Upload
felipe-leon
View
219
Download
0
Embed Size (px)
Citation preview
Why Frankfurt-Examples Don’tNeed to Succeed to Succeed
felipe leon
California State University
neal a. tognazzini
College of William and Mary
Decisive counterexamples to substantive philosophical theses are few
and far between. It is relatively uncontroversial that Gettier (1963) gave
us one. It is much more controversial whether Frankfurt (1969) did.
Indeed, there is a huge literature surrounding so-called Frankfurt-style
counterexamples (FSCs), with no clear consensus one way or the other.
Many have contributed to the project of attempting to construct a fool-
proof FSC; many have argued against the possibility of successfully
completing that project.1 In this paper, we do not aim to add directly
to either of these projects. Rather, we aim to avoid both of them in a
way that we hope will shed light on the debate as a whole.
1. The Debate Over the Success of FSCs
But in order to sidestep, we need to know what it is we are sidestep-
ping. So allow us to give a bit of background. Prior to 1969, most phi-
losophers accepted the Principle of Alternative Possibilities (PAP),
which is the thesis that an agent is morally responsible for an action
only if the agent could have done otherwise.2 But, as is now well
known, Harry Frankfurt (1969) challenged this thesis by constructing
what appears to be a counterexample to it. Frankfurt’s basic insight
was the following: it seems that there are situations in which certain
factors block an agent’s access to any alternative courses of action, but
that nevertheless play no role in bringing about the action that the
1 For an excellent collection of papers on this topic, see McKenna and Widerker
2003.2 This is how Frankfurt 1969 puts it. But it should be noted that the ‘could’ indicates
ability and not mere possibility.
WHY FRANKFURT-EXAMPLES DON’T NEED TO SUCCEED TO SUCCEED 551
Philosophy and Phenomenological ResearchVol. LXXX No. 3, May 2010� 2010 Philosophy and Phenomenological Research, LLC
Philosophy andPhenomenological Research
agent actually performs. In cases like this, where the agent still acts for
his own reasons despite having no alternative course of action open to
him, it is intuitive to suppose that the agent is morally responsible
despite the lack of alternatives. Such cases would show that PAP is
false. For concreteness, let’s consider one of these sorts of cases, struc-
tured after Frankfurt’s own (and inspired by John Fischer (1982)):
Jones goes into a voting booth, deliberates in the normal way about
whom to vote for, decides to vote for the Democrat, and then votesfor the Democrat. Unbeknownst to Jones, however, his neural activitywas being monitored from afar by Black, who had implanted a chip
into Jones’ brain that would enable Black to directly stimulate partsof Jones’ brain. Had Jones shown any sign that he would vote Repub-lican instead, Black would have directly manipulated Jones via thechip to make it so that Jones decided to vote for the Democrat
instead, and did vote for the Democrat. But, as it turned out, Blackdid not need to step in, since Jones did not show any sign that hewould vote Republican instead.
Now, as we mentioned above, cases like this are supposed to elicit the
following two intuitions:
(1) The agent, S, could not have done other than perform action A.
(2) S is morally responsible for performing A.
But if (1) and (2) are both true in the case described, where ‘S’ is Jones
and ‘A’ is voting for the Democrat, then PAP must be false, because we
would then have a case in which Jones is morally responsible for an
action even though he could not have done otherwise than perform that
action. Consideration of these two intuitions leads to a natural under-
standing of what it would be for an FSC to succeed. Since, as you have
no doubt guessed by this point, the argument of this paper will turn on
distinguishing two different notions of ‘success’, let’s call this first under-
standing ‘the traditional definition of success’, which can be stated as
follows. An FSC can be considered a success if and only if it describes a
metaphysically possible scenario in which an agent is morally responsi-
ble for some action even though the agent could not have done other-
wise than perform that action. Without a doubt, an FSC that succeeded
in the traditional sense would show that PAP is false.3
The major point of contention between proponents and opponents
of FSCs has been whether any FSC can be constructed in which (1)
and (2) really are both true at the same time with respect to the same
3 At least on one charitable understanding of the principle. See section 5 below.
552 FELIPE LEON AND NEAL A. TOGNAZZINI
action. That is, the debate has focused on whether any FSC succeeds,
in the traditional sense. Although the case described above sure seems
like a case in which Jones could not have done other than vote for the
Democrat, opponents of FSCs have proven extremely adept at finding
alternative courses of action open to Jones. For instance, even in the
case above, it does seem like there is something open to Jones—after
all, it is still open to Jones to show a sign that he is not going to vote
for the Democrat. And so it looks like it’s not true that the above situ-
ation eliminates all alternatives, because Jones could have exhibited the
relevant sign instead.
And of course this is hardly the final word on the matter. The pro-
ject of attempting to construct a (traditionally) successful FSC contin-
ues full steam ahead. We’re happy to leave this project to the side,
however. For we shall argue that there is a different (and perhaps pref-
erable) sense in which FSCs succeed even if they don’t succeed in the
traditional sense stipulated above. This brings us to a new account of
success, which we defend below: an FSC can be considered a success in
this new sense if and only if it describes a metaphysically possible sce-
nario in which both an agent, S, is morally responsible for a particular
action, A, and either (1) S could not have done other than A or (2) if S
could have done other than A, then this ability is irrelevant to S’s
moral responsibility for A. Call this ‘the revised definition of success’.
Now it should be a bit clearer what we mean when we say, in the title
of the paper, that FSCs don’t need to succeed to succeed. We mean
merely that FSCs don’t need to satisfy the traditional definition of ‘suc-
cess’ in order to satisfy the revised definition, since the revised defini-
tion adds clause (2) as a disjunct.4
But does our revised definition of success even make sense? PAP seems
to be a straightforward conditional, and how could one possibly provide
a counterexample to a conditional without constructing a metaphysically
possible scenario in which the antecedent is true while the consequent is
false? Such questions are quite reasonable. In the next two sections, we
will discuss some general considerations about philosophical claims and
thought-experiments to appease these sorts of worries. Afterward, we
will bring the discussion to bear on the case of FSCs in particular.
4 The point about irrelevance that is included in the second disjunct of our revised
definition should sound familiar to those who have followed debates about the dif-
ference between robust and non-robust alternatives. John Martin Fischer (1994) ini-
tially introduced the idea that the alternative possibilities in question must be
sufficiently robust specifically to directly our attention to the difference between
alternatives that are relevant to moral responsibility and alternatives that are not.
Our argument in this paper is importantly similar to Fischer’s arguments about
robustness. Indeed, our project might be seen as a way of interpreting and explain-
ing more fully Fischer’s point.
WHY FRANKFURT-EXAMPLES DON’T NEED TO SUCCEED TO SUCCEED 553
2. General Considerations: Thought-Experiments
There are some important distinctions to be made that will help us
understand the difference between the traditional and the revised defini-
tions of ‘success’. First, let’s draw attention to the difference between
two sorts of thought-experiment: modal and non-modal. Modal
thought-experiments are those whose utility depends on whether the
thought-experiment justifies a modal claim—a claim that it’s possible
or necessary that something is or isn’t the case.5 For example, consider
the Cartesian thought-experiment involving conceiving or imagining
yourself existing apart from your body. This is a paradigm case of a
modal thought-experiment because it is useful only if it justifies the
modal claim that it’s metaphysically possible for your mind to exist
apart from your body. If it turns out that the thought-experiment does
not justify this modal claim after all, it can be put to no further use.
By contrast, non-modal thought-experiments are those whose utility
does not depend on whether they justify modal claims. Such experi-
ments have a variety of functions, but they typically involve showing
that a given philosophical position or thesis is inadequate. Here’s one
example. In debates about the nature of personal identity, those who
argue against certain ‘‘empiricist’’ accounts, such as psychological con-
tinuity accounts, sometimes use thought-experiments according to
which a person undergoes fission—two people are somehow created
from an original person. Richard Swinburne, for example, uses such a
thought experiment to show that if some psychological continuity
account is correct, then both of the resultant persons should be the
same person (by transitivity of identity), which is supposed to be
absurd—no one can be ‘‘of two minds’’ in such a literal sense.6 What
makes this a non-modal thought-experiment is that, for the purposes of
Swinburne’s argument, it really doesn’t matter whether the fission sce-
nario is metaphysically possible. Even if it turns out that fission cases
are metaphysically impossible, the thought-experiment can still be put
to good use because, arguably, it has succeeded in getting us to see that
psychological continuity theories of personal identity make identity
depend on the wrong sort of things (such as who else exists). And one
need not see into other possible worlds in order to see this latter point.
Or so the defender of the possibility of non-modal thought-experiments
would argue.
We’ll come back to the distinction between modal and non-modal
thought-experiments below in relation to the debate over FSCs. For
5 The relevant modality here is metaphysical.6 See Shoemaker and Swinburne 1984, pp. 12–19.
554 FELIPE LEON AND NEAL A. TOGNAZZINI
now, keep it in the back of your mind as we move to a second sort of
general consideration relevant to our argument.
3. General Considerations: Types of Philosophical Claim
The second set of important distinctions deals not with thought-experi-
ments but with types of philosophical claim. Interesting philosophical
claims (like PAP, for instance), often come in the form of some sort of
conditional.7 But there are (at least) three importantly different sorts of
philosophical claim that are each some form of conditional:
(1) Ordinary conditionals
(2) Strict conditionals
(3) Constituent ‘‘conditionals’’8
Each type of claim has a set of truth-conditions, corresponding condi-
tions of evaluation, and characteristic proper methods of evaluation
and defeat. Let’s look at each kind of claim in turn.
Ordinary conditionals are the least common sorts of significant
philosophical claims. For they look only to the actual world—as
opposed to other possible worlds—and are thus contingent claims.
Ordinary conditionals are true just in case, actually, the consequent is
true whenever the antecedent is true. Thus, they can be defeated by
showing an actual instance where the antecedent is true and the conse-
quent is false. The most frequently relevant means of evaluating ordin-
ary conditionals is via observation and observation-sensitive theory.
Strict conditionals are probably the most common ways to construe
significant philosophical claims. They deal with what is necessarily the
case. Unlike ordinary conditionals, other worlds besides the actual
world are relevant to their truth-conditions, defeat conditions, and
means of evaluation. Strict conditionals are true just in case, necessar-
ily, the consequent is true whenever the antecedent is. They can thus be
defeated by showing a possible instance in which the antecedent is true
and the consequent false. The most frequently relevant means of evalu-
ating strict conditionals is thus via a modal thought-experiment, in the
sense outlined above.
7 Or a biconditional, though we are leaving biconditionals to one side here, since
PAP itself is not a biconditional.8 We put this term in scare quotes because the formulation of constituent conditionals
need not explicitly use words like ‘if’ and ‘then’. We will drop the scare quotes in
what follows.
WHY FRANKFURT-EXAMPLES DON’T NEED TO SUCCEED TO SUCCEED 555
The final major kind of philosophical claim is one that asserts that
one (or more) entity is a constituent, or set of constituents, of another
entity. Arguably, these used to be the most common sorts of claim in
philosophy. Such claims have to do with essential properties. These
aren’t claims about contingent properties or extrinsic, Cambridge prop-
erties; rather, they are claims about what properties ‘‘enter into the
very being’’ of the thing being analyzed. The most cautious of these
sorts of claim merely assert that one or more entities are or are not
constituents of the essence of another entity. But the holy-grail type of
constituent-claim in philosophy is the analysis. Analyses attempt to
state the full essence—the recipe or blueprint—of some entity type or
token, as in ‘‘A = df BCD’’.
It is important to pause for a moment here and reflect on the impli-
cations of the truth-conditions of this last sort of philosophical claim.
Notice that their truth-conditions do not make essential reference to
other possible worlds. There is a good reason for this: a non-constituent
property can be necessarily co-extensive with a constituent property. So,
for example, the property of having three sides is necessarily co-exten-
sive with the property of having three angles (modulo certain qualifica-
tions about, for example, being a closed figure), but the former isn’t a
constituent—doesn’t enter into the being—of the latter. Or consider
Descartes’ account of epistemic justification in terms of clear and dis-
tinct perceptions of the intellect. On this account, it isn’t essential to
clear and distinct perceptions that they are true. For according to
Descartes, if it weren’t for an all-powerful god to ensure the connection
between clear and distinct perceptions and true beliefs, the former
wouldn’t be an infallible indicator of the latter. However, Descartes’
god ensures that the two are necessarily co-extensive. So again, being
an essential property is one thing, being a necessarily co-extensive
property is another, and the latter isn’t an infallible indicator of the
former. But if that’s right, then the methods of evaluation for constitu-
ent conditionals may diverge from those of strict conditionals.
The moral here is that constituent claims and analyses can be evalu-
ated and defeated without appealing to other possible worlds. And if
that’s right, then the door is open for evaluating them via non-modal
thought experiments—that is, via thought-experiments that don’t
depend for their utility on whether they justify claims about what is
necessarily or possibly the case.
4. Companions vs. Soul-mates
Perhaps a metaphor will help to really drive home the difference
between philosophical claims stated as strict conditionals, on the one
556 FELIPE LEON AND NEAL A. TOGNAZZINI
hand, and philosophical claims stated as constituent conditionals, on
the other. Strict conditionals do tell us something interesting about the
relationship between two properties. What they tell us is that one prop-
erty is a companion of another, in the sense that wherever the other
goes, the one tags along. One property is a companion of another, in
this sense, just in case it shows up in every metaphysically possible
world where its companion shows up.
But mere companionship doesn’t tell us anything interesting about
the relationship between the two properties that are companions. Per-
haps there is a deep conceptual connection between the two, and per-
haps not. When there is a deeper conceptual connection between the
two, we might call the properties soul-mates, rather than mere compan-
ions. Roughly, one property is the soul-mate of another just in case it
‘‘enters into the very being’’ of its soul-mate—just in case it is part of
the conceptual fabric of its soul-mate. Of course, if two properties are
soul-mates, then they will ipso facto be companions, as well, but the
converse does not hold. Indeed, it is precisely because the converse
does not hold that we need to distinguish between strict conditionals
and constituent conditionals. Whereas strict conditionals tell us about
companions, constituent conditionals tell us about soul-mates.
5. What Sort of Philosophical Claim is PAP?
So we have two sorts of thought-experiment: modal and non-modal.
And we have three sorts of philosophical claim: ordinary, strict, and
constituent conditionals. With these distinctions in mind, let’s turn
back to the issue at hand. Regarding FSCs, then, there are now two
important questions we need to answer:
(1) What sort of philosophical claim is PAP?
(2) What sort of thought-experiment is Frankfurt’s attack on PAP?
Those familiar with the literature on FSCs will know that PAP is usu-
ally (though perhaps somewhat surprisingly) stated as an ordinary con-
ditional, as follows:
PAPoc: An agent is morally responsible for an action only if
the agent could have done otherwise.
But if this were the way PAP were intended, then Frankfurt’s counte-
rexamples wouldn’t even stand a chance, because nowhere does Frank-
furt say that his counterexample scenarios must exist in the actual
WHY FRANKFURT-EXAMPLES DON’T NEED TO SUCCEED TO SUCCEED 557
world in order for PAP to be false. It’s more plausible to interpret
Frankfurt as thinking that all we need to do in order to show PAP
false is to find a possible world in which an agent is morally responsible
for an action even though the agent could not have done otherwise.
(This is, remember, the traditional definition for the success of an
FSC.) So, this suggests a second reading of PAP, put in terms of a
strict conditional:
PAPsc: Necessarily, an agent is morally responsible for an
action only if the agent could have done otherwise.
If this is how we read PAP, then at least Frankfurt’s alleged counterex-
amples don’t obviously miss the mark. But our discussion above tells us
that there is an even stronger way to read PAP. We could read it as a
constituent conditional instead:
PAPcc: A conceptually necessary constituent of being morally
responsible is having the ability to do otherwise.
This reading of PAP is stronger because it ties the ability to do other-
wise to moral responsibility in a much deeper way than PAPsc. In the
terminology of the previous section, another way to think about it is to
say that while PAPsc
only tells us that moral responsibility and the
ability to do otherwise are companions—in the sense that whenever you
have moral responsibility, the ability to do otherwise tags along—
PAPcc
tells us that the two notions are soul-mates, in the sense that the
ability to do otherwise is really part of the conceptual fabric of moral
responsibility. A less metaphorical way to put it would be to say that
PAPcc
tells us that when an agent is morally responsible, he is so partly
in virtue of his being able to do otherwise. If the in virtue of relation
obtains between two facts in this way, then it’s not enough to capture
that relation merely to say that both of the facts are true or even that
necessarily, both of the facts are true. In order to capture the relation,
something needs to be said about the conceptual connection between
the two facts in question. PAPcc
is an attempt to capture this concep-
tual connection.
6. What Sort of Thought-Experiment is an FSC?
Now, we have said that Frankfurt himself is plausibly interpreted as
attacking PAPsc. And it is PAP
scthat seems to have been the focus of
most contemporary discussions of FSCs. Let’s grant for the moment
that FSCs fail to show that PAPsc
is false. That is, let’s grant that
558 FELIPE LEON AND NEAL A. TOGNAZZINI
Frankfurt (and others) failed to find a metaphysically possible world in
which it’s both true that an agent is morally responsible and that the
agent could not have done otherwise. If FSCs were modal thought-
experiments, then this information would be enough to justify our
scrapping them altogether. Since (we are granting for the moment) no
FSC can justify the modal claim in question, FSCs are useless. But are
FSCs modal thought-experiments? Or can they still be put to good use
even if they don’t justify the modal claim in question? We think there
is a good case to be made that FSCs are most plausibly thought of as
non-modal thought-experiments, and we will gesture at the case for this
below. For now, let’s note that if FSCs can plausibly be considered
non-modal thought-experiments, it follows that even if they can’t be
used to refute PAPsc, they may nevertheless be able to refute PAP
cc,
and thus still tell us something interesting about moral responsibility in
the process.
What might they tell us? Non-modal FSCs might tell us that the
ability to do otherwise is not a requirement of moral responsibility. But
we don’t mean ‘requirement’ here in the sense that one thing is a
requirement of another just in case necessarily, you can’t have the
other without the one. Rather, we mean it in the sense that even if it’s
true that the ability to do otherwise always tags along with moral
responsibility, it’s nevertheless false that they are connected in the same
way that a constituent is connected with the thing it is a constituent of.
Part of the problem here, we are suggesting, is that there are at least
three different ways to interpret the phrases ‘is a requirement of’ and
‘is a necessary condition for’, corresponding to the three different types
of philosophical claim identified above. Perhaps (and we stress we
aren’t making any commitments about this at this point) the ability to
do otherwise is a requirement of or a necessary condition for moral
responsibility in the sense of PAPsc. It still may not be in the sense of
PAPcc. In other words, even if FSCs don’t constitute a counterexample
to PAPsc, they still may constitute a counterexample to PAP
cc. Thus,
FSCs don’t need to succeed to succeed.
7. Objection: Sinning Against Plantinga9
We suspect that some will have a problem with our main line of argu-
ment. For our account relies on the idea that a property can be neces-
sarily coextensive with a thing’s essential properties without itself being
essential to the thing. But how can this be? Isn’t it part of philosophi-
cal orthodoxy that a property is essential to a bearer just in case the
9 Thanks to Kevin Timpe for raising this objection forcefully.
WHY FRANKFURT-EXAMPLES DON’T NEED TO SUCCEED TO SUCCEED 559
bearer has the property in all possible worlds in which it exists? While
we agree that it is indeed orthodoxy, we think that this sort of modal
account is nonetheless inadequate. We’ll briefly discuss two reasons for
thinking so: (1) the modal account admits of counterexamples, and
more fundamentally, (2) the modal account is too coarse-grained to
adequately differentiate essential from non-essential properties. Let’s
discuss these criticisms in turn.
Alvin Plantinga’s account of essential propertyhood is representative.
Plantinga (1974) tells us that a property, P, is essential to its bearer, x,
if and only if x is P in all possible worlds in which x exists.10 So, sup-
pose that Ralph has the property of being a Californian in some possi-
ble worlds, but not in others. On Plantinga’s analysis, then, being a
Californian isn’t essential to Ralph. By contrast, suppose that Ralph
has the property of being a person in all possible worlds in which he
exists. In this case, Plantinga’s account would entail that being a per-
son is essential to Ralph.
We think Kit Fine (1994) provides a nice counterexample to
Plantinga’s analysis. Consider the set whose only member is Socrates,
and call that set ‘singleton Socrates’. There is no possible world in
which Socrates exists yet lacks the property of being the member of
singleton Socrates. So, by Plantinga’s modal definition of essence,
being a member of this singleton set is essential to Socrates. But intu-
itively, being a member of this set is not essential to or constitutive of
Socrates. As Fine puts it, ‘‘Strange as the literature on personal iden-
tity may be, it has never been suggested that in order to understand
the nature of a person one must know to which set he belongs. There
is nothing in the nature of a person…which demands that he belongs
to this or that set or which even demands that there be any sets.’’
(Fine 1994, p. 5.)
This brings us to our second complaint against Plantinga’s modal
account of essential properties: it’s too coarse-grained. For in Fine’s
case of Socrates and his singleton set, the set of worlds containing Soc-
rates and the set of worlds containing singleton Socrates are the same
set. Therefore, on the modal analysis of essential properties, we can’t
make sense of the intuitive asymmetry here: it is part of the nature of
singleton Socrates that it contains Socrates as a member, but it’s no
part of the nature of Socrates that he be a member of singleton Socra-
tes. Something has gone wrong here, and it is the adoption of a modal
analysis of essence. As Fine’s case nicely illustrates, possible worlds
are just too coarse-grained to do the work of singling out essential
properties.
10 This is a slight paraphrase of Plantinga’s account on p. 56.
560 FELIPE LEON AND NEAL A. TOGNAZZINI
8. Reply: Toward a Non-Modal Account of Essential Properties
But if modal analyses are inadequate, what sort of analysis is adequate?
We don’t think we need to give a positive account (not here, anyway),
but perhaps it would be helpful to offer some suggestive remarks about
how such an account might go. Here again, Fine’s article is helpful. Fine
distinguishes between modal and definitional accounts of essential proper-
ties. Modal accounts are of the sort Plantinga gives. Definitional
accounts are of the sort that Aristotle gave. Thus, just as one can give the
definition of what a word means, so one can give the definition of what
an object—or kind of object—is. As Aristotle put it, ‘‘clearly, then, the
definition is the formula of the essence.’’ (Metaphysics 1031a12).11
To be sure, on definitional accounts, claims of essential properties
are inextricably tied to modal properties and modal claims. Thus, if F
is essential to x, then x is F in all possible worlds in which x exists. But
the difference is that on the definitional account, the relationship
between modal properties and essential properties is asymmetric: the
modal properties supervene on the essential properties, but the essential
properties do not supervene on the modal properties. Thus, the con-
verse of the conditional above holds for the modal account of essence,
but not for the definitional account—x might be F in all possible
worlds in which x exists, but for all that, F might not be essential to x
(hence the case of Socrates and singleton Socrates).
9. A Potentially Successful Non-Modal FSC
The main motivation behind FSCs, it seems, and behind rejecting PAP
in general, is that alternative possibilities in and of themselves seem
altogether irrelevant to our ascriptions of moral responsibility, even if
they are there. Indeed, one often hears talk of ‘irrelevance’ in the litera-
ture on FSCs, but such talk tends to arise within a discussion of PAPsc
.
Adopting our framework above, however, provides a more nuanced
way of understanding the idea of irrelevance, in terms of a rejection of
PAPcc
instead. If that’s right, then, as we have pointed out, it’s at least
coherent to suppose that FSCs succeed in establishing the irrelevance
claim even if they fail to establish the falsity of PAPsc. But enough
about what it is merely ‘‘coherent to suppose’’. What considerations
can be adduced in favor of the claim that we should view FSCs as non-
modal thought-experiments, despite the opposite tendency in the litera-
ture? To start, let’s look at a variant on an FSC that we think does
pretty well at bringing out the irrelevance intuition. After we have the
11 Both the reference and the definitional construal of Aristotle’s account of essence
are Fine’s. See Fine 1994, p. 2.
WHY FRANKFURT-EXAMPLES DON’T NEED TO SUCCEED TO SUCCEED 561
case in front of us, we’ll make some suggestive (but ultimately incon-
clusive) remarks about how this case might help to show that FSCs
may still be useful even if there is no modal claim that they justify.
The case we have in mind is provided by John Martin Fischer, in
the following anecdote:
Some months ago, I was at the local supermarket. When I got to thecheckout counter, the checker asked, ‘‘Would you like a paper bag orplastic?’’ I thought for a moment and replied, ‘‘Plastic is fine.’’ Then
the checker smiled and said, ‘‘It’s a good thing—I see we only haveplastic!’’ (Fischer 2006, p. 62, n. 22)
We think it is plausible to say that Fischer is responsible for bringing
home plastic in this case, even though he couldn’t have brought home
paper. Were his wife to scold him for bringing home the groceries in
plastic bags on account of the fact that plastic is less friendly to the
environment than paper, for example, it wouldn’t be a good excuse for
Fischer to say, ‘‘But they were out of paper, so I had to take plastic!’’
(It would be an excuse, of course, but not a good one.) He chose plastic
for his own reasons and is therefore responsible for bringing home
plastic even though he couldn’t have brought home paper. Such a sce-
nario is like an FSC, of course, but it’s clearly not a case that rules out
all alternative possibilities. After all, Fischer could have chosen paper
instead. He would still have brought home plastic, of course, since that
was all they had, but he could have chosen something different. Never-
theless, we think that this sort of case makes plausible the claim that
whether there are any alternative courses of action open to Fischer is
just irrelevant to our assessment of Fischer’s choice and action.
What does seem relevant is the reason for which Fischer acted. Since he
believed that the store had paper bags in stock, his choice of plastic was
not at all influenced by the fact that plastic was all the store really had.
He thus made his decision for his own reasons. Of course, there is cer-
tainly much more that needs to be said about the details of this situation.
We can’t mount the full argument against PAPcchere, but we think that
considering the details of cases like Fischer’s can provide the resources
for making a plausible case that sometimes an agent’s ability to do other-
wise, even if it is there, is altogether irrelevant to the agent’s moral
responsibility. For the purposes of this paper, we are happy to leave our
discussion of this case here, merely as a suggestion for future work.
10. Final Thoughts
To conclude, we’d like to apply some of the above considerations to
the wider debate between compatibilists and incompatibilists about
562 FELIPE LEON AND NEAL A. TOGNAZZINI
determinism and moral responsibility, of which the debate over FSCs is
most often a part.
It is often compatibilists who get excited about falsifying PAP, since
that principle figures in one of the most celebrated arguments for in-
compatibilism. That argument (which has sometimes been called the
‘indirect’ argument for incompatibilism) can be stated as follows:
(1) If determinism is true, then no one is ever able to do otherwise
with respect to any action.
(2) An agent is morally responsible for an action only if the agent
could have done otherwise.
(3) Therefore, if determinism is true, then no one is ever morally
responsible for any action.
Premise (1) is supported by arguments such as Peter van Inwagen’s
(1983) Consequence Argument, and premise (2) is Frankfurt’s familiar
PAP. Compatibilists will want to resist this argument in some way, and
although at one time it was thought that the best way to resist was to
deny premise (1), ever since Frankfurt’s article many compatibilists
think that the best way to resist is to deny PAP, and thus deny premise
(2). One might think, then, that the argument of this paper should be
welcomed by compatibilists.
But this reasoning moves a little too quickly. Remember that com-
patibilists, by definition, are committed to the existence of metaphysi-
cally possible deterministic worlds in which there are morally
responsible agents. Given the plausible claim that determinism rules
out the ability to do otherwise, compatibilists will then be committed
to the existence of metaphysically possible worlds in which some agent
is morally responsible even though he could not have done otherwise.
And it is precisely this claim that we have avoided defending in this
paper. For all we have said about PAPcc
, incompatibilism may still be
true. Since we have distinguished between different readings of PAP,
we would also distinguish between different readings of premise (2) in
the above argument for incompatibilism. What we have argued in this
paper is that such an argument may be unsound when premise (2) is
interpreted as PAPcc, even if it is sound when premise (2) is interpreted
as PAPsc. This amounts to a concession that incompatibilism may well
be true, even if it turns out that the ability to do otherwise is not part
of the conceptual fabric of moral responsibility.12
12 There are of course some incompatibilists who already believe this. Our argument
will no doubt be welcomed by these incompatibilists.
WHY FRANKFURT-EXAMPLES DON’T NEED TO SUCCEED TO SUCCEED 563
So we should be wary about thinking that the argument of this
paper will be welcomed with open arms by compatibilists. That said,
however, we do think that our argument puts pressure even on those
who would defend the weaker PAPsc. For if a plausible case could be
made that PAPcc
is false, then wouldn’t it be mysterious if PAPsc
were
nevertheless still true? Or, to put it another way, one might wonder
why moral responsibility and the ability to do otherwise might be
thought to be companions. A natural answer is that they are soul-
mates, and any two properties that are soul-mates are automatically
companions. But if our argument above is right, then FSCs may indeed
succeed in showing that the two properties are not soul-mates after all.
And then what sort of explanation will be left for their supposed com-
panionship? Upon relinquishing a claim about soul-mates, it seems that
this should push one at least to suspend judgment regarding the paral-
lel claim about companionship. And insofar as that’s true, the compa-
tibilist will find this a happy result indeed.13
Finally, a word about the currently popular thesis that has come to be
known as ‘source incompatibilism’ (McKenna 2001). Although typically
the incompatibilist thesis has focused on the threat that determinism
would pose to the ability to do otherwise, and hence via PAP to moral
responsibility, more and more incompatibilists are becoming convinced
that there is a deeper threat from determinism. Rather than threatening
the ability to do otherwise, these source incompatibilists think that deter-
minism threatens some important notion of sourcehood or origination,
which is required for moral responsibility. This is not the place to discuss
the view in detail, but we would like to at least make one suggestion for
source incompatibilists that is inspired by the argument of this paper.
Many source incompatibilists formulate their view employing a
notion ‘fundamentality’ that is hard to interpret. To take just one
example, consider the following formulation from Michael McKenna:
[Source incompatibilists] hold that the fundamental threat to moralresponsibility from determinism derives, not from the thought that in
a deterministic world an agent cannot do otherwise; it derives, instead,from the thought that, if determinism is true, an agent’s actions donot originate in her. (McKenna 2001, p. 40)
Many other source incompatibilists have formulated the view similarly,
with language about how sourcehood, rather than the ability to do other-
wise, is more fundamental to moral responsibility, or is what gets to the
heart of the matter concerning moral responsibility. But many of these
13 We note that Michael McKenna has, independently, suggested something like this
line of reasoning to us (in conversation).
564 FELIPE LEON AND NEAL A. TOGNAZZINI
source incompatibilists also think that something like PAP is true, and if
that’s right, then it’s hard to understand what could possibly be meant by
saying that sourcehood is ‘‘more fundamental’’ to moral responsibility. If
both sourcehood and the ability to do otherwise are necessary for moral
responsibility, then shouldn’t they both be considered fundamental?
A source incompatibilist who adopted the argument of this paper,
on the other hand, can make sense of this intuition about fundamental-
ity in a very straightforward way. Such a theorist can simply say that
the sourcehood condition is more fundamental because it is relevant to
moral responsibility in a way that the ability to do otherwise is not.
Though both sourcehood and the ability to do otherwise may be neces-
sary companions of moral responsibility, it’s only sourcehood that is
partly constitutive of it. And therein lies its fundamentality.14
References
Fine, Kit. 1994. ‘‘Essence and Modality’’. Philosophical Perspectives 8:
1–16.
Fischer, John Martin. 1982. ‘‘Responsibility and Control’’. Journal of
Philosophy 79: 24–40.
—— 1994. The Metaphysics of Free Will: An Essay on Control. Oxford:
Blackwell.
—— 2006. ‘‘Responsibility and Alternative Possibilities’’. In his My Way:
Essays on Moral Responsibility. Oxford: Oxford University Press.
Frankfurt, Harry. 1969. ‘‘Alternate Possibilities and Moral Responsi-
bility’’. Journal of Philosophy 66: 829–839.
Gettier, Edmund. 1963. ‘‘Is Justified True Belief Knowledge?’’ Analysis
23: 121–123.
McKenna, Michael. 2001. ‘‘Source incompatibilism, Ultimacy, and the
Transfer of Non-Responsibility’’. American Philosophical Quarterly
38: 37–52.
McKenna, Michael and Widerker, David. 2003. Moral Responsibility
and Alternative Possibilities: Essays on the Importance of Alternative
Possibilities. Ashgate.
Plantinga, Alvin. 1974. The Nature of Necessity. Oxford: Oxford Uni-
versity Press.
Shoemaker, Sydney and Swinburne, Richard. 1984. Personal Identity.
Oxford: Basil Blackwell.
Van Inwagen, Peter. 1983. An Essay on Free Will. Oxford: Clarendon
Press.
14 For helpful comments on previous drafts, we are grateful to John Martin Fischer,
Michael McKenna, Kevin Timpe, and an anonymous referee for Philosophy and
Phenomenological Research.
WHY FRANKFURT-EXAMPLES DON’T NEED TO SUCCEED TO SUCCEED 565