FINKING FRANKFURT

Daniel Cohen* & Toby Handfield†

1. Laertes is about to stab Hamlet. Claudius, watching from the wings,

desperately wants Laertes to stab Hamlet and has implanted a device in Laertes’
brain that, if activated, will force him to stab Hamlet. Claudius intends to activate
the device if it appears that Laertes will change his mind at the last moment. Laertes,
however, stabs Hamlet of his own accord and Claudius’s device is never activated.
By appealing to examples such as this, Harry Frankfurt (1969) presents a
powerful objection to the view that moral responsibility requires the ability to do
otherwise (the principle of alternative possibilities, or PAP). Frankfurt’s argument relies
on two claims: firstly, that Laertes is responsible for stabbing Hamlet and secondly,
that Laertes lacks the ability not to do so. The former claim seems obviously true.
The latter claim is, however, more controversial.
In particular, the latter claim depends on how ‘ability’ is understood. In this
paper, we shall discuss an analysis of ability in terms of agents’ dispositions. Relying
on a dispositional analysis of ability, Michael Smith (1997, 2003) has argued that
agents in Frankfurt examples, such as Laertes, should be thought of as possessing
the ability to do otherwise, contra Frankfurt. We will grant that Smith is right about
agents such as Laertes, but argue that another classic Frankfurt case evades Smith’s
analysis. Frankfurt’s willing addict (1971), in particular, cannot be shown, on Smith’s
account, to possess the ability to do otherwise. Provided one shares the intuition that
the willing addict is indeed responsible, our argument suggests that Smith fails,
ultimately, in his defence of PAP.
We do not intend to argue here that no account is available according to which
the willing addict possesses the ability to do otherwise. We will only argue that no
dispositional account is available. Nevertheless, if we are right, the burden of proof
will fall on the defender of PAP to provide an account that does not, however tacitly,
rely on dispositions.

2. Smith (2003) suggests that a variety of puzzle cases, including Frankfurt’s

examples, can be explained in terms of a dispositional analysis of the ability to do
otherwise. So, to the extent that Laertes may be said to have the disposition not to

*Philosophy Program, RSSS, Australian National University and School of Philosophy & Bioethics, Monash
University, email: cohen at coombs.anu.edu.au.

† School of Philosophy & Bioethics, Monash University, email: toby.handfield at arts.monash.edu.au.

This is a preprint draft of a paper forthcoming in Philosophical Studies.

stab Hamlet – despite the presence of Claudius and his device – we may salvage the
thought both that Laertes has the ability to refrain from stabbing and that this ability
is necessary for him to be responsible for his behaviour.
Simply, Smith’s view is that Laertes is responsible only if he possesses two
crucial capacities, one being the capacity to form correct evaluative beliefs (a capacity
for moral judgment), and the other being a capacity to bring one’s actions into
accord with one’s evaluative beliefs (a capacity for moral action, one might say.)
These capacities are akin to dispositions. To ascribe a capacity to someone is to
say something about what they could do, without necessarily saying what they will
do. It is to locate agents in modal space: to say something about how things will go
under various possible circumstances. If, as we may imagine the case, Laertes’
behaviour does not accord with his evaluative beliefs, Laertes’ possession of the
capacity for moral action means that it was in his power to act otherwise: to act in
accordance with his evaluative beliefs and to refrain from stabbing Hamlet.
Dispositions are typically thought of as dispositions to give a certain
manifestation, if a certain stimulus phenomenon occurs. Smith’s capacities have
clearly described manifestation conditions – forming correct evaluative beliefs,
acting on one’s evaluative beliefs – but he is chary of giving an account of the
stimulus conditions for these capacities. This is not to say that Smith’s capacities
need be brute propensities, which can manifest without any stimulus condition at all.
Rather, the idea appears to be that the capacities do not have any very well
understood, and perhaps only indeterminate, stimulus conditions. Smith does say,
however, that the capacity must be suitably “multi-track” (2003, p. 123). That is to
say: they must manifest similarly under a reasonably wide range of similar
circumstances.
To try and characterize in counterfactual terms, then, what is meant by saying
that Laertes has the capacity to do otherwise, one might say: Had things been
otherwise, in a wide variety of suitably minor ways, Laertes would have acted otherwise. Of
course, in the situation described, this counterfactual is probably false, because of the
presence of Claudius. Thus, if one relies upon a counterfactual characterization of
Laertes’ capacities, it is somewhat dubious to claim that Laertes has the capacity to
do otherwise in this case.
However, it has been known for some time that dispositional properties cannot
be analysed in terms of simple counterfactuals, of the sort given above. Smith draws
on this recent work on dispositions to suggest that certain counterexamples to the
conditional analysis of dispositions – finks and masks – are of precisely analogous
structure to Frankfurt’s cases.1 Finks and masks show that the simple conditional
analysis of dispositions is false, but they don’t show that objects lack dispositions,
even when finks and masks are present. Analogously then, Smith suggests that

1 For another, undeveloped, suggestion along these lines see Vihvelin 2000, note 34.

2
Frankfurt cases show that simple conditional analyses of the capacities for moral
judgment and moral action are false, but that the cases do not show that we lack
these capacities.
A fink is a feature of a situation such that, were a disposition to be triggered by
its stimulus, the fink would very quickly act to remove the disposition before it could
be manifested (Martin 1994). David Lewis’s influential example of a fink is a sorcerer
who wishes to protect a glass from breaking (1997). If the glass were struck, the
sorcerer would very quickly change the molecular structure of the glass so that it
would no longer be fragile. Thus it would not be true to say of the glass that, if
struck, it would break. Nonetheless, insofar as the glass is never struck, there is a
strong intuitive appeal in the thought that the glass is fragile.
A mask is like a fink in that it acts to prevent the manifestation of a disposition,
but rather than removing the disposition itself, a mask (or antidote) interferes with
the manifestation process (Johnston 1992, Bird 1998). A poison is disposed to cause
death if ingested. For some poisons, however, if I ingest them I shall not die, because
I will take an antidote which prevents the poison from manifesting its disposition.
By considering Laertes in light of these possibilities, we may now understand
how Smith defends the dispositional analysis of ability. Laertes is like the fragile
glass protected by the sorcerer. Just as the glass remains fragile, at least when it is
not in fact dropped, Laertes may be thought to possess the ability to do otherwise, at
least while Claudius’s device is not activated. Claudius, like the sorcerer, is a fink
because he will destroy Laertes’ disposition not to stab Hamlet, but only in those
worlds where his disposition would otherwise have been manifested.2
Even though both the glass and Laertes will not manifest their dispositions in
the presence of a fink, they may still be said to have their dispositions, when not
actually finked. This is why, in identifying Laertes’ ability, Smith says that we must
‘abstract away’ from all features external to the relevant features of his brain, at least
when these features don’t interfere with his behaviour (Smith 2003, pp. 126–7). The
irrelevant features will, of course, include Claudius and his device.
We welcome Smith’s account of the phenomena. However, there is crucial work
to be done in exploring the range of cases which can be analysed in this way. The
appeal to examples only shows that some Frankfurt situations are compatible with
the possession of the disposition to do otherwise. It does not show that all Frankfurt
situations are so compatible. Is it possible that some Frankfurt situations are
“radical”, such that their obtaining suffices to eradicate the disposition to do
otherwise?

2 Depending on whether Claudius’s device removes Laertes’ disposition to do otherwise, or whether

the device simply prevents this disposition from being manifested, Claudius may in fact be either a
fink or a mask. We shall assume he is a fink, but nothing of importance turns on things being one way
or the other.

3
 We here argue that there could indeed be such radical finks. Moreover, we
show that the possibility of such finks throws up a problem for Smith’s dispositional
defence of PAP. The willing addict, as we shall argue, is intuitively responsible for
her behaviour, yet she cannot be said to possess the ability to do otherwise, at least
on Smith’s dispositional analysis.

3. In philosophical discussion, finks are typically conceived of as something

extrinsic to the bearer of the disposition. The presence of Lewis’s sorcerer, for
instance, is not an intrinsic feature of the glass. Similarly with masks: it is not an
intrinsic property of a given poisonous substance to be such that, upon being
ingested, the poisoned party will also ingest an antidote. Clearly this is an extrinsic
property of the substance, because intrinsically perfectly similar substances could
lack this property.
Could there be intrinsic finks for a disposition? Following Sungho Choi (2005),
we suggest not. Consider a glass made of a substance which is intrinsically such that,
when struck, it acquires a very durable molecular structure. Consequently, if it were
struck the glass would not break. One might have thought that this is just another
fink. The glass is fragile, but it is in a situation such that, if exposed to the stimulus,
it would cease to be fragile.3
We believe this would be a mistake. It is intuitively plausible that such a glass
is better described as non-fragile. If one has no clear intuition about this glass,
consider Choi’s heuristic (pp. 499–500) for determining whether an object has an
intrinsic disposition: does any intrinsic duplicate of the object, subject to the same
laws of nature, obviously possess the disposition? The object may have intrinsic
duplicates of which it is true to say that, if struck, they would break. But all such
possible glasses will have this property in virtue of some extrinsic factor, such as a
fink or mask. None of these objects obviously have the disposition, given the
complication of these external factors. Given this, we suggest it is very plausible to
deny that the glass is fragile at all.
This pattern appears to generalize. If an object is alleged to have a disposition,
but also to have an intrinsic fink or mask to that disposition, it is very difficult to see
why we should accept this description, rather than simply deny that the object has
the disposition at all.
We make the following conjecture, then: dispositions cannot be intrinsically finked
or masked. If an object with a disposition were to acquire an intrinsic property which
finked or masked the disposition, then it would simply cease to have the disposition.
Or, to put it another way, intrinsic finks and masks are radical: upon instantiation
they eradicate the original disposition.

3This appears to be what Lewis (1997, p. 157) thought. He suggests a glass could be both fragile and
non-fragile, provided one of those dispositions is finkish, and the other is not.

4
4. The case of Laertes and Claudius is a clear example of an extrinsic mask. An
intrinsic duplicate of Laertes, subject to the same laws of nature, may clearly possess
the disposition not to stab Hamlet. Therefore there is no difficulty in maintaining,
with Smith, that Laertes possesses the disposition not to stab Hamlet, and in virtue
of this, the ability not to do so.
Frankfurt’s willing addict (1971), however, is less easily assimilated to this
paradigm. The willing addict has a disposition to crave narcotics. This craving,
moreover, is strong enough to force the addict to ingest narcotics. (It is in virtue of
this that she is, indeed, addicted). This disposition will be triggered whenever she
goes without narcotics for a significant period of time. The addict, however, never
goes without narcotics for that period of time, because she willingly chooses to
ingest them at regular intervals. We may imagine that, due to her constant ingestion
of narcotics, the willing addict may not even realise that she is addicted. Frankfurt
claims that, like Laertes, the willing addict is responsible for her behaviour, despite
not being able to do otherwise. We have seen that Smith is able, however, to show
that Laertes may, in fact, possess the ability to do otherwise, if abilities are
understood dispositionally. We will argue that Smith cannot make this claim in the
case of the willing addict. On Smith’s account, the addict will be able to do otherwise
only to the extent that she possesses the disposition not to ingest narcotics, a
disposition that might be finked (or masked) by her addiction (Smith 1997).
However, does the willing addict possess such a disposition? This depends crucially
upon the relation between the addict and her addiction.
Either the addiction is intrinsic to the addict or it is extrinsic to the addict.
Neither option is trouble-free.
The first option will seem clearly to follow if we think of the addict as at least
partly constituted by her body. An intrinsic duplicate of the addict’s body, subject to
the same laws of nature, will plausibly share the disposition to ingest narcotics. If
this is the case, however, Smith cannot say that the willing addict has the ability to
do otherwise: whatever disposition the addict may have possessed to refrain from
taking drugs has been eradicated by the radical mask of her addiction. Assuming
that the addict is indeed responsible for her behaviour, Smith cannot thus claim that
the ability to do otherwise is necessary for moral responsibility. We seem forced to
deny PAP, vindicating the original purpose of Frankfurt’s examples.
Smith might, however, bite the bullet. He may say that Laertes is responsible
because he does possess the ability to do otherwise: his disposition to do otherwise is
finkish, but is nonetheless present. The willing addict’s ability, on the other hand, is
not simply finkish, but actually eradicated by her addiction. On this basis, the addict
cannot be held responsible for her behaviour. This move is problematic because the
intuition that the willing addict is responsible appears fundamentally on par with the
intuition that agents such as Laertes are responsible. (Plausibly, they are instances of
the same intuition, so that they stand or fall together.) Denying that the willing
addict is responsible is only plausible to the extent that one also denies that Laertes

5
is responsible. Any putative analysis of responsibility must, moreover, respect this
parity of intuitions.4
Perhaps, however, Smith may defend the intuition that the willing addict is
responsible, despite lacking the ability to do otherwise at the time of her addiction.
That is, he may argue that the willing addict is responsible for her addictive
behaviour because she freely chose to become addicted at some time before she was
addicted. On this view, Laertes and the willing addict are only superficially on par,
for the willing addict is not strictly responsible for her addictive behaviour as such,
but rather only for some earlier behaviour that caused her currently to behave as she
does.5
This suggestion is problematic for a couple of reasons. Firstly it is not obvious
that a free choice to initiate some chain of events will always render one responsible
for all the events that follow, unless it were reasonably foreseeable that those events
would follow. However, our intuitions about the responsibility of the willing addict
for her behaviour don’t seem contingent on any assumptions about what she foresaw
at the time she became addicted. Secondly, it’s not clear that our intuitions about the
responsibility of the willing addict at all depend on there being some time when she
could have avoided being addicted. Even if the willing addict were born addicted to
narcotics, as long as she always consumes them on her own volition (without, let’s
imagine, even realising she is addicted), it seems clear that she is responsible for her
behaviour. It is clear, however, that this is not an intuition about her choice to
become addicted. Rather, it appears that the willing addict’s deliberative choice to
consume narcotics, by itself, is sufficient for her to be responsible.
The second option – saying that the willing addict’s addiction is not an intrinsic
part of her – is perhaps more superficially attractive, then. But on what basis would
we say that the addiction is not intrinsic to the addict? Any psychological duplicate
of the addict would surely also share the same addiction. Perhaps psychological
duplication is too coarse and inclusive a relation to capture only the intrinsic features
of a moral agent, as opposed to the intrinsic features of the behavioural system in
which moral agency is realised. Perhaps by taking psychological duplicates, we are
including features of the merely “phenomenal self”, in addition to those of the
“noumenal self”. Smith himself suggests that it is just certain “relevant properties” of
the brain that are to be duplicated in identifying an agent’s capacities (Smith 2003, p.
122). However, it is not clear what reasons could be given for duplicating only the
non-addicted parts of the willing addict’s brain. The suggestion that only the non-

4 Smith may, of course, deny that Laertes is responsible, but this would be substantially to abandon
the project of developing a compatibilist analysis of ability. If Claudius’s counterfactual intentions are
analogous to causal determinism, they would appear equally to preclude the ability to do otherwise.
But if one maintains PAP, as Smith does, then determinism would appear to preclude responsibility.

5 Thanks to an anonymous referee who suggested this argument.

6
addicted parts of the addict’s brain constitute her “real self” (see Wolf 1990) would
seem conveniently to beg the question. Even if the distinction between real and
superficial selves could be made sense of, it’s not clear why an agent’s addictive
desires won’t be part of her real self.
Another way of pressing this worry is to compare the willing addict with the
unwilling addict, who “hates his addiction and always struggles desperately, although
to no avail, against its thrust” (Frankfurt 1971, p. 12). On the dispositional analysis
of ability, if the willing addict is said to have the ability to resist temptation, then, by
the same token, the unwilling addict must have this ability also. The unwilling
addict, however, seems blameless for giving in to temptation. The most natural
explanation for this is that the unwilling addict lacks the ability to resist temptation,
but Smith cannot say this, insofar as he credits the willing addict with the same
ability.6

5. In conclusion, the possibility of radical finks leaves Smith with an ugly

dilemma. Either he can fall in line with Frankfurt and abandon the claim that the
ability to do otherwise is a necessary condition for moral responsibility (PAP), or he
can deny some of the key intuitive phenomena regarding the responsibility of the
willing and unwilling addicts. We have not, however, argued against PAP. Rather,
we hope to have shown that the ability to do otherwise, if it is thought to support
judgements of responsibility – cannot be analysed dispositionally.7

6There is some suggestion that Smith would claim that the unwilling addict does, in fact, possess the
ability to resist temptation but that he deserves to be excused because the failure to exercise this
ability does not explain his behaviour (Smith 2003, p. 127). This approach is problematic because it
suggests that the unwilling addict, due to his addiction, is unable to exercise his abilities. If this is the
case, however, surely we must say the same about the willing addict. And if the willing addict is not
able to exercise her abilities, then she too deserves to be excused.

7Thanks to Neil Levy, an anonymous referee, and an audience at the University of Bristol for helpful
comments on an earlier version of this paper. Thanks also to Neil for our title.

7
References

Bird, Alexander. 1998. “Dispositions and Antidotes”. The Philosophical Quarterly 48:
227–34.
Choi, Sungho. 2005. “Do Categorical Ascriptions Entail Counterfactual
Conditionals?” The Philosophical Quarterly 55: 495–503
Cullity, Garrett and Gaut, Berys (eds.). 1997. Ethics and Practical Reason. Clarendon:
Oxford University Press.
Frankfurt, Harry. 1969. “Alternate Possibilities and Moral Responsibility”. The
Journal of Philosophy 66: 829–39.
Frankfurt, Harry. 1971. “Freedom of the Will and the Concept of a Person”. The
Journal of Philosophy 68: 5–20.
Johnston, Mark. 1992. “How to Speak of the Colours”. Philosophical Studies 68:
221–63.
Lewis, David. 1997. “Finkish Dispositions”. The Philosophical Quarterly 47: 143–158.
Martin, C. B. 1994. “Dispositions and Conditionals”. The Philosophical Quarterly 44:
1–8.
Smith, Michael. 1997. “A Theory of Freedom and Responsibility”. In Cullity and
Gaut 1997. Reprinted in Smith 2004.
Smith, Michael. 2003. “Rational Capacities, or: How to Distinguish Recklessness,
Weakness, and Compulsion”. In Stroud and Tappolet 2003. Reprinted in
Smith 2004.
Smith, Michael. 2004. Ethics and the A Priori: Selected Essays on Moral Psychology and
Meta-Ethics. Cambridge: Cambridge University Press.
Stroud, Sarah, and Tappolet, Christine (eds.). 2003. Weakness of Will and Practical
Irrationality . Clarendon: Oxford University Press.
Vihvelin, Kadri. 2000. “Freedom, Foreknowledge, and the Principle of Alternate
Possibilities”. Canadian Journal of Philosophy 30: 1–24.
Wolf, Susan. 1990. Freedom Within Reason. New York: Oxford University Press.