Running Sydney Shoemaker's argument about the impossibility of p-zombies in reverse

This is a blog post about the logical and metaphysical possibility of p-zombies. I don't think its conclusion has a strong bearing either way on the nomological possibility of p-zombies. In fact I think the fading qualia argument provides strong evidence that p-zombies are not nomologically possible. I write about these terms for describing different kinds of possibility in the philosophy of mind section of this post.

In a 1975 paper titled "Functionalism and Qualia" the philosopher Sydney Shoemaker presented an argument that p-zombies are logically (and therefore metaphysically) impossible. In this post I lay out Shoemaker’s argument as I understand it, and then argue why I think it does not work. Then I'll argue that, in fact, one can run a modified version of it as part of an argument against certain materialist positions.

Here is my reconstruction of Shoemaker’s argument:

• Suppose for contradiction that the absent qualia scenario is possible (i.e. that p-zombies are possible).
• While our p-zombie counterparts may lack the conscious experience of pain, when you pinch them they still believe that they are pain since:
  • Beliefs are mental states defined in terms of our behaviors or dispositions.
  • p-zombies share all of our behaviors and dispositions.
• But then if absent qualia are possible, it is possible to believe that you are in pain without actually being in pain.
• If it is possible to believe that you are in pain without actually being in pain, then we cannot be certain that we are in pain when we are being pinched.
• But we are certain that we are in pain when we are being pinched.
• So the absent qualia scenario is not possible.

This is an especially strong example of a response to arguments about p-zombies. Most other responses grant that p-zombies are conceivable, but argue that this is not sufficient to deploy them to draw any useful conclusions about what kinds of worlds are metaphysically possible. Shoemaker’s argument on the other hand would show that p-zombies are not a coherently conceivable concept in the first place. Indeed the argument appears to show that p-zombies are impossible a priori. As a result anyone relying on the conceivability of p-zombies would be guilty of the same fallacy as anyone who tries to argue that e.g. it is metaphysically possible that Fermat’s Last Theorem is false. Sure, you can conceive this to be the case in a very abstract sense, but it's only as a result of lacking the capacity to properly think through the relevant proofs and consequences.

I think the argument fails but does so in a nevertheless interesting way. My main issue is that I don't share Shoemaker's intuitions about how p-zombie beliefs work. In particular while some beliefs might supervene only on behaviours and dispositions (and therefore are possessed by p-zombies too), I think beliefs that are about our own conscious experiences contain what philosophers call phenomenal concepts¹ as part of their mental content. The beliefs held by philosophical zombies are different in mental content to the beliefs held by us because p-zombies lack phenomenal concepts.

Here's an example that illustrates this point based on a related kind of mental state: desires.

If I desire to feel a certain way, and my corresponding p-zombie twin has an analogous desire, there are two ways interpreting what happens when my desire is fulfilled, either:

• The p-zombie’s desire to feel a certain way was identical to mine, and therefore was not fulfilled, because it can’t actually feel anything.
• The p-zombie in fact had a different desire to mine, the p-zombie’s desire was just a desire to behave as if it had felt something, and that desire has been fulfilled.

I think the second of these is the only one that makes sense, since p-zombies lack phenomenal concepts - but these concepts form an essential part of our own desires regarding feelings. I think we should also adopt an analogous interpretation for p-zombie-beliefs: a p-zombie’s belief that it is in pain is a true belief, since for a p-zombie the conceptual content of that belief contains only the concept of behaving in certain ways, rather than actually feeling anything. And so for p-zombies, while it can be helpful to describe them as being in pain under what Dennett calls the intentional stance, this description has no phenomenal content: it doesn't actually feel like anything to be a p-zombie in pain.

The reason I think this argument is especially interesting is because if you agree with my case against the second premise then you can run the rest of the argument as an argument against certain materialist theories. In particular, any theory which shows that p-zombies would have phenomenal concepts with identical content to ours forming parts of their beliefs would be failing to account for how we can be certain that we feel the feelings we claim to.

The question of whether p-zombies (as conceived) have phenomenal concepts or not appears in this 2006 paper by David Chalmers as a crux for one line of defence that a materialist could take in response to that paper’s “master argument”. In the paper Chalmers takes a narrower approach to rebutting this line of defence which is based on the adequacy of phenomenal concepts to explain Mary’s change in epistemic state which takes place in the thought experiment described in the standard version of the knowledge argument.