epigram, Holmes, shaped the attitudes of
many theorists about the problem of legal reasoning when he pro-
nounced: "The life of the law has not been logic: it has been experi-
'
ence." 258 In considering the possible models for legal reasoning, it
may commonly be felt that no formal "logic" could capture the
processes by which the legal system adapts and responds to the "felt
necessities of the time" that supposedly determine the rules of law
which govern us. 259 My contention is that practical reasoning, under-
stood in terms of the logics of satisfactoriness and satisfaction, is a
distinct form of reasoning. We rely on it in ordinary reasoning: we
make decisions about what to do in light of our needs and purposes.
In one sense, at least, practical reasoning dissolves the Holmesian di-
chotomy: it provides a logic of the "felt necessities." Practical reason-
ing may thus serve to bridge the gap which Holmes posits between
logic and experience. In any event, practical reasoning is a model for
judicial justification. It explains the relevant features of judicial argu-
mentation better than does either the analogical or deductive thesis,
and it avoids the problematic consequences posed by the deductive
thesis. It is, in sum, a. better theory.

V. PRACTICAL REASONING AND LEGAL THEORY

I have argued that practical reasoning provides a better theory of
the nature and validity of judicial justification than does either the an-
alogical or deductive thesis. As part of the argument, I have con-
tended that we should aim for an account of argumentation that is not
dependent on any particular theory of law. Judicial justification is im-
portant enough a topic in jurisprudence that we should aspire to ex-
plain it as an independent concern. Developing a theory of judicial
justification is, however, only one part of the larger task of investigat-
ing the nature of law and the legal system. Issues in philosophy of law
are commonly thought to depend on an understanding ofjudicial justi-
258. O.W. HOLMES, supra note 93, at 1.
259. Id.
110 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

fication. Moreover, the worth of a theory of justification can be mea-

sured, in part at least, by the extent to which the theory illuminates
other questions in jurisprudence. In this last section, I will explore
how the practical reasoning thesis helps us understand other problems
in legal theory.
One longstanding controversy has been the debate over the exist-
ence of "gaps" in the law. Are there legal claims which we cannot
determine to be either valid or invalid as a matter of law? Legal theo-
rists of varying stripes have taken the position that a mature legal sys-
tem must be without gaps: for every possible controversy the system
should generate one right answer. It is sometimes asserted that Black-
stone held this view, 2" and great jurists have relied on this claim in
their opinions. 26 '
The question of the law's gaplessness is related to other jurispru-
dential concerns. If there are no gaps, unease might be felt at the pros-
pect of judges deciding cases originally or creatively. Judicial
originality seems to mean that the legal obligations of the parties have
been reshaped by the court's decision, and judge-made changes in the
law, it is suggested, raise substantial problems of unfairness in the law.
Among other things, when the law changes, parties whose cases are
resolved under the new law are treated differently than those whose
disputes were adjudicated under the old law. Moreover, the parties
may have relied on the older standard at the time they decided to act.
Conversely, if judicial law-making is a proper part of the task of decid-
ing cases, then it would seem that the law is incomplete. Any case
that falls in one of the gaps would therefore seem to be decided by the

260. See, e.g.,Linkletter v. Walker, 381 U.S. 618, 623 (1965).

261. Consider, for example, the remarks of Chief Justice Shaw in his opinion for the court in
Norway Plains Co. v. Boston & Maine R.R., 67 Mass. (I Gray) 263, 267-68 (1854):
It is one of the great merits and advantages of the common law, that, instead of a series of
detailed practical rules, established by positive provisions, and adapted to the precise cir-
cumstances of particular cases, which would become obsolete and fail, when the practice
and course of business, to which they apply, should cease or change, the common law
consists of a few broad and comprehensive principles, founded on reason, natural justice,
and enlightened public policy, modified and adapted to the circumstances of all the particu-
lar cases which fall within it. ...
Another consequence of this expansive character of the common law is, that when new
practices spring up, new combinations of facts arise, and cases are presented for which
there is no precedent in judicial decision, they must be governed by the general principle,
applicable to cases most nearly analogous, but modified and adapted to new circumstances
by considerations of fitness and propriety, of reason and justice, which grow out of those
circumstances. The consequences of this state of the law is, that when a new practice or
new course of business arises, the rights and duties of parties are not without a law to
govern them; the general considerations of reason, justice, and policy, which underlie the
particular rules of the common law, will still apply,' modified and adapted, by the same
considerations, to the new circumstances.
1985] PRACTICAL REASONING AND JUSTIFICATION 111

judge, not as a matter of law but as a matter of personal whim or

prejudice.
In their arguments about the existence of gaps, theorists have fre-
quently relied on claims about judicial argumentation. H.L.A. Hart,
for example, has advanced the claim that on occasion the law simply
"runs out." On such occasions, he contends, judges have discretion to
decide the case at hand, and their decisions must be regarded as origi-
nal, if not necessarily creative.26 2 Hart implies that if there is no law
in such a case, there cannot be, as a matter of law, a right answer. The
2 63
judge must make new law if he is to decide the case at all.
Ronald Dworkin, on the other hand, has been a vigorous advo-
cate of the position that in a well-developed legal system there is, in
principle, a right answer to each legal controversy.2 64 In a variety of
writings, he has advanced a skein of related contentions about the
existence of legal gaps. First, he has been a diligent and devoted critic
of what he calls the "no-right-answer" thesis - the claim that there
are gaps in the law. In some of his writings, he challenges the coher-
ence of certain arguments which might be advanced in support of the
existence of gaps.2 65 Other of Dworkin's contentions aim at support-

262. See H.L.A. HART, supra note 75, at 132.

263. Id.
264. Professor Dworkin has not taken a clear position on the logical nature of judicial justifica-
tion. Various remarks indicate that he is in sympathy with the deductive thesis. He is apparently
committed to the notion that statements of law are true or false. See, e.g., R. DWORKIN, supra note 17,
at 283; Dworkin, No Right Answer?, in LAW, MORALITY AND SociETY: ESSAYS IN HONOUR OF
H.L.A. HART 58, 62-65 (P. Hacker & J. Razed. 1977). And, on at least one occasion, he demonstrates
his apparent agreement with the deductive thesis: "In easy cases legal rights can be deduced, in some-
thing close to a syllogistic fashion, from propositions reported in books that are available to the public."
R. DWORKIN, supra note 17, at -337. However, there is too little basis on which to claim that these
remarks represent a full-fledged commitment to deductive logic as the nature of justification.
265. Dworkin has attempted to defuse Hart's view of judicial discretion to decide hard cases. By
including, as a part of what must be counted as "law," principles and policies which might guide the
judge in deciding cases even where there are no rules, Dworkin has advanced the notion that the judge
is never without law. Courts have therefore no real discretion, at least not of the sort that might imply
that the judge makes new law. See R. DWORKIN, supra note 17, at 14-80.
Moreover, in two related essays he has sought to demonstrate that certain arguments in favor of
the no-right-answer thesis are logically incoherent. See id., at 279-90; Dworkin, supra note 264. He
offers a variety of distinctions in an effort to make this second point. First, he distinguishes two differ-
ent versions of the no-right-answer thesis. "On the first version, two lawyers debating whether the
plaintiff has a right to a decision, or whether, on the other hand, the defendant has a right to win, may
both be wrong, because the right answer is that neither party has a right to win. On the second version,
neither lawyer is right, but neither is wrong either; what each says is, for some reason, neither true nor
false." R. DWORKIN, supra note 17, at 331.
He further distinguishes between two different standpoints from which the conclusion that some
issue has no right answer - in the sense of the second version of the claim noted above - might be
advanced. There are claims that there is no right answer which are made, as Dworkin puts it, "within
the enterprise," and other claims that law lacks a right answer which are made "outside the enterprise."
Claims made within the enterprise are claims about the relative reliability of three different legal resolu-
112 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

ing directly the proposition that if settled law is sufficiently complex,

only one legal answer will be tenable. 26 6 Finally, he advances a related
claim about the obligations of a judge in a mature legal system. In
their deliberations, he argues, judges should assume that there is a
right answer and should undertake in every case to find it.26 7 The
practical reasoning thesis can help clarify this dispute about the pur-
ported completeness of the legal system. In particular, it can help us
evaluate some of Dworkin's claims.
Dworkin's principal arguments in favor of the right answer thesis
rely on his theory of judicial decision making, which he expounds in
his article "Hard Cases."' 26 On this theory, the right answer in a par-
ticular controversy follows from the best available constitutional the-
ory of the legal system in which the controversy arises. 2 6 9 No actual
jurist, he acknowledges, could be expected to accomplish the intellec-
tual feat of building such a theory, so he posits Hercules, a mythical
judge who is capable of this labor. 270 Assuming Herculean capabili-
ties, Dworkin articulates a test for the truth of a legal proposition: "A
proposition of law may be asserted as true if it is more consistent with
the theory of law that best justifies settled law than the contrary prop-
27 1
osition of law."
Dworkin uses the Hercules story to support two important pro-
positions about the existence of gaps. First, in a mature legal system
there will be "enough" law so that only one theory will justify settled
law, and only one answer to a legal controversy will be consistent with
that best theory.27 2 Second, all possible answers to a legal controversy

tions of a particular controversy: (i) plaintiff has a right to win; (ii) defendant has a right to win; and
(iii) there is no right answer to the dispute. A claim made within the enterprise that there is no right
answer "is a judgement of the same character, and is equally fallible, as either of the other available
alternatives." Id. at 285. Claims made outside the enterprise that there is no right answer are of an
entirely different character. They are "external, critical" claims about the enterprise of judging and our
standards for the validity of judicial decisions. Id.
266. See infra notes 268-72 and accompanying text.
267. See infra notes 276-78 and accompanying text.
268. R. DwoRKIN, supra note 17, at 81.
269. Id. at 105-23.
270. Id. at 105.
271. Id. at 283.
272. See id. at 286: Suppose "that the legal system these judges administer is very advanced, and
is thick with constitutional rules and practices, and dense with precedents and statutes. The antecedent
probability of a tie is very much lower; indeed it might well be so low as to justify a further ground rule
of the enterprise which instructs judges to eliminate ties from the range of answers they might give.
That instruction does not deny the theoretical possibility of a tie, but it does suppose that, given the
complexity of the legal materials at hand, judges will, if they think long and hard enough, come to think
that one side or the other has, all things considered and marginally, the better of the case." See also
Dworkin, supra note 264, at 84: "For all practicalpurposes, there will always be a right answer in the
seamless web of our law." (Emphasis added.)
Professor Dworkin's conception of a gapless mature legal system bears some resemblance to the
1985] PRACTICAL REASONING AND JUSTIFICATION 113

can be ranked according to their consistency with the best theory of

law for the legal system. Various features of practical reasoning cast
doubt on these propositions.
Let us suppose that judges are equipped with a single theory
which best justifies settled law. On the practical reasoning thesis, this
theory will invoke a set of goals and purposes of the law relative to
which past decisions are satisfactory. The judge should justify his de-
cision in any controversy in terms of that statement of legal goals and
purposes. In Section IV I noted that, relative to any statement of the
goals of the law, more than one rule or holding could be satisfac-
tory.27 3 Of course, as between competing rulings, we can discriminate
some from the rest by appealing to further goals and purposes: by ad-
ding those goals to the premise set, some of the rulings which were
satisfactory relative to the original set of premises may be defeated.
This means that, for any particular controversy, the court may be able
to reach a decision. But, as a general claim about the legal system,
there is no more reason to assume that only one possible ruling will be
satisfactory relative to the legal system's purposes and goals than there
is to assume that one plan of action will be satisfactory for an individ-
ual. In short, the practical reasoning thesis suggests that we cannot
assume that only one answer to a controversy will be consistent with
the best theory for a legal system. There may well be more than one
right answer.
Further, practical reasoning's defeasibility casts doubt on any ar-
gument for the law's gaplessness that depends, as does Dworkin's
claim, on the effectiveness of judicial reasoning to rank all possible
legal answers in terms of their consistency with the best theory of the
legal systems.274 On Dworkin's view, the ranking is determined by the
answers' relative consistency with a theory that justifies settled law.
But the practical reasoning thesis suggests that further, valid premises
might always obtain. We could never preclude the possibility that,
upon considering further valid premises, settled law might become un-

logician's notion of a complete formal system of logic. A formal system of logic is complete if and only
if every logically valid statement of logic is a theorem of that system. See, e.g., G. HUNTER, supra note
123, at 92-95. Dworkin's apparent kinship with deductive theorists of judicial justification, see supra
note 264, together with his rejection of the no-right-answer thesis, has led some commentators to sug-
gest that Dworkin conceives of a well-developed legal system as comparable to a complete logical sys-
tem. See, e.g., Farago, Intractable Cases: The Role of Uncertainty in the Concept of Law, 55 N.Y.U. L.
REV. 195, 220-29 (1980). I see no good independent reason to suppose that completeness, or anything
comparable, could be proven of the legal system. While it is possible that a proponent of the deductive
thesis might be seduced into thinking that completeness of the legal system is an attainable goal, I doubt
whether such a view is properly attributable to Dworkin.
273. See supra text accompanying notes 222-29.
274. This part of Dworkin's argument has been challenged on other grounds by the late John
Mackie. See Mackie, The Third Theory of Law, 7 PHIL. & PUB. AFF. 3, 9 (1977).
 UNIVERSITY OF COLORADO LAW REVIEW [Vol. 57

settled. 5 So, it is doubtful that even Hercules could state settled law
with any finality. We would therefore have to expect that in at least
some legal controversies there might be no effective procedure for gen-
erating a single right answer.
Another strand of Dworkin's writing urges the right answer the-
sis not just as the correct claim about the existence of gaps, but also as
a standard for evaluating judicial performance.2 1 6 It is Dworkin's
view that judicial decisions enforce antecedently existing rights of the
parties. As a consequence of the political theory which justifies the
institution of judging, courts are bound by a general requirement of
articulate consistency in their decisions. At least in the case of Hercu-
les, this requirement obligates him to ascertain the force of available
legal authority. He has fulfilled this obligation only when he has con-
sidered the bearing on a particular case of any and all relevant author-
ity.2 77 Thus, the right answer thesis commends itself as a prophylactic
against the prospect that judges, when confronted with a welter of
conflicting authority, might fail to enforce the pre-existing rights of
the justified party. In short, Dworkin argues, we may well hold judges
in a mature legal system accountable for deciding cases as if there is a
27
right answer, even though there is not. 1
The practical reasoning thesis is, to some extent at least, consis-
tent with Dworkin's argument. As we noted, a judge may appeal to
further goals and purposes of the legal system in an effort to discrimi-
nate among competing rules and decisions. Practical reasoning's de-
feasibility defies the prospect of a complete premise for justification
and hence undermines the general claim that there are no gaps. But,
for most controversies, it may well be that only one answer is satisfac-
tory relative to a suitably expanded set of legal goals and purposes.
That is, it may well be that a judge, if he thinks long and hard enough,
will be able to justify the choice of one claim but not the other by
reference to some goal or purpose.
Of course, to acknowledge the possibility that one ruling can usu-
ally be justifed over its rivals does not establish that a single right an-
swer will necessarily obtain. The practical reasoning thesis suggests
that no judge should expect a priori that some particular controversy

275. Cf. A. KENNY, supra note 7, at 92-94.

276. See R. DWORKIN, supra note 17, at 286; Dworkin, supra note 264, at 84. In this connection,
it is worth noting that Hercules' responsibilities in deciding the case before him are prior, in Dworkin's
development of the thesis, to any claim about the necessity of a right answer. See R. DWORKIN, supra
note 17, at 113-23. The right answer thesis seems to depend on the claim about judicial obligations; we
can agree with the latter without committing ourselves to the former.
277. R. DWORKIN, supra note 17, at 115-23.
278. Id. at 286.
1985] PRACTICAL REASONING AND JUSTIFICATION 115

will be undecidable: as between any two competing rules or decisions,

judges may well be able to establish one over the other. This does not
guarantee, however, that there is only one right answer.
It is always possible that there are other arguments in favor of the
right answer thesis, and the practical reasoning thesis may cast no
light on those other positions. But, where a controversy depends on a
claim about judicial justification, the superiority of the practical rea-
soning thesis means that we may begin with that understanding of ju-
dicial argumentation to help resolve our broader concerns.