A couple of weeks ago I went to this year’s TSC in Budapest. While I didn’t enjoy it as much as I did last time in Copenhagen (though Budapest, even during a heat wave, was quite enjoyable), I liked some of the talks, as in the discussion, you can hear some hidden assumptions that aren’t always spelled out in the printed versions. Now, Stuart Hameroff, as usual, was trying to sell his vision of microtubules realizing quantum computations. But his vision was so computational (he didn’t have much time to dwell on the neutral monism speculation), that he got asked whether he accepts the fact that computational account of consciousness (be it quantum or not) commits him to the multiple realizability about consciousness. He replied that quantum theorists must bite the bullet, and say that in principle, consciousness can be multiply realized.
Now, as I tried to argue in my paper in Budapest, it means that this is a purely computational view, i.e. the view that claims there is no other causal mechanism that would support consciousness. Now, if you have a large enough computer, you can change the scale of the “quantum-brain-computer”, make it compute the right thing, and you’d have consciousness. Of course, there is some air of mystery around quantum computation (not everyone is really sure if it would be hypercomputation, if this is any computation at all), so it’s pretty hard to say why you couldn’t do the right computations on a standard computer (if it’s fast enough), but in principle, or so it seems to me, you could get rid of the mysterious quantum microtubules and use some other standard computing mechanism. If so, then a question arises: why bother with all these quantum computation at all?
So a much safer position for a quantofile is to say that multiple realization is not possible, because you need a special quantum-something for consciousness, and this special something is not computational. (I’m not saying that this position is especially attractive but easier to defend against my argument.)
Hi Marcin, I have no familiarity with the subject matter at hand, but I can venture a couple of comments about multiple realization. (Maybe Gualtiero can comment on the stuff related to computation).
First, the issue about nomological multiple realization as opposed to “in principle” multiple realization is worth noting. The former concerns whether a property is multiply realized by certain other properties given the actual laws of nature. The latter question seems to concern situations not involving the actual laws. I have no idea how we can successfully answer the latter question, so I will stick to the former. (I realize some people think that matter can take whatever form they can conceive, but that approach has been shown to be woefully unreliable when applied to the only matter we have had contact with).
Anyway, the second comment is, as Ken Aizawa and I press in a bunch of forthcoming paper, that multiple realization of a property appears always to be by particular kinds of properties at a certain level. Thus it makes no sense, and can lead to grave confusion, to simply talk of a property being multiply realized. We always need to know multiply realized by WHICH properties and at WHICH level.
This comes up in the case at hand, since we have three levels of properties (at least): the psychological level, some kind of computational level, and the quantum level (though we may have other levels of properties that realize the computational level in addition to the quantum).
So two questions immediately arise: Is consciousness multiply realized at the computational level? And, are the relevant computational states, which putatively realize consciousness under the actual laws, multiply realized by quantum and other lower level states?
I have no idea how to answer those questions for Hameroff or myself. But consider this:
Assume quantum or other states multiply realize the computational states, but certain computational states uniquely realize consciousness. Then consciousness will not be multiply realized — at the computational level; but it will be multiply realized at the quantum level.
On the other hand, assume only certain quantum states are capable of realizing the relevant computational states, and that those computational states uniquely realize consciousness. THEN consciousness will be uniquely realized at both computational and quantum levels. That would be a quantum computational view of consciousness that rejects MR. (Though, again, god knows what to say about the “in principle” issue. Better to ask someone who thinks they can answer those literally fantastic questions).
Hope that was clear and somewhat helpful, best, Carl
Marcin,
It is not clear how the quantophile is supposed to get consciousness to be uniquely realized.
For example, one might simply stipulate that consciousness is quantum mechanical property that can only be realized in one way. Surely stipulation will have the theory turn out to imply that multiple realization is not possible.
But, suppose the theory says only that consciousness is realized by some property found in quantum mechanical individuals. On that theory, it could still turn out, in principle, that the property is found in other kinds of individuals, not described in quantum mechanics.
Or, finally, the theory could be that consciousness is realized by some properties at the quantum mechanical level. That theory could, in principle, be found to have consciousness as multiply realizable.
So, it’s not clear how to get quantum consciousness to be uniquely realized, short of stipulation.
Thanks for this interesting post and thread. A few quick points:
1. As far as I know, quantum hypercomputation is just about as speculative (i.e., unlikely to be physically possible) as any other kind of hypercomputation.
2. A quantum computational theory of consciousness might deny that the quantum computations in question can sustain consciousness when realized classically, because they would lack some quantum feature (such as superposition of states). Perhaps, however, Marcin would count this theory as computational-in-name-only. (And by the way, I don’t know of any reason to maintain such a thesis.)
3. Assuming that quantum computation is not hypercomputational and ignoring views like the one just mentioned above, I agree with what I take to be Marcin’s main point, namely, that the quantum computations in question could also be realized in a classical computer.
4. In response to Carl, I would say that there is no such thing as “the computational level”. There are many levels, some of which may be computational. Maybe there is one computational level, maybe there is more than one, or manybe there are no levels that are computational. It depends on what you are studying and the properties of its levels.
5. For what it’s worth, I think computational theories of (phenomenal) consciousness, whether quantum or otherwise, are a non-starter.
Thanks for all the comments.
1. The levels. Yes, of course, with multiple realization you must be aware of what’s realized and how; the levels enter. I mostly agree with Carl on that. Yet I’m not so sure if the quantum consciousness theorist is ready to accept a multiple-level view. It’s a kind of a strict reductionism, and to talk of levels, you actually need to accept there are real levels. In some version of reductionism, viz. in Heil & Martin, you cannot get any levels. So I’m not so sure if the proponents of quantum consciousness would accept this as a shared assumption. I’m not sure they ever leave the quantum level, for that matter 😉
2. Gualtiero: computational accounts of consciousness are criticized for not taking care of experience, but this is something that quantum theories purport to offer. I share Chalmers’ doubts whether they succeed in that, though I think I can use exactly the same argument as above against his idea of intrinsic qualities of experience as realized on a microlevel – why should those microqualities matter at all? (More on that in my presentation, working on a draft of a draft now).
3. Ken: I think that quantophile could use something else than stipulation; there can be some properties that are nomologically impossible to be realized on a macrolevel or microlevel (electrons cannot be said to be hot or cold, for example). I haven’t come across such an argument anywhere, so probably there isn’t even a candidate for that (but I only browsed the papers, having no time to read all the stuff).