In Defense of Tracing

Matthew T. Flummer
PENULTIMATE DRAFT: FINAL DRAFT FORTHCOMING IN THE JOURNAL
OF PHILOSOPHICAL RESEARCH
Abstract:
John Martin Fischer and Mark Ravizza (1998) claim that reflection on tracing cases
partially motivate the idea that moral responsibility is essentially historical. Fischer and
Ravizza contend that in cases in which an agent is morally responsible for an action
despite lacking the right kind of control, we must appeal to tracing. In a pair of recent
papers, Andrew Khoury (2012) and Matt King (2014) have argued that tracing is not a
necessary feature of moral responsibility. King argues that in tracing cases, the agent's
responsibility can be fully explained either by appeal to either recklessness or by
negligence. Khoury notes that the agent in a tracing case is claimed to be responsible for
the consequences of his action even though he does not satisfy the control condition at
the time of action. But he argues that agents cannot be responsible for the consequences
of their actions. If no one is morally responsible for consequences, then tracing is
unnecessary. In this paper, I will argue that both Khoury and King's respective arguments
fail to show that tracing is not a necessary part of a successful theory of moral
responsibility.

I. INTRODUCTION
Recently there has been a lively debate over whether moral responsibility is
essentially historical. Some properties are dependent upon their histories in some way these properties are historical properties. Others are not - these are current time-slice, or
snapshot properties. Historicists claim that facts about an agent's past are essential to the
agent's moral responsibility. So two people with qualitatively identical mental states at
the time of action might differ with respect to their respective moral responsibility for
these actions. Reflection on manipulation cases might motivate the idea that moral
responsibility is essentially historical (see Mele 1995 and Fischer and Ravizza 1998).
John Martin Fischer and Mark Ravizza claim that reflection on other types of cases do so
as well.
Consider two drunk drivers, Drunk Dale and Wasted Wally:

Drunk Dale: Dale went to a bar to blow off some steam after work. Several of his
friends showed up and, having a good time with his buddies, Dale became very
drunk. On the way home, he lost control of his car and swerved onto the sidewalk
and hit and killed a pedestrian.

Wasted Wally: Wally made some enemies extremely angry with him recently. To
get even with him, they decided to kidnap him, inject alcohol directly into his
bloodstream, and put him behind the wheel of a car. On the way home, he lost
control of his car and swerved onto the sidewalk and hit and killed a pedestrian.

By hypothesis, neither Dale nor Wally has the control necessary to be directly morally
responsible for killing the pedestrian at the time of the accident.1 But it also seems that
Dale is responsible for killing the pedestrian and Wally is not. Fischer and Ravizza
contend that in cases in which an agent is morally responsible for an action despite
lacking the right kind of control, we must appeal to tracing. That is, if an agent is to be
responsible for an action over which she lacks the requisite control, then we must be able
to trace back to a prior action over which she did have the requisite control. Here is how
they characterize tracing,

When one possesses control over one's act at t1, and one can reasonably be
expected to know that so acting will (or may) lead to the agent lacking control
over her action at t2, one can be held responsible for so acting at t2 (Fischer and
Ravizza 1998: 50-51)

Fischer and Ravizza note that this is just one way to formulate a tracing principle.
Tracing, however, has not been without detractors. In a pair of recent papers,
Andrew Khoury and Matt King have argued that tracing is not a necessary feature of
moral responsibility (Khoury 2012 and King 2014). King argues that in typical cases that
appeal to tracing, the agent's responsibility can be fully explained either by appeal to
recklessness or by appeal to negligence. Since the agent's responsibility can be explained
in one of these ways, tracing is superfluous. Khoury notes that proponents of tracing
often claim that the agent in the tracing case is responsible for the consequences of his
action even though he does not satisfy the control condition at the time of action. But he

argues that agents cannot be responsible for the consequences of their actions because
persons are only responsible for their subjective mental states and the actions that flow
from those mental states. Consequences don't necessarily flow from an agent's mental
states, therefore agents are not responsible for consequences. And if no one is morally
responsible for consequences, then tracing is unnecessary.
In this paper, I will argue that both Khoury and King's respective arguments fail to
show that tracing is not a necessary part of a successful theory of moral responsibility.
First, I will argue that King's proposal fails. I will do this by demonstrating that it
depends on a particular theory of action individuation and that this theory does not show
that tracing is unnecessary. Next, I provide reasons to doubt the soundness of Khoury's
argument. I will then argue that even if we grant Khoury's argument about responsibility
for consequences, he has still failed to show that tracing is unnecessary because tracing is
still needed for cases of impaired decisions.

II. RECKLESSNESS AND NEGLIGENCE

King claims to offer two strategies to show why Drunk Dale is responsible for
running over the pedestrian that do not depend on tracing. He claims that Dale's
responsibility can be exhaustively modeled on cases of recklessness or exhaustively
modeled on cases of negligence (King 2014: 465-466). He asks us to consider the
following cases:

Suppose a soldier wants to incapacitate a group of enemies sheltered in a pillbox.

So, the soldier lobs a grenade into the pillbox, incapacitating the enemies. We can

imagine that for several seconds after he has thrown the grenade, indeed, for a few
moments after the grenade has landed in the pillbox, he still hasn't achieved his
goal. Only when the grenade finally explodes does he incapacitate them.
Similarly, we can imagine a different soldier with the same goal, incapacitating a
group of enemies, but this time he is being pursued by them. So, the soldier sets a
mine along his path, intending for his pursuers to trip it, and thus be incapacitated.
Surely enough, his ploy works. However, the mine isn't triggered for several
minutes after he sets it. He is a good half-mile away when it finally goes off and
the enemies are incapacitated. A final example: suppose another different soldier
is planning to rescue some captured comrades. His plan involves incapacitating
several groups of soldiers who are positioned among the guardhouses and towers.
So, he goes about setting explosives at each structure, timed such that they go off
all at once. He is a safe distance away when they do explode, and he successfully
incapacitates all the enemies. But it has been nearly an hour since he set the first
charge (caution required he take his time). (King 2014: 466)

King claims that there are three interesting features of each case. First, all three soldiers
are responsible for incapacitating the enemies. Second, tracing isn't needed because each
soldier satisfies the control necessary for being morally responsible for incapacitating the
enemies (in whatever way one wishes to spell out that control). King claims this is
because each soldier's actions are paradigmatic cases of intentional action. Finally, all
three cases show how an agent can control the outcome in question even though he is
both physically and temporally removed from the outcome. He explains,

Nevertheless, this time lapse does nothing to modify the account we need to give
in order to explain the soldiers' responsibility, nor does the fact that each soldier
fails to exert any direct physical control over the outcome at the moment it
occurs...We don't need a special tool, like tracing, to explain responsibility in such
cases...In all of the above cases the soldier is clearly responsible for incapacitating
the enemies, an action over which they seemingly exercise control. Yet they are
(to various degrees) removed from where the action is when the outcome actually
obtains. (King 2014: 466-467)

King points out that though the soldiers in the above cases act intentionally in setting the
explosives, drunk drivers don't typically intentionally run over pedestrians. This is so in
the case of Drunk Dale above. Dale obviously does not run over the pedestrian
intentionally.
King claims that if we consider a different soldier case, we can draw a relevant
parallel:

Consider a soldier who lays a mine to incapacitate his pursuers, only to have it
detonated by a child from a nearby village who was wandering through the forest.
(King 2014: 467)

The soldier does not intentionally harm the child. But perhaps he could have reasonably
foreseen that his laying of the mine might have caused such an unfortunate consequence.

Thus this case is relevantly similar to a drunk driving case in which the drunk driver does
not intentionally run over a pedestrian; but he could have reasonably foreseen that his
driving drunk might have caused harm to someone.
King claims that there are two interpretations of Drunk Dale's case:

In both these sorts of cases, then, the first step involves distinguishing between
two possible interpretations of [Dale's] case. What matters here is whether [Dale]
actually foresaw that getting drunk might lead to him injuring others on the road
or, instead, whether he should have foreseen that this was the case (but didn't). If
he actually considers the possibility of future harm at the time of his continued
drinking, then his case can be usefully modeled on recklessness, which I argue
involves exercising control in much the same way as the soldiers' cases above
(involving intentional action). Call this the reckless interpretation.
If [Dale] doesn't consider the possibility of harm, but should have, then we
can model his case on cases of negligence. Call this the negligence interpretation.
(King 2014: 467-468)

Either Dale foresaw the bad consequences of his getting drunk or he didn't. If Dale
foresaw that getting drunk and getting behind the wheel might lead to disaster, then his
case can be modeled on cases of recklessness. If, however, Dale did not foresee such an
outcome but should have, then his case can be modeled on cases of negligence.
Before going on it is important to note two points. First, King does not mention that
theories of responsibility that rely on tracing distinguish between direct responsibility and

indirect responsibility. Direct responsibility refers to responsibility that results from the
agent satisfying control and epistemic conditions at the time of action. Indirect
responsibility refers to responsibility for an action (or consequence) even though the
agent failed to satisfy the control and/or epistemic conditions at the time of action. The
agent's responsibility results from tracing back to a time at which the agent did satisfy the
control and epistemic conditions. With this distinction in mind, the tracing theorist can
claim that Dale is morally responsible for running over the pedestrian despite not
satisfying the control condition when he does so. The key will be the point in time at
which Dale satisfies the control condition. It seems that King's argument implies that
Dale is directly morally responsible for the accident. But the proponent of tracing claims
that Dale is only indirectly morally responsible for the action because, by hypothesis, he
does not exercise suitable control at the time of action (For instance, Dale is not reasons
responsive at the time of the accident). He is intuitively morally responsible for the
action; therefore, we must trace back to a point in time at which he did have the requisite
control.
Second, in order to be able to claim that Drunk Dale is directly morally
responsible for killing the pedestrian, King must appeal to a specific way of individuating
actions. His argument depends on a description of drunk driving cases in which the drunk
driver's running over the pedestrian is a temporally extended action that begins with his
drinking and ends with running over the pedestrian. To illustrate this, King asks us to
consider the case of Jill:

Suppose Jill pops a pill she knows makes her likely to become an unconscious
zombie who suffers random and serious muscle spasms 30 seconds after taking it.
She then walks into a china shop. While she certainly lacks the requisite control
for the duration of the pill's effect, she is also certainly responsible for all the
china she breaks due to her spasms. (King 2014: 470)

Proponents of tracing will agree with King that Jill is morally responsible for breaking
the china. But it is also clear that she lacks the requisite control at the time of her
breaking the china. Normally, a proponent of tracing will claim that she is not directly
responsible for breaking the china because of this. But, they will argue, she is indirectly
responsible for it because we can trace back to a prior action at which she did have the
requisite control. King claims that this is not the case. He states,

But this needn't be because she is responsible for some prior action, the taking of
the pill, which is causally related to the absence of certain mental states which led
to the later harm. It is rather that she is already related to the later harm at the
time she takes the pill. Her foreknowledge of its effects coupled with her beliefs
about the shop and the likely harm she would cause in it are sufficient for
explaining her responsibility. She acts recklessly by entering the shop after having
taken the pill, knowing as she does the risk of harm she'll be running. (King 2014:
470, emphasis added)

According to King, Jill performs a single temporally extended action that begins with her
taking the zombie pill and ends with her breaking china. He states, If all of the
soldiers ... and all manner of temporally-extended-action-performing agents exercise
sufficient control to be responsible, then Jill does as well. And if Jill is responsible in
virtue of exercising control, so is [Dale] (King 2014: 471).
Proponents of tracing claim the case of Drunk Dale involves two separate actions:
running over the pedestrian and getting drunk. Dale seems to be morally responsible for
both actions. But he also does not have the relevant control over running over the
pedestrian. But the tracing theorist can explain why he is morally responsible for running
over the pedestrian, when he lacks control, by tracing back to the prior action getting
drunk, which does not lack control. King's strategy, on the other hand, is to claim that
Dale actually has control over running over the pedestrian, so tracing is unnecessary.
Dale has control over running over the pedestrian because his killing the pedestrian and
his getting drunk are a part of a single, temporally extended action. And since Dale has
the requisite control over his getting drunk, then he likewise has control over his killing
the pedestrian.2 With this in mind, I now turn to the details of King's proposal.
The recklessness interpretation requires the following: The agent in question must
actually think of the possibility of harm as a result of his action before the action takes
place. Consider this in regards to Drunk Dale. When Dale began drinking, if he actually
thought about the possibility of harm because of his continued drinking, then his
responsibility for later running over the pedestrian fits the recklessness interpretation.
King states,

But should he injure someone, (as he in fact does), we need not trace back to his
responsibility for continuing to drink. If the soldier controls incapacitating the
enemies when the charges he set go off, it is plausible to suppose that [Dale]
exercises the requisite control over the possible, considered harms. There is no
need to trace backward anymore than in ordinary action; the structure required
extends control forward even when [Dale] is no longer able to directly affect the
course of the car. (King 2014: 468)

King claims that had Dale considered the possible bad outcomes, then he exercises
control over possible bad outcomes, and therefore acted knowingly when he caused the
accident. For instance, Dale cannot claim that he is not responsible because he didn't
mean to hurt anyone. But, King contends, he did not mean not to hurt anyone either.
The proponent of tracing might object at this point that it is implausible to assume
that Drunk Dale exercises control at the time of the accident. But King claims that
reckless actions don't lack the requisite control. He asks us to consider the following case:

Rick thinks to himself that if he exceeds the speed limit on slick roads, swiftly
switching lanes, he might well cause an accident. But, so be it, he does so anyway.
Sure enough, he causes an accident, hitting a sedan and injuring those inside.
(Ibid)

King admits that though this is not an intentional action it is close enough to an ordinary
case of action. He states,

Again, on the assumption that responsibility requires control, and since Rick is
surely responsible, he plausibly satisfies the control condition over the accident. If
Rick is responsible in virtue of exercising control, and if [Dale's] case mirrors
Rick's, then I conclude that [Dale], too, is responsible in virtue of exercising the
same control. (King 2014: 468)

To sum up so far: King claims that if Dale thought about the possible harms from
drinking too much and driving, then his case can be modeled on cases of recklessness.
Cases of recklessness don't lack the control necessary for moral responsibility. Therefore,
Dale controls possible harms from drinking too much and driving and is therefore
morally responsible for doing so. Since Dale doesn't lack control, tracing is unnecessary.
King claims that if we construe the drunk driver's action as a temporally extended action
and the drunk driver thought about the possible harms that could come about because of
his his action, then we won't need to appeal to tracing because cases of this type can be
fully explained with the recklessness interpretation.
King claims that if Dale does not think about the possible harms that would come
about by his actions, then we can explain his responsibility with the negligence
interpretation. To illustrate the negligence interpretation, King presents the case of
Negligent Norma: If Norma is distracted while backing up out of her driveway and fails
to check her mirrors, and as a result she hits a child, she is responsible for hitting the
child because she was negligent. Norma fails to take due care to limit the risk of harm her
backing out poses to those around her by failing to look behind her (King 2014: 473).

King claims that the explanation for why Norma is responsible for hitting the child in this
case need not appeal to tracing and if this is so, then tracing won't be needed to explain
Dale's case either.

III. TRACING FOR TEMPORALLY EXTENDED ACTIONS

In this section, I will argue that King's proposal does not show that tracing is
unnecessary. First, the case of Norma is not analogous with Dale's case. To see this,
consider what King calls the canonical schema for tracing:
1.
2.
3.
4.

Agent A did something x without control at t1;

A performed action y with control at t0;
A's action y at t0 made A have property P; and,
P explains why A lacked control at t1 and why A did x.

Dale's case fits this schema perfectly. Dale ran over the pedestrian without control at t1.
He drank too much with control at t0. His drinking too much at t0 made him drunk. And
being drunk explains why Dale lacked control at t1 and why he ran over the pedestrian.
Norma's case does not fit the schema because she did not lack control over her backing
her car up - or at least from the vignette King offers, we have no reason to doubt that she
had control. So even if King's negligence interpretation can explain why Norma is
morally responsible for harming the child, it will not help to explain why Dale is morally
responsible for running over the pedestrian. The reason why tracing is appealed to in the
first place in cases such as Dale's is to explain how someone who lacks control over an
action can be morally responsible for that action.
King anticipates this objection. He claims that one might argue that we should
treat Norma's case differently than Dale's. But this would simply take us to the
recklessness interpretation, to which I turn now. According to the temporally extended

version of action individuation that King's argument requires, there is a single action that
begins with Dale drinking and ends with his running over the pedestrian. If Dale didn't
think about the possible harms of his continued drinking, then we model his case on the
recklessness interpretation. Dale's control over his drinking extends forward to his
running over the pedestrian. But even if we do this, it seems that we need tracing.
Recall Fischer and Ravizza's characterization of tracing:

When one possesses control over one's act at t1, and one can reasonably be
expected to know that so acting will (or may) lead to the agent lacking control
over her action at t2, one can be held responsible for so acting at t2 (Fischer and
Ravizza 2000: 50-51).

But also note that they claim that there are other ways that tracing principles can be
formulated (ibid.). The most plausible interpretation of the way King individuates action
is by taking a componential approach. According to this way of individuating actions, a
larger action has various smaller actions among its parts (for more on this see Ginet
1990). For instance, Dale's larger action has various components including drinking too
much, driving drunk, running over the pedestrian, etc. (each of which are actions).3
Now, when Dale performs the action component, running over the pedestrian, he does
not exercise the requisite control over this component (this might be, for instance,
because he was not responsive to reasons). In order to explain why he is morally
responsible despite lacking this control, one must trace back to a previous action
component over which he did have control.

We might formulate a tracing principle with this in mind as follows:

When one possesses control over a part, P1, of some temporally extended action,
A, at t1, and one can reasonably be expected to know that performing P1 will (or
may) lead to the agent lacking control over a part, P2, of A at t2, one can be held
responsible for P2.

King anticipates my objection. He claims that appealing to tracing in cases of

recklessness is problematic. He states, Explanation by tracing is an exceptional
explanation, meant to handle exceptional cases. While we can normally explain
responsibility by appeal to exercises of control at the time of action, these cases resist
such treatment because the usual control is absent. So we're forced to look to the agent's
past (King 2014: 464). He therefore claims that if we appeal to tracing in cases of
reckless action, then we aren't using tracing as an exceptional explanation to cover
exceptional cases. But there is no reason to accept King's claim that tracing is an
exceptional explanation. The more obvious implication is that we are directly responsible
for much less than we may have thought. In response, King claims,

This would be a remarkable result, however, since the prevailing view in the
literature is that tracing is needed to explain a certain range of cases, not all cases.
Moreover, this result would seemingly undermine the primary motivation for an
appeal to tracing. To get the right result in cases like drunk-driving, so the thought

goes, we must employ tracing. If tracing is to be used everywhere, this seems to

weaken its initial motivation. (King 2014: 7)

But this doesn't follow. If we accept the supposition that we are directly responsible for
much less than we may have thought, this doesn't mean that we'll need tracing in all cases
nor does it mean that we must use tracing everywhere. Perhaps we still only need it in a
certain range of cases - it is just that the range of cases may be bigger than we might have
thought.4

IV. RESPONSIBILITY FOR CONSEQUENCES

In typical tracing cases, a vignette is presented in which an agent is intuitively
morally responsible for some consequence despite not satisfying some part of the control
condition. Proponents of tracing explain why the agent is morally responsible by tracing
to a point in time at which the agent did satisfy the control condition. In the typical drunk
driving case, this point is usually the decision of the agent to keep drinking or to get
behind the wheel after drinking too much. Andrew Khoury argues that we should not
appeal to tracing in these cases because agents cannot be morally responsible for the
consequences of their actions.
Consider Khoury's argument that no one is blameworthy for consequences:
1. If some thing is not that in virtue of which one is blameworthy then one cannot be
blameworthy for that thing.
2. The consequences of one's actions are not that in virtue of which one is
blameworthy.

3. Therefore, one cannot be blameworthy for the consequences of one's actions

(Khoury 2012, 195).5
Khoury claims that if (1) is denied, then it would seem that blameworthiness is unfair. To
show this he asks us to consider cases of resultant moral luck. He states, Resultant moral
luck occurs if the external factors that affect how one's action turns out affect one's
responsibility (Khoury 2012: 195). He asks us to consider the following cases:

Suppose an assassin carefully aims his rifle at the intended victim and pulls the
trigger. The victim is killed. Imagine that in the second case, everything is the
same as the first except for the fact that once the trigger is pulled a bird flies into
the path of the bullet and the victim is spared. Imagine that the experiences of the
assassins are qualitatively identical (up to the point at which the triggers are
pulled). (Khoury 2012: 195)

Khoury asserts that the two assassins seem equally blameworthy. It seems that if the only
difference between them is that a bird deflected the bullet of the unsuccessful sniper, then
their level of blameworthiness should be the same. He states, The idea is that
blameworthiness ought not to depend on factors that are external to the agent and are, for
that reason, simply a matter of luck (Khoury 2012: 195-196, emphasis added).
In his defense of premise (2), Khoury asks us to consider the following story from Orson
Scott Card's novel Ender's Game:

Ender is a student at an elite military Battle School. He quickly moves to the top
of his class after showing many tactical talents and is promoted to attend
Command School. For his final exam at Command School he plays a video game
that simulates a large battle. Ender's ships are greatly outnumbered by the alien
fleet orbiting around their alien planet. Displaying his willingness to win at all
costs he decides to use a special weapon which destroys the planet and with it all
the alien ships and some of his own. It is later revealed that the game was not a
mere simulation, but that his actions actually controlled the movement of troops
and that the simulated events actually took place culminating in alien genocide.
(Khoury 2012: 198)

According to Khoury this is a case where the normal correlation between expectable
consequences and actual consequences is severed. Of course, Ender did not expect his
actions in a video game to have such real world consequences. Khoury claims that in this
scenario, Ender's blameworthiness is completely determined by his subjective mental
states.
Khoury then asks us to consider a case in which Ender believes that his actions in
the video game will actually cause alien genocide in the real world but his belief is false his actions don't cause genocide. Khoury claims that Ender's blameworthiness rests not
on what consequences actually happen, but on facts about him at the time of action - such
as what consequences he could reasonably expect to happen and his reasons for acting.
So Ender would seem to be just as blameworthy had it been the case that his actions
actually cause genocide and he knew that this would happen. Khoury states, What

Ender's blameworthiness turns on is what consequences he could reasonably expect

would result from his action (and the reasons for which he is acting), not what
consequences actually do result from his action. We would not, for instance, let him off
the hook were he to plea that luckily no harm was caused (Khoury 2012: 199).

V. RESULTANT MORAL LUCK AND TRACING FOR IMPAIRED DECISIONS

Khoury's argument depends on the rejection of resultant moral luck. In this section, I will
provide reasons not to reject resultant moral luck. I will then argue that even if Khoury's
argument is sound, he has still not shown that tracing is unnecessary. This is because not
all tracing cases involve responsibility for consequences. For instance, there are types of
cases that do not involve responsibility for consequences. Yet tracing is needed in these
types of cases.
Khoury's argument might be convincing to those who are already uncomfortable
with resultant moral luck. Here, I only wish to provide some reasons to doubt Khoury's
rejection of resultant moral luck. Khoury offers three considerations in support of doing
so: First, blame ought not to depend on factors that are external to the agent and are, for
that reason, simply a matter of luck. Second, the intuition that two agents are equally
blameworthy despite only one bringing about a bad consequence. And third, accepting
resultant moral luck seems to make blameworthiness unfair. Let us take each of these in
turn.
With regard to the first consideration, I note the distinction between something's
being simply a matter of luck as opposed to something's being partly a matter of luck.
Consider the successful assassin's being blameworthy for the consequence that the

president is killed. Rather than being simply a matter of luck, I would claim that this
consequence obtaining is partly a matter of luck. Consider this: it is not simply a matter
of luck that the world cooperated with someone's actions. That is the way the world
normally works. If it were simply luck which consequence was brought about then the
consequences that were brought about by agents performing intentional actions would be
wildly divergent from what normal agents actually experience in the world. The
successful assassin was successful in doing what he intended to do despite luck. The
obtaining of the consequence does depend partly on factors that are external to the agent
(the world cooperating with the assassin's actions and intentions). But this does not make
it the case that the consequence is solely dependent on factors external to the agent and
simply a matter of luck. If someone's actions bring about a consequence in the normal
way, I see no reason not to claim that they are blameworthy for bringing about this
consequence. Agents who intentionally bring about bad consequences can be
blameworthy for bringing about those consequences because they were successful in
completing what they were trying to do.
What about the unsuccessful assassin? It does seem to be simply a matter of luck
that he did not kill his intended target. But, as Nathan Hanna notes in his discussion of a
similar case, Luck just doesn't befall him. It interferes with him, with his plans and
intentions (Hanna 2014: 683). The successful sniper doesn't have bad luck that the world
cooperated with his intentions - that's the way things usually work out in our world. The
unsuccessful sniper has good luck that the world doesn't cooperate with his intentions he's lucky that he is not blameworthy for bringing about the consequence, that the target
is dead. There seems to be an intuitive difference between someone's blameworthiness

being simply a matter of luck and someone's lack of blameworthiness being simply a
matter of luck. The unsuccessful assassin is not blameworthy for bringing about the
consequence that the target is killed. And it seems to be simply a matter of luck that he
does not bring this consequence about. His not being blameworthy is simply a matter of
luck. But his not being blameworthy because of luck seems much less problematic than
the claim that the successful assassin's blameworthiness was a matter of luck.
Now let us consider the second consideration: the intuition that two agents are
equally blameworthy despite bringing about different consequences. Though Khoury has
argued that no one is blameworthy for consequences, he has not argued that
consequences are totally irrelevant to moral responsibility. On the contrary, consequences
are epistemically relevant to moral responsibility. He states,

The tendency to think that agents are blameworthy for the consequences of their
actions is based on a confusion between epistemic and metaphysical
considerations. Some consequence-types are such that there is a high correlation
between their occurrence and an agent with particular criticizable mental states
that ground blameworthiness; fatalities from drunk driving, for example. But this
correlation is explained by the fact that the consequence is typically expectable in
these kinds of cases. And it is the expectability of the consequence, rather than the
occurrence that explains the agent's blameworthiness. (Khoury 2012: 202)

By making this move, he attempts to explain away the intuition that the successful sniper
is more blameworthy than the unsuccessful one.

In the cases of the two assassins, I admit I don't have the same intuition as
Khoury. I believe that the successful assassin is more blameworthy than the unsuccessful
one because the successful one brought about the consequence, that the target died. And I
don't think that the obtaining of the consequence is only epistemically relevant to the
assassin's blameworthiness. It seems much more plausible that the successful assassin is
actually more blameworthy than the unsuccessful one. One reason in favor of this
intuition is that if Khoury is right, then all drunk drivers like Drunk Dale, who freely and
knowingly get drunk and get behind the wheel, are equally blameworthy regardless of
whether or not they bring about any bad consequences. Likewise all people who attempt
murder are equally blameworthy regardless of whether or not they succeed at bringing
about the consequence that they were intentionally trying to bring about. This seems to be
implausible. But I am willing to admit that this may only lead to a dialectical stalemate
between those who have one intuition and those that have another.
Third, is it the case that accepting resultant moral luck makes blame unfair? Some
ways that it could be unfair would be for us to use inconsistent principles or apply
principles inconsistently in blaming people. But the claim that people are blameworthy
for a consequence that they intentionally tried to bring about can be applied consistently
to all who actually bring about those consequences that they were intentionally trying to
bring about (c.f. Hanna 2014: 692). If what consequences one brings about were simply a
matter of luck, then perhaps this unfairness claim would be justified. But I have already
argued that this is not the case. Would holding people responsible for the consequences
they intentionally bring about be unfair if the obtaining of those consequences was partly
a matter of luck? Consider an example from Robert Kane:

A husband, while arguing with his wife, in a fit of rage swings his arm down on
her favorite glass-top table top intending to break it. Again, we suppose that some
indeterminism in his outgoing neural pathways makes the momentum of his arm
indeterminate, so that it is undetermined whether the table will actually break
right up to the moment when it is struck. Whether the husband breaks the table is
undetermined and yet he is clearly responsible, if he does break it. (It would be a
poor excuse to offer his wife, if he claimed: Chance did it, not me. Though
indeterminism was involved, chance didn't do it, he did.). (Kane 2007: 27)

Now, this case is different from the cases being discussed thus far - for instance, this case
features the problem of luck for indeterministic theories of free will and the previous
cases feature resultant moral luck. But the general principle still stands. In this case, the
obtaining of the consequence, that the table breaks, is partly a matter of luck (because of
the indeterminism). And yet, it does not seem unfair to say that the husband is morally
responsible for bringing this consequence about.
Khoury responds to a similar objection. He states,

One might hold that when the consequence that occurs is expectable this allows
for blameworthiness for the consequence (for which one lacks [a good] excuse).
In reply, such a move strikes me as ad hoc. The only motivation for this response
is that it saves the pre-theoretical intuition that agents can be blameworthy for the
consequences of what they have done. But in scenarios in which actual

consequences and expectable consequences do line up, it is the fact that the
consequence is expectable that makes it the case that the agent is blameworthy,
not the occurrence of the consequence. (Khoury 2012: 204)

Khoury makes two claims in this response. First, the move that he considers seems ad
hoc. And second, the fact that the consequence is expectable makes it the case that the
agent is blameworthy, not that the consequence obtains. Let us consider each of these
claims with regard to my proposal in the previous paragraph. First, is my proposal ad
hoc? In my proposal, I am simply appealing common intuition that agents are
blameworthy for the consequences that they intentionally bring about. It is not clear to
me how this attempt could be ad hoc. Second, this claim brings us back to the discussion
of conflicting intuitions above.
Finally, suppose my reasoning thus far is unpersuasive. I admit that this is
probably the case for most people who share Khoury's uneasiness regarding resultant
moral luck. But even if Khoury's argument is sound and no one is blameworthy for the
consequences of their actions, tracing will still be needed to explain responsibility for
impaired decisions. Suppose that when Drunk Dale decides to drive home he is so drunk
that he does not exercise the control necessary to be directly morally responsible for the
decision (for instance, he might not be responsive to reasons). Intuitively, he seems to be
morally responsible for the decision. In order to explain why Dale is morally responsible
for his impaired decision, we will need to trace back to a prior action over which Dale did
have control and which he could have reasonably expected to put himself into an

impaired state. With this in mind, we could formulate tracing for impaired decisions like
this:

When one possesses control over one's act at t1, and one can reasonably be
expected to know that so acting will (or may) lead to the agent lacking control
over her decision at t2, one can be held responsible for so deciding at t2.

The reason that Khoury's proposal will not be able to accommodate impaired decisions is
because of the way he defines the term consequence. According to Khoury,
Consequence will be taken to mean an event or state of affairs (causally related in the
appropriate way to an action of the agent) under a description that makes no reference to
the mental states of the agent in acting (Khoury 2012: 197). So according to this
denotation of the term, Dale's decision to drive drunk is not a consequence. Therefore, we
must appeal to tracing to explain why he is morally responsible for his impaired
decision.6

VI. CONCLUSION
In a recent article in defense of tracing Fischer and Tognazzini state,

Much is at stake in evaluating a critique of tracing. We do not see how a theory of

moral responsibility could adequately handle the range of drunk-driving cases,
Martin Luther cases, and manipulation cases without some sort of tracing
component; tracing just seems both highly plausible and theoretically

indispensable. Indeed, theorists as diverse as the libertarian Robert Kane and the
semicompatibilists John Martin Fischer and Mark Ravizza have embraced
tracing as a component of their theories of moral responsibility.
If tracing cannot be vindicated, this would seem to be a devastating
challenge to such theories as Kane's and Fischer'sand any control-based theory
of responsibility...(Fischer and Tognazzini 2009)

I have attempted here to vindicate tracing from the arguments presented by King and
Khoury. King argued that we don't need tracing because tracing cases can be modeled on
cases of negligence and recklessness. Khoury argued that tracing is unnecessary because
no one is responsible for the consequences of their actions. The best case scenario for my
interlocutors is that Khoury's argument against responsibility for consequences is sound.
This would allow King some wiggle room and allow him to block the tracing for
consequences objection I had to his argument. But even if this is the case, King must
appeal to some form of tracing to explain responsibility for temporally extended
component actions and Khoury must appeal to tracing to explain responsibility for
impaired decisions. Either way leaves the necessity of tracing intact.7

References
Fischer, J. M. and Ravizza, M. 1998. Responsibility and Control: A Theory of Moral
Responsibility. Cambridge: Cambridge University Press
Fischer, J. M. and Tognazzini, N. 2009. The Truth about Tracing, Nous 43: 531-56.
Hanna, Nathan. 2014 Moral Luck Defended, Nous 48:4, 683-698.
Kane, Robert. 1998. The Significance of Free Will. New York: Oxford University Press.
___________. 2007 Libertarianism, Four Views on Free Will, ed. by Ernest Sosa, Blackwell
Publishing, 5-43
Khoury, Andrew. 2012. Responsibility, Tracing, and Consequences, Canadian Journal of
Philosophy 42(3-4), 187-207.
King, Matt. 2014. Traction without Tracing: A (Partial) Solution for Control-Based Accounts
of Moral Responsibility. European Journal of Philosophy 22(3): 463-482.
Mele, Alfred 1995 Autonomous Agents. New York: Oxford University Press.
Shabo, Seth. Forthcoming. More Trouble with Tracing, Erkenntnis.
van Inwagen, Peter. 1983. An Essay on Free Will. Oxford: Clarendon Press.

1 I wont be specifying which control condition is at issue. For the purposes of this
paper, control condition will simply be a placeholder for any plausible account of
the control condition for direct moral responsibility.
2 Thanks to Andrew Khoury for helping me make this point more clearly.
3 One might claim that King might also make use of a coarse-grained approach to
action individuation and claim that Dales drinking heavily and Dales running
over the pedestrian are different descriptions of the same event (See Davidson
1980). A tracing principle for action descriptions similar to the principle I supply
for action components would still be needed
4 This is not a novel idea. C.f. Kane 1998 and van Inwagen 1983.
5 For Khoury, consequence is a term of art that refers to an event or state of
affairs (causally related in the appropriate way to an action of an agent) under a
description that makes no reference to the mental states of the agent acting
(Khoury 2012, 197). For instance, in the assassin cases below, that the target was
killed is the consequence of the first assassins action. This is an event that makes
no reference to the mental states of the assassin.
6 C.f. Shabo Forthcoming, footnote 3.
7 My thanks to Alfred Mele, Gabriel de Marco, Taylor Cyr, Andrew Khoury, and an
anonymous referee for helpful comments on previous drafts of this paper.