Hmm, upon a further re-reading of the Chalmers paper I cited, I see that where he and I differ is in his continual use of something that your run-of-the-mill physicalism explicitly disallows (except for MWI coherence, as I mentioned): the woulda-coulda-shoulda's of life.
If p, then q. If p happened to me (e.g., I was playing chess, with the board in a specific configuration, and it was 5:37:28 PM in a room with three windows, with sunlight streaming in at such-and-such an angle, and the room was 73.26 degrees Fahrenheit, and there was a glass of water with condensation in a certain pattern on it to my left, and the game pieces were in a certain precise orientation (e.g., maybe my knight on G4 was facing 32.6 degrees left of center, away from me, and was offset 1.27 mm away and 3.45 mm to the right within the square), etc., etc.), then I would do q (e.g., wipe my brow, grab the glass of water and take a sip, and then move my knight to F6 and declare, "Check!").
Chalmers is saying that, what if p hadn't happened to me. What if, say p* happened to me. For example, what if the knight on G4 had been facing 32.5 degrees left of center, or 30 degrees, or 10? What if it had been facing right? What if it was ten degrees warmer in the room (which arguably would affect my comfort, and hence interfere with my concentration)? Then I would have done q*. In other words, I might still have moved my knight to F6, but I might have said "Check!" with more or less conviction, louder or softer, or perhaps with a more suspenseful pause. I might even have moved a different piece altogether, foregoing the check opportunity for a better position (or, just missing the opportunity altogether). I might not have wiped my brow (had the room been colder), or I might have taken a gulp rather than a sip (had the room been warmer). The tactile interaction with the condensation pattern on the glass might even affect my actions.
In other words, at that instant in time, p->q, but p*->q*. For a given arbitrary set of physical processes in a rock for which p->q is true (an easy condition to satisfy), it is overwhelmingly unlikely that for any such p*, p*->q*. In fact, across the full spectrum of p*_i's, we could say p*_i->q*_i, and this couldn't be done in a rock: there just isn't enough complexity.
A system which could replicate all the possible q*_i's, given all the possible p*_i's, would pass Chalmers's more stringent test, and as such, a lookup table would suffice, but would be larger than the universe, so it's not physically possible. On the other hand, a software representation of my neural networking in my brain might just suffice, and hence, on this basis, Chalmers (and I presume functionalists in general) would say that such a software-based neural network would be conscious, whereas the rock wouldn't.
However, this is the deadly trap Chalmers allows the functionalist to fall into. You see, in real life (under physicalism, minus MWI), p* doesn't happen. We could test an arbitrary number of p*_i's (by resetting the system to the original state, and then running it again with a different input), but doing so doesn't lead to combinatorial explosion: it leads to linear explosion. Add 10 test cases of inputs, and your table only needs 10 extra outputs. In theory, then, consciousness is NOT the possibilities, the woulda-coulda-shoulda's. Instead, it's just outputs. It's just bitstrings of a certain complexity. This of course trivializes into ultra-panpsychism (not just that grossly functional objects, like thermostats, are conscious, but every physical process, including the random motions of heat).
That everything is conscious, and more richly conscious than our minds, trivializes our own existence. Why balk at killing a person, if the entire bitstring representing the rest of their life (assuming determinism, ignoring QM) naturally occurs a near infinite number of times in nature if you slice it right and define the states correctly. That bag of blood and neurons, etc., that you "killed" was just one instance of a nearly ubiquitous consciousness, and that "murder" would be as wrong as stirring a cup of tea that might otherwise have instantiated the same complexity. It might not be directly contained in the cup of tea, depending on how much time the tea needs to replicate that complexity. Assuming the cup of tea isn't big enough, then at least including the interactions of the tea with your GI tract, bloodstream, kidneys, urine, and trip through the sewer system, into the ocean, until the complexity is sufficient. Given that the heat complexity of a cup of tea is orders of magnitude more complex than the neural networking of a human brain, we correspondingly would need data from a time period orders of magnitude smaller, perhaps only days, hours, or minutes. Of course, in stirring the cup of tea, you might destroy this instance of this consciousness (which would be murder), but you might instantiate a different consciousness. On the other hand, if you murder a real person, that's all you did, right? Wrong. The person's body, both as an object with heat, and as an object which will decompose (yet another physical process), will contain extreme complexity as well, so that it's pretty much a wash, break-even. Some indeterminate number of consciousnesses (including the one we presumably would attribute to the "person" killed) would be prevented, and some roughly equal number of consciousnesses would be created.
No, this is far too absurd. Something more than bitstrings must be responsible for consciousness, if we are to assign any value to it. Of course, we could drop functionalism and stop worrying about it (in which case, saying that a silicon brain must have consciousness because it's functionally identical is an unfounded claim), or we have to worry about it, about the sheer absurdity of it.
Without dropping functionalism, then, perhaps, as Chalmers points out, we must account for the woulda-coulda-shoulda's. But that's not functionalism in a deterministic universe. That's applying something more. Perhaps that something more is Cartesian dualism. Perhaps that something more is MWI coherence. I used to be a Copenhagen man, but MWI is seeming more and more likely, with coherence of extremely-close branches providing the "dualistic" nature. It's not confirmable, in the sense that the Copenhagen and MWI interpretations are equally valid empirically (here again, we're talking about the question of whether we're running on a Pentium or an Analytical Engine), but under Occam's razor, MWI is far more plausible (consciousness is no longer trivialized to meaninglessness, in that rocks aren't conscious, because they can't replicate the functionality in a deterministic universe with MWI). And it still allows for functionalism, just not in an objectively verifiable way. In effect, it makes our objective universe just one more "subjective" observer, viewing only its slice, its branching, within MWI.
Of course, this leads back to the question of a silicon brain. If we assume MWI is true, and we require functionalism to be true, then a rock can't be conscious, but perhaps software can be. Not by virtue of what the software does functionally within the objectively viewed universe, but because of the totality of MWI splits, in which the software performs functionally like an organic brain under all circumstances. Well, all ordinary circumstances. It won't react to the flu the same way a real brain will, and it doesn't functionally replicate a brain when a PET scan is run (i.e., the PET scan will be able to tell the difference). But as far as responding to the physical senses (which is our baseline for consciousness, not how it looks in a PET scan), the functionality is there, so it must be conscious.
Something for me to think about. I'm still not 100% convinced that software can be conscious under even MWI, but I'm more convinced than ever that it cannot be conscious without MWI (or some other version of physicalism where the woulda-coulda-shoulda's are taken into account). Once again, we see consciousness and quantum mechanics being linked in some way. Yet it's not necessarily that one is mysterious and so is the other, so they must be related. It's that analysis of one generally leads to analysis or consideration of the other. That correlation, not based on mystery but on function, is what keeps bringing QM to discussions of consciousness.
LongeCityAdvocacy & Research for Unlimited Lifespans
