Thursday, April 3, 2014

The neural prosthetic argument against naturalism

While it is unclear whether my mental functioning could survive my getting getting a prosthetic brain, surely it could survive my getting a prosthetic brain part:

  1. For any 0.5 centimeter cube in my brain and any machine that functions in exactly the same way with respect to inputs and outputs on the cube boundaries as the neural matter did, it is possible that replacing the cube with the machine would not change my mental functioning.
Claim (1) strengthened by removing "it is possible that" is in fact a key argument for functionalism: roughly, one repeats application of the strengthened claim until the whole brain has been replaced by a functional isomorph. So claim (1) certainly doesn't beg the question against functionalism. And it's pretty plausible.

Yesterday I argued that if functionalism is true, basic mental states are perfectly natural. In comments, Brian Cutter offered some excellent criticisms (though I responded back), but even if Cutter's criticisms are right, we still have:

  1. If functionalism is true, the realizers of basic mental states have to be at least fairly natural.
But if we replace a cube of neural matter whose state is a part of the functional realizer of a basic mental state M by a sufficiently complex prosthetic while keeping fixed edge interaction, we can make the corresponding realizer as messy as we like. By (1), mental functioning could be unchanged by this, while (2) tells us that if functionalism is true, we'd have to lose mental state M. So we've argued that
  1. If (1) and (2) are true, functionalism is false.

Now, it is actually pretty plausible that:

  1. If naturalism is true, functionalism is true.
The naturalistic alternatives to functionalism just don't seem great. So, we have an argument against naturalism based on the possibility of neural prostheses.

Anyway, probably any naturalistic alternative to functionalism will be heavily biological in nature. It will tie mental functioning to organic rather than functional features of our brains. And in so doing, it is apt to violate (1) as well. Or at least it will violate a strengthened version of (1) which says that (1) necessarily holds for any mental being whose cognitive organs have the same kind of functional density that our brains have. For the replacement of a cube by a prosthetic need not change functional density, and then one could do a second replacement, and continue. Finally, by S4 one would conclude that it is possible that mental functioning could continue after total prosthetization of the brain, which would violate the organicity of our naturalistic alternative to functionalism.

So, surprisingly, gradual replacement considerations may favor dualism, not functionalism.


William said...

What reasons are there to accept the sorites argument in the third paragraph?

If the sorites argument is invalid, one might expect that as the biological components are replaced that the system as a whole would change (deteriorate) in an increasing way, even though each replaced component would seem to to working as intended when looked at in isolation.

Alexander R Pruss said...

Well, (1) says that there could be *no* change in mental functioning. It would be odd (contradictory!) that there would be a number of small steps, none of which produced *any* change in mental functioning, but suddenly after the last the mental functioning was different.

William said...

The remaining issue which might cause progressive drift from normal (allowing (1)) is the dynamic nature of the system. Any given cm3 of the brain is probably changing constantly enough so if we measured it over time it would show a change in functioning over time that we certainly cannot predict currently.

So we might not know if the future changes in the substitute, or lack thereof, would duplicate the unreplaced future functions of the tissue.

Alexander R Pruss said...

Then suppose determinism is true. It would be odd if determinism were incompatible with consciousness.

William said...

True. A deterministic machine is a functionalist one.