...worlds.
Prima facie conceivability, according to Chalmers, is sometimes prone to human failures, but ideal conceivability (conceivability under ``ideal rational reflection'') is not.
...concepts.
Since I don't find the distinction terribly useful, from now on in this paper, I will not distinguish between imagining and conceiving of---I will use the two terms interchangeably.
...all.
We should, of course, however, be especially suspicious of intuitions that are cognitively similar to intuitions we know to be unreliable.
...otherwise.
There is a reply to Hill and McLaughlin from Chalmers on this issue, but I am going to skip Chalmers' reply in an effort to press on to more engaging debates.
...necessities'',
A strong necessity, according to Chalmers, is a non-Kripkean a posteriori necessity; that is, an a posteriori necessity with a necessary primary proposition. In contrast, a Kripkean a posteriori necessity, according to Chalmers, has a contingent primary proposition and a necessary secondary proposition. A ``primary proposition'' is what is expressed by a sentence when evaluating the terms involved using their primary intensions. A ``secondary proposition'' is what is expressed by a sentence when evaluating the terms involved using their secondary intensions.
...object
Well, this isn't quite right because many possible worlds are, no doubt, quite small. For the most, part, however, the ones we care about will be very large.
...elegant?
Lewis feels that it is the number of different kinds that count, when tallying up the ontological cost, and that postulating more things of a kind that we already suppose exists has little or no ontological cost. I take it that most of us feel differently, however, especially when it concerns beings with phenomenal states and moral status.
...significant.
Chalmers might object that there is no ontological cost to CPT because even if it posits more possible worlds, possible worlds don't exist, so there's no ontological cost involved. Furthermore, if we are to consider this an ontological cost, then surely we already suffer a huge ontological cost when we deviate at all from talk of only the actual world---that is even when we start talking about nomologically possible worlds, and all the more so for talk of metaphysically possible worlds. In comparison to the costs we all agree to incur, the extra cost is modest.

I am going to leave this debate for another day.
...influence.
It appears that Hill and McLaughlin have summarily dismissed interactionism as extremely unlikely. For the purposes of this paper I will go to along with this, though perhaps such a quick dismissal is unfair.
...deterministic.
For the sake of brevity, I'm ignoring the complication of quantum randomness, which does not seem able to rescue free will, unless perhaps it is fit into an interactionist theory. I also don't mean to disparage compatibilism, except to assert that the resulting sense of free will is not quite the same as the sense with which we started.
...behavior.
Perhaps a bit of clarification is warranted. I have said that our intuitions tell us that a zombie duplicate world is possible, and I have also said that our intuitions tell us that zombie duplicates of us as individuals are impossible. How can both of these assertions be true? Our intuitions about phenomenal causation really direct us toward interactionism. It is only once we have rejected interactionism, and have tentatively accepted the thesis of physicalism that a zombie duplicate world seems possible.
...behavior.
 The most philosophically straight-forward (though also the least practical) way to to make a functional duplicate of a person using a digital computer, would be to determine the position and momentum of every fundamental particle in the person's body. This information would then be fed into the computer, along with a program designed to ``perfectly simulate'' the microphysical laws of the world. That is, given the positions and momentums of a set of fundamental particles at time t, the program will calculate the positions and the momentums of the particles at , , , etc. It might be argued that a physical system cannot have its causal structure duplicated in this manner because the world is inherently analog and information will be lost by a digital computer due to its limited precision in arithmetic calculations and in time quantization. Surely, however, there is some level of limited-precision arithmetic which would be close enough to the real thing to be good enough. For instance, if every Real number were represented with a google digits of precision, the computer simulation would not stray measurably from reality within the lifespan of the universe. Mutatis mutandis for time quantization.
...world.
For a more detailed explanation of my notion of ``perfect simulation'' see footnote 12.
...possibilities.
In the case of the antimatter world, this is not quite right. Until the 1950's it was believed that nature would be unaffected if particles were all replaced by their antiparticles. It was discovered, however, that right and left would also need to be exchanged in order for there to be no observable change. This symmetry relationship is known as CP. More recently it was discovered that the direction of time would also need to be reversed for there to be no observable change. This symmetry relationship is known as CPT. Without reversing the direction of time, the only noticeable difference would be in a very obscure form of radioactive decay.
...identification.
I am told that Hill has a whole book defending type-materialism. If he addresses the above objections in it, it might be an interesting future project to consider his replies.
...intensions.
Hill and McLaughlin use the term ``reference-fixing properties'' instead of ``primary intensions'', but I am taking the liberty of translating back into Chalmers' terminology.
...CPT.
It should also be noted that if we can show that there are strong necessities, it also shows that cosmic hermeneutics is impossible even if physicalism is true, and consequently it removes one of the premises of the knowledge argument, rendering it unsound.
...does.
It might be argued that AI is an a priori venture, and it may yet succeed. If it does, however, we'll never know without a posteriori research that the computational processes devised by AI researchers are the same as those implemented by our brains. And if they aren't, the two may give different results in some situations.
...program.
A potential objector might claim that cognitive psychology is only needed because, although the problems to be solved are solvable with a priori methods, they are too difficult for us to solve with our limited cognitive abilities without using a posteriori methods. This objection, once again, seems to postulate a superhuman notion of conceivability that I will address shortly.
...stuff
``Watery stuff'' is a term that is meant to have the same primary intension as ``water'', but its secondary intension is the same as its primary intension. That is, its secondary intension isn't rigidified, as is the secondary intension of ``water''.
...convincing.
I haven't yet read ``The Paradox of Phenomenal Judgment'', which is Chapter 5 of Chalmers' book The Conscious Mind. It's quite possible that there is a reply to my argument, or to something quite like it, to be found in this chapter. This suggests another future project.
...either.
I do think a carefully articulated functional identification, as I attempted to provide in a previous paper, leaves it very difficult to say just what is supposed to be missing in a putative zombie duplicate.

Douglas Alan
Wed Sep 23 21:01:36 EDT 1998