The Mad Scientist Meets the Robot Cats: Compatibilism, Kinds, and Counterexamples

Author(s): Mark Heller

Source: Philosophy and Phenomenological Research , Jun., 1996, Vol. 56, No. 2 (Jun.,
1996), pp. 333-337
Published by: International Phenomenological Society

Stable URL: https://www.jstor.org/stable/2108523

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to Philosophy and
Phenomenological Research

This content downloaded from

70.28.13.33 on Tue, 29 Oct 2024 23:13:35 UTC
All use subject to https://about.jstor.org/terms
Philosophy and Phenomenological Research
Vol. LVI, No. 2, June 1996

The Mad Scientist Meets the Robot

Cats: Compatibilism, Kinds, and
Counterexamples

MARK HELLER

Southern Methodist University

In 1962 Hilary Putnam forced us to face the possibility of robot cats.' More
than twenty years later Daniel Dennett found himself doing battle with mad
scientists and other "bogeymen."2 Though these two examples are employed
in different philosophical arena, there is an important connection between
them that has not been emphasized. Separating the concept associated with a
kind term from the extension of that term, as Putnam and others3 have urged,
raises the possibility of accepting counterexamples to compatibilistic analy-
ses without rejecting compatibilism. Even if no compatibilist analysis of our
concept of free action is acceptable, a compatibilist account of the essential
nature of free action may be.
Putnam has suggested that, in general, conceptual analysis is not the same
as discovering a kind's essential properties. He asks us to consider the possi-
bility of our discovering that all the things that we have been calling cats are
automata controlled by mischievous Martians. Given such a discovery, would
we have discovered that there really are no cats (there are only fake-cats) or
that cats are really robots? The objects in question, being completely mechan-
ical and controlled by creatures on another planet, do not fit our concept of
cat. If we hold (mistakenly, according to Putnam) that the concept determines
the extension, then we should conclude that there really are no cats.
However, Putnam encourages us to accept instead that the extension is de-
termined by paradigm cases, so that a cat is anything that is of the same kind

1 "It Ain't Necessarily So," Journal of Philosophy 59 (October 1962), pp. 658-71. Also
see Putnam's discussion of water, H20, and XYZ in "The Meaning of 'Meaning',"
Mind, Language, and Reality (Cambridge: Cambridge University Press, 1975), pp.
215-7 1.
2 Daniel C. Dennett, Elbow Room: The Varieties of Free Will Worth Wanting
(Cambridge: The MIT Press, 1984).
3 Most notably, Saul Kripke, "Naming and Necessity," in Davidson and Harman, eds,
Semantics of Natural Language (Dordrecht: Reidel, 1972).

THE MAD SCIENTIST MEETS THE ROBOT CATS 333

This content downloaded from

70.28.13.33 on Tue, 29 Oct 2024 23:13:35 UTC
All use subject to https://about.jstor.org/terms
as the paradigm cases. To discover what it is to be a cat we put a cat before
us and ask "what is essential to this thing?" We do not first have to have a
concept of a cat or even know that the thing in question is called a "cat." Dis-
covering the essential nature of the kind becomes an empirical matter. If we
discover that all the paradigm cats are robots, we are discovering something
about cats, since the paradigm cases, and other creatures of the same kind,
cannot fail to be cats. Putnam's treatment of the robot cat example is very
plausible and widely accepted. A similar treatment of free action will allow us
to avoid the standard criticisms of compatibilism.
Consider the following relatively simple statement of what it is for an ac-
tion to be free.

(FA) Necessarily [S's action A is free if and only if: if S wanted to do

A, S would do A, and if S wanted to refrain from doing A, S
would refrain from doing A.]4

If FA is true, then so is compatibilism, because FA allows for S's actions to

be free even if S's desires are completely determined by previous events-as
long as S's actions follow from her desires they are free actions, regardless of
how S came to have the desires she has.
And here is where the mad scientist rears her ugly head. Consider the
imaginary example of a mad scientist who implants a device in S's brain that
allows the scientist to directly control S's desires. The device does not di-
rectly control S's actions; whatever influence the scientist has over S's ac-
tions is in virtue of the influence that she has over S's desires. Though we
would not count S's action as free if it were caused by a desire that was itself
caused by the mad scientist, the right hand side of FA's biconditional is
satisfied. If S had had the opposite desire-if the mad scientist had given S
the desire to refrain from doing A-S would not have done A. Thus, as an
analysis of our concept of free action, FA suffers death by counterexample.
However, demonstrating that FA fails as conceptual analysis does not
demonstrate that it fails to capture what is essential to free action. Suppose
we discover tomorrow that the imagined example above is a true description
of how the world actually is. We have a particular action before us, and we
ask "what is essential to this kind of action." We cannot find that it is essen-
tial to actions of this kind that they not be caused by desires caused by mad
scientists, because it has turned out that these actions are caused in that way.

4 The requirement that if S wanted to do A, S would do A can avoid being trivially

satisfied in every case in which S does do A if we accept weak centering into our coun-
terfactual logic. Weak centering is the thesis that when evaluating counterfactuals with
true antecedents some non-actual worlds in which the antecedent is true may be as close
to the actual world as it is to itself. For a discussion of weak centering, see David
Lewis, Counterfactuals (Cambridge: Harvard University Press, 1973).

334 MARK HELLER

This content downloaded from

70.28.13.33 on Tue, 29 Oct 2024 23:13:35 UTC
All use subject to https://about.jstor.org/terms
By discovering that nothing fits our concept of free action, we would not
have discovered that no actions are free. We would have discovered that free
actions do not fit our concept of free action. This would be parallel to discov-
ering that cats do not fit our concept of cat.
And the situation is no different for determinism than for the mad scien-
tist. Granting that the incompatibilist is right about our concept of free ac-
tion, if we discover that determinism is true we are discovering that nothing
fits our concept of free action. But this discovery would not seem to alter the
fact that there are paradigms of free action. Those paradigms, and any other
actions of the same kind, are free actions even though they do not fit the con-
cept. A priori rumination about the mad scientist no longer has any rele-
vance.5 We must discover empirically what it is to be a free action-what it
is to be one of these acts. It might turn out that what we discover is that FA
is the correct account of what it is to be a free action.
Much of the literature on free will is directed at presenting either coun-
terexamples to proposed analyses or revisions of analyses to avoid counterex-
amples. If we think of free acts as forming a kind and then apply the lesson
of the robot cats, we will see that the energy that has gone into attacking and
defending various compatibilist analyses has been, if not wholly misspent, at
least not as central as had previously been thought. It could even turn out that
our concept is self-contradictory (I suspect this is so), without this in any
way implying that there is no free action or that compatibilism is false.
Similar considerations show that separating the concept of free action
from the essential nature of free action can provide a response to general con-
ceptual arguments that directly support incompatibilism. In recent times such
direct arguments have been forcefully championed by Peter van Inwagen.6 I
will mention just one of his arguments here, but what I say is easily general-
izable to his others. Van Inwagen's modal argument rests heavily on his
principle:

(Ii) Np, N(pDq) I-Nq

where "Np" abbreviates "p and no one has, or ever had, any choice about
whether p."
I find fB extremely plausible; it seems an obvious truth about our concept
of having a choice about something (which is closely connected to our con-
cept of free action). But the fact that j is an obvious truth about our concept

5 We must be careful about what we label "a priori." It may be a priori that if anything fits
our concept of free action then it is not caused by a desire that is produced by a mad sci-
entist. But it is not a priori that our free actions are not caused in this way. For our free
actions may not fit our concepts.
6 Peter van Inwagen, An Essay on Free Will (Oxford: Clarendon Press, 1983), Chapter
III, especially section 3.10.

THE MAD SCIENTIST MEETS THE ROBOT CATS 335

This content downloaded from

70.28.13.33 on Tue, 29 Oct 2024 23:13:35 UTC
All use subject to https://about.jstor.org/terms
of having a choice does not make it an obvious truth about having a choice.
It could turn out that the kinds of actions that are in the extension of "having
a choice" do not fit our concept of someone's having a choice about some-
thing. Even if determinism is true, the actions which are of the same kind as
the paradigm free actions are still cases of having a choice about things. If fB
is intended to be a truth about the nature of having a choice rather than about
the concept of having a choice, then if determinism is true each of the
paradigm cases is a counterexample to f.
The strategy I have been proposing provides defense against well known
criticisms of compatibilism, but it does not guarantee that there actually are
free acts. An act is free if and only if it is of the same kind as the paradigms.
But if there is no single kind of which the paradigms are all instances, noth-
ing counts as being of the same kind. I have little to say here about kind
identification. Being of the same kind is a vague matter. The vagueness is
even more extreme when the kind in question is not a natural kind, and the
notion of kind at work here should not be restricted to natural kinds. If cats
turn out to be robots, they would not form a natural kind, but they would
still form a kind, and "cat" would refer to any members of the kind of which
the paradigm cats were members. Likewise, free acts could form a kind even
if not a natural kind. But if the acts we take to be free form no kind at all,
then "free act" does not refer, and there are no free acts.7
In this brief paper I have argued that a certain strategy popular in other
domains can help the compatibilist in her domain. I do not intend this strat-
egy as an argument for compatibilism. I intend it only as a means to defend
compatibilism against attacks by counterexample. Nothing that I have said
rules out the possibility that what is in fact essential to an action's being free
is that there be some undetermined event in that action's causal history. My
point is that what it takes to discover what is essential to free action is not
conceptual analysis. So general conceptual arguments and arguments by

7 The fact that the present strategy does not ensure that there are free actions distin-
guishes it from the Paradigm Case Argument that van Inwagen considers and rejects in
sections 4.2 and 4.3 of An Essay on Free Will. Van Inwagen's criticism of the
Paradigm Case Argument is based in large part on his ability to produce counterexam-
ples to the most promising proposed compatibilist analyses. Such criticism depends
on the assumption that concept determines extension, and the fact that proponents of
the Paradigm Case Argument do not explicitly deny this assumption marks another im-
portant difference between that argument and the present strategy.
8 And this marks a third important difference between my proposal and the Paradigm
Case Argument. The Paradigm Case Argument was meant to offer positive support for
compatibilism. But drawing this conclusion depends upon confusing epistemic possi-
bility with metaphysical possibility. Assuming that we do not know determinism to be
false and even assuming that we do know that "free act" refers, what follows is that
there is an epistemic possibility of a deterministic world in which our actions are free.
It does not follow that there really is any such possible world.

336 MARK HELLER

This content downloaded from

70.28.13.33 on Tue, 29 Oct 2024 23:13:35 UTC
All use subject to https://about.jstor.org/terms
counterexample do not have the significance for the free will debate that they
have been taken to have.9

9 I received helpful comments on earlier drafts of this paper from Bill Bracken, Jim Hud-
son, Hilary Kornblith, Bill Tolhurst and the Iowa State University Philosophy De-
partment. I read a version of this paper at the APA Pacific Division meetings where I re-
ceived helpful comments from Terry Horgan, Takashi Yagasawa, and my commentator
Richard Warner. I also thank the referees for this journal.

THE MAD SCIENTIST MEETS THE ROBOT CATS 337

This content downloaded from

70.28.13.33 on Tue, 29 Oct 2024 23:13:35 UTC
All use subject to https://about.jstor.org/terms