Monday, November 22, 2021

Functionalism implies the possibility of zombies

Endicott has observed that functionalism in the philosophy of mind contradicts the widely accepted supervenience of the mental on the physical, because you can have worlds where the functional features are realized by non-physical processes.

My own view is that a functionalist physicalist shouldn’t worry about this much. It seems to be a strength of a functionalist view that it makes it possible to have non-physical minds, and the physicalist should only hold that in the actual world all the minds are physical (call this “actual-world physicalism”).

But here is something that might worry a physicalist a little bit more.

  • If functionalism and actual-world physicalism are true, there is a possible world which is physically exactly like ours but where there is no pain.

Here is why. On functionalism, pain is constituted by some functional roles. No doubt an essential part of that role is the detection of damage and the production of aversive behavior. Let’s suppose for simplicity that this role is realized in C-fiber firing in all beings capable of pain (the argument generalizes straightforwardly if there are multiple realizers). Now imagine a possible world physically just like this one, but with two modifications: there are lots of blissful non-physical angels, and all C-fiber equipped brains have an additional non-physical causal power to trigger C-fiber firing whenever an angel thinks about that brain. It is no longer true that the functional trigger for C-fiber firing is damage. Now, the functional trigger for C-fiber firing is the disjunction of damage and being thought about by an angel, and hence C-fiber firing no longer fulfills the functional role of pain. But now add that the angels never actually think about a rain while that brain is alive, though they easily could. Then the world is physically just like ours, but nobody feels any pain.

One might object that a functional role of a detector is unchanged by adding a disjunct to what is being detected. But that is mistaken. After all, imagine that we modify the hookups in a brain so that C-fiber firing is triggered by damage and lack of damage. Then clearly we’ve changed the functional role of C-fiber firing—now, the C-fibers are triggered 100% of the time, no matter what—even though we’ve just added a disjunct.

We can also set up a story where it is the aversive behavior side of the causal role that is removed. For instance, we may suppose that there is a magical non-physical aura normally present everywhere in the universe, and C-fiber firing interacts with this aura to magically move human beings in the opposite direction to the one their muscles are moving them to. The aura does nothing else. Thus, if the aura is present and you receive a painful stimulus, you now move closer to the stimulus; if the aura is absent, you move further away. It is no longer the case that C-fibers have the function of producing aversive behavior. However, we may further imagine that at times random abnormal holes appear in the aura, perhaps due to a sport played by non-physical pain-free imps, and completely coincidentally a hole has always appeared around any animal while its C-fibers were firing. Thus, the physical aspects of that world can be exactly the same as in ours, but there is no pain.

The arguments generalize to show that functionalists are committed to zombies: beings physically just like us but without any conscious states. Interestingly, these are implemented as the reverse of the zombies dualists think up. The dualist’s zombies lack non-physical properties that the dualist (rightly) thinks we have, and this lack makes them not be conscious. But my new zombies are non-conscious precisely because they have additional non-physical properties.

Note that the arguments assume the standard physicalist-based functionalism, rather than Koons-Pruss Aristotelian functionalism.


Brian Cutter said...

These are great cases. I wrote an paper a while back arguing along related lines that physicalists can accommodate the possibility of zombies. (It's unpublished, and I set it aside because I decided I didn't like the paper much.) My strategy was a bit similar, in that I argued that physicalists can allow for the possibility of zombies in worlds that *add* ingredients to the actual world, rather than by substraction. But I hadn't considered cases quite like the ones you describe, and I think I like your cases better. (One of my cases was this: imagine a world just like ours, but in which your counterpart, Alex* has a non-physical part that plays a neuron-like functional role, which is functionally redundant to some physical neuron. (So, Alex* is like your body plus this extra non-physical thing.) Then consider Alex-minus, the largest proper part of Alex* which is purely physical. (So, Alex* minus the non-physical part.) Alex-minus is a perfect physical duplicate of you-as-you-actually-are. Given Sider's assumption that consciousness is a maximal property (large proper parts of a conscious thing are excluded from being conscious in virtue of being large proper parts of something else that's conscious), it follows that Alex-minus is not conscious; so, a zombie.

swaggerswaggmann said...
This comment has been removed by a blog administrator.
Alexander R Pruss said...


I like your cases, too.

Maybe we should write a paper together.

I wonder whether the results are favorable to functionalism or unfavorable.

Logically speaking, it should be favorable: it helps handle zombies, allows for variant qualia with the same physical base, etc. (Probably doesn't help with the knowledge argument. Mary will have to know, in addition to the physical stuff, that physicalism is true. But that doesn't make a sigificant difference.) Similarly, the Endicott observations on supervenience help functionalists handle the conceivability of an afterlife.

However, socially, I think the idea that functionalism is compatible with zombies and afterlife make functionalism less attractive. Lycan in his piece on substance vs property dualism thinks that the compatibility of substance dualism with an afterlife makes substance dualism less attractive. (Odd, I agree.) And Endicott thinks that the loss of supervenience is a reason to modify functionalism.

Brian Cutter said...

Yeah I was thinking the point should help functionalists, but only modestly. (You're probably right on the sociological point though.) There are strong intuitions that zombies are possible, and these cases would show that functionalists can accommodate those intuitions. Still, I don't think they help a huge amount against zombie arguments. Even if functionalists can accept the possibility of zombies, they won't be able to accept zombies worlds that are minimal physical duplicates of our world. Such worlds seem to be conceivable (indeed, these are the sorts of worlds most people have in mind when they try to conceive of zombie worlds). So it seems easy to tweak the conceivability/zombie argument to get around these cases.

I suppose one could say that the straight "zombies are possible" intuition is on firmer footing than the "minimal zombies are possible" intuition, since the latter makes demands on the totality of reality in a way the former doesn't. Relatedly, some think there's something problematic about conceiving of absences/negatives, which "minimal" here would imply. But even the standard zombie scenario involves conceiving an absence of consciousness, so arguably the two are on a par.

Alexander R Pruss said...

The general lesson is that functionalists are rather more like dualists that people think:
- zombies: possible
- knowledge gained by Mary: sort of (Mary would have to know that physicalism is true, not just all of physics, to know what she'd experience)
- inverted spectrum: maybe not, but some other qualia transformations, yes
- disembodied existence after death: possible
- supervenience: no.