Can the Zombie Thought Experiment Support Physicalism?

Kathryn Kanuck Lammens

University of Florida

ktkanuck@ufl.edu

Chalmers's zombie thought experiment has led to two different reactions: some philosophers say it shows that physicalism is false, and others say it does nothing of the sort because in general we cannot rely on conceptual analysis in arriving at metaphysical conclusions. I reject the view that conceptual analysis is not useful in arriving at metaphysical conclusions. I argue that close attention to the details of the zombie thought experiment supports the view that physicalism is a real option in responding to the mind-body problem. To defend this, I (i) look into what the conditions are for proper performance of a thought experiment and explain the relevance of background information to proper performance, and (ii) draw an analogy between Chalmers's zombie thought experiment and a version of Putnam's Twin Earth thought experiment to (iii) draw attention to a mistake the anti-physicalist makes in her performance of the zombie thought experiment.

Chalmers's zombie thought experiment has led to two very different reactions. First, there are those who say that it shows that physicalism is false. Second, there are those who say that it does nothing of the sort because in general we cannot rely on conceptual analysis in arriving at metaphysical conclusions. In this paper, I reject the view that conceptual analysis is of no use in arriving at metaphysical conclusions. But I also argue that close attention to the details of the zombie thought experiment actually support the view that physicalism is a real option in responding to the mind-body problem. To defend this, I first look into (i) what the conditions are for proper performance of a thought experiment and explain the relevance of background information to proper performance, and then (ii) draw an analogy between Chalmers's zombie thought experiment and a version of Putnam's Twin Earth thought experiment to (iii) draw attention to a mistake the anti-physicalist makes in her performance of the zombie thought experiment.

If we are to rely on thought experiments to help us draw metaphysical conclusions, we must make sure that they are performed correctly. A thought experiment that is not properly performed will likely give us misleading and/or confused results. When we properly perform a thought experiment we usually get a reliable result from which we can draw metaphysical conclusions. The following are plausibly the conditions under which a thought experiment is properly performed:

There is not space here to defend this in detail, so I perforce state the conditions without a full defense in order to focus on the one of special interest to our current topic.Sometimes when we properly perform a thought experiment, we give an incorrect answer to the canonical question because there is something about the concept or the thought experiment itself which prevents us from reliably drawing metaphysical conclusions from its results. We can do everything right in performing a thought experiment and still get the wrong result because various other factors may run interference. I call proper performance that gets a true answer to the canonical question "successful performance."

I will argue that although the anti-physicalist properly performs the zombie thought experiment, she gives the wrong answer to the canonical question, i.e., she does not successfully perform it, because she lacks background information, the having of which could result in changing her answer to the canonical question. In particular, the background information may eliminate some background assumptions she makes that may not be true, but which may influence her response to the thought experiment.

For the purposes of this paper, I am most interested in (iii). For we can see the relevance of background information to proper performance of a thought experiment by focusing on what it is to fully possess a concept. Surely someone who participates in a thought experiment should fully possess the concept that the thought experiment is about, if the thought experiment is to be of any use. I will explain concept possession in terms of the abilities I think a participant in a thought experiment should have. I propose the following conditions:

A bit of clarification is needed to explain the two different ways that it can be indeterminate whether a concept applies. For some thought experiments, even when a target scenario is relevantly completely described, it is open whether or not it applies, i.e., it is objectively indeterminate. This happens when a target scenario describes a truly borderline case for a vague term whose application conditions we are interested in. Epistemic indeterminacy can be understood as the following: given what A knows about the actual world, it is indeterminate whether C applies. It can be epistemically indeterminate whether a concept applies, for example, if there is something about the world that A does not know, but which can affect C's application. A participant may not always be able to tell when it is epistemically indeterminate whether a concept applies, but she should be sensitive to the possibility of epistemic indeterminacy. If she is presented with different stipulated background information, she may answer the canonical question differently, and then reconsider her response to the original presentation of the thought experiment. Epistemic indeterminacy is primarily an issue in thought experiments involving natural kind concepts.

The role of background information in proper performance becomes clear when we spell out the modal element of clause (ii) above, namely, A possesses C iff A can say accurately whether a concept applies, fails to apply or it is objectively indeterminate or epistemically indeterminate whether C applies. The modal element of (ii) leaves room for us to specify some conditions under which someone who possesses C must be able to judge accurately whether C applies. I am interested in only one here: A must be able to say correctly whether C applies, etc., only if she is in an epistemic position make judgments about C's application. Sometimes being in the right epistemic position requires being provided with relevant information about the actual world. This is exactly what happens when we inquire about the application conditions for a natural kind concept such as that of beryllium. When there is information about the actual world that may be relevant to evaluating a target scenario, the experimenter should provide the participant with what I will call "stipulated background information." It is a lack of stipulated background information in the zombie thought experiment that (I suggest) causes the anti-physicalist to respond to the zombie thought experiment incorrectly.

Stipulated background information is information about the actual world (whether accurate or merely supposed) that may help a participant in a thought experiment answer a canonical question. It may also help the participant in a thought experiment to understand correctly whatever she has in fact successfully imagined by displacing any possibly false background assumptions she may rely on in responding to a thought experiment.

Information about the actual world is not usually relevant to what we will say in response to the canonical question in a thought experiment. That is, nothing about the actual world is usually relevant to the successful performance of a thought experiment. For some concepts, however, what the actual world is like is relevant to their application conditions, paradigmatically natural kind concepts. Once we know that water is actually H2O, for example, we cannot strongly conceive of water being anything other than H2O, where being able to strongly conceive of p means we can imagine that, given the way things are in the actual world, it could have been that ~p. We cannot strongly conceive of XYZ being water because once the facts are in we see that necessarily, H2O is water. If we cannot strongly conceive of XYZ being water, then XYZ is not water. However, we can weakly conceive of water being XYZ because we can conceive of us being wrong that water is H2O. It is weakly conceivable that p iff what we can determine a priori leaves it open that p though further information about the actual world may change that. Sometimes successful performance of a thought experiment requires that the participant be given stipulated background information.

Although stipulated background information may help minimize epistemic indeterminacy, it will not always lead to results that give us answers to questions about whether the target scenario presents us with a situation which shows the correct application of the concept. If, for example, the experimenter tells the participant to suppose that the actual world is a certain way, but it is not clear that the actual world really is that way, the answer that the participant gives to the canonical question likely will not tell us anything directly about the application conditions for the concept determined by the actual context. Why? Because there is no telling whether what we were asked to suppose is true about the actual world. Supposing that the actual world is a certain way in cases where we do not know what the actual world is like in that respect may help lend plausibility to certain positions we otherwise might think thought experiments rule out, as I will explain in my evaluation of the zombie thought experiment. Furthermore, it may help a participant realize that without stipulated background information, it is epistemically indeterminate whether a concept applies. If information about the actual world can affect the results of thought experiments, as I argue it does, then we need to take a closer look at the anti-physicalist results of the zombie thought experiment.

In this section, I will lay out a variation of Putnam's Twin Earth thought experiment and the zombie thought experiment and evaluate our responses to them. Then I will draw an analogy between the two and draw attention to a mistake the anti-physicalist makes as a participant in the zombie thought experiment.

The following is a variation of the Twin Earth thought experiment, call it TE*, which, I argue, is analogous to the zombie thought experiment:

Canonical question: Is there water on Twin Earth?

The answer that one gives to the canonical question may seem to depend at least in part on what further information one has about the actual world, i.e., the context one is in, and, perhaps, Twin Earth. Suppose that the participant has no further information about the actual world that might be relevant to the thought experiment; suppose, for instance, that it is 1750, and no one knows that what Earthlings call 'water' is H2O because it is prior to the development of atomic theory. It seems that many participants will say, "Yes, there is water on Twin Earth." After all, what we call 'water' and what Twin Earthlings call 'water' share all of their superficial properties and function in the same way. Although there may be a natural tendency to answer the question this way, it is not the correct reflective response of someone who is paying attention to what is going on. The proper response to this canonical question is "I don't know," or "It is epistemically indeterminate," because we do not know whether the stuff on Earth and the stuff on Twin Earth is the same stuff. When provided with the relevant background information, epistemic indeterminacy should be eliminated, and the participant should say that there is no water on Twin Earth.

Now that we have evaluated TE*, we should do the same with the zombie thought experiment and see what has gone wrong. The following is one way we may present the target scenario of the zombie thought experiment:

Canonical question: Must the being be conscious?

If the zombie thought experiment is analogous to TE* and we accept the Kripke/Putnam account of a posteriori necessities, then the answer one gives to the canonical question should depend on what further information one has about the actual world. Since we do not know what property the concept of pain picks out in the actual world, we are in the same position that those in 1750 were in with respect to water, and, so, we might say, an ideal participant should answer, "I don't know." However, most participants answer, "No, the being could fail to be conscious." That is, it seems that we can conceive of a being that is functionally identical to us, etc., which fails to have conscious experience.

It is fairly straightforward to see how answering this way can allow us to infer that physicalism is false. If, given the target scenario, it looks as if the being need not be conscious, then one can conclude that zombies are conceivable. If zombies are conceivable, then, one may argue, zombies are possible. If zombies are possible, then physicalism is false. The physicalist must take seriously the intuition that the being described in the target scenario need not be conscious despite being physically identical to us, since it seems that it is one that most ideal participants have.

One who thinks that the Kripke/Putnam explanation of a posteriori necessity is correct may argue that the reason TE* and the zombie thought experiment are likely to yield the wrong result is that the participant does not have the relevant empirical information that would enable her to describe correctly what she has successfully conceived of when she imagines the target scenario. As a result, it looks as if zombies are conceivable. Further, an ideal participant may fail to be sensitive to the relevance of background information, simply because she has not been exposed to a variety of different thought experiments which differ on that parameter (unlike in the case of the Twin Earth thought experiments). However, if the participant is presented with variations of the zombie thought experiment that contain different stipulated background information, she may become sensitive to the significance of certain background assumptions she actually brings to bear in her answer to the original canonical question.

I say this in part because it seems that if a participant were presented with stipulated background information, e.g., the thing that plays the pain role is actually identical with the property of being in a certain neurological state, then she would reconceive the zombie scenario and change her answer to the canonical question to "Yes, the being must be conscious." She will realize that she originally made some mistake. The physicalist can argue that the participant in the zombie thought experiment is trying to conceive of something that is not possible, but fails to realize it until presented with the relevant background information.

Here is one diagnosis of why we tend to get an anti-physicalist result to the zombie thought experiment: Suppose that any competent speaker knows a priori the following conditional (where 'C' is background information, 'M' is a phenomenal concept, and 'P' is a phenomenal property):

(1) If C, then M refers to P.

When we stipulate the background information that in the actual world C is the case, the participant changes her answer to the canonical question about the zombie, who has P. But, it may also be the case that the participant knows the following conditional a priori:

(2) If C*, then M does not refer to P.

If the participant assumes that C* is true in the actual world, the assumption may go unnoticed by the participant. If this is what is going on in the zombie thought experiment, then, unless we ask the participant to consider other related thought experiments which differ in stipulated background information, she will answer the canonical question in accordance with her background assumption, which may not be true. As a result, the thought experiment will not be successfully performed. Someone who fully possesses the concept in question may easily make this kind of mistake: she cannot say accurately about the zombie target scenario whether the being must be conscious because she is not in the correct epistemic situation, and, even worse, she may be relying on a false background assumption in answering the canonical question.

I do not pretend here to offer an argument that this is actually what happens in the zombie thought experiment; that would take more time than I have here. What I do argue is that since it is a genuine possibility that the anti-physicalist is relying on a possibly false background assumption in responding to the zombie thought experiment, we can call into question the apparent anti-physicalist result of the zombie thought experiment. Failure to have the relevant empirical information may lead a participant to misunderstand what she has successfully conceived of. Since we do not know what property the concept of pain picks out, it does not look as if issues about the truth of physicalism are resolved by the results of the zombie thought experiment.

Kripke, Saul. (1980) Naming and Necessity. Cambridge, MA: Harvard University Press.

Putnam, Hilary. (1975) "The Meaning of Meaning." Philosophical Papers, Vol. 2: Mind,

Language and Reality. Cambridge: Cambridge University Press, p. 215-271.