Some more questions: Philosophy Forums

Philosophy Forums » Guest Speakers » David Chalmers » Some more questions » Some more questions

Some more questions

bemoosed
Initiate

Usergroup: Members
Joined: Mar 29, 2009
Location: Seattle

Total Topics: 2
Total Posts: 4

Posted 05/13/09 - 08:42 AM: Subject: Some more questions	#1
Prof. Chalmers, I have a few more questions if you have the time... 1) Another question regarding the "Absent Qualia..." chapter of "The Conscious Mind"... in introducing the thought experiment of replacing neurons with equivalent silicon implementations, you say that it's only necessary to get the chips' input-output dependencies right, and that, say, internally a table-lookup could be used. Given that, I imagine replacing all my neurons with their lookup-table isomorphic chips, and indeed find that my qualia are unchanged. Now I begin a process of replacing a larger and larger set of chips with a single superchip. I start by creating a chip that has the inputs and outputs of two neurons. I combine the lookup tables of the two chips being replaced, install the combined table in the superchip, and replace the two neuron-level chips with the superchip, finding my qualia still unchanged. It seems to me that I could install a larger and larger superchip, replacing one more original chip at a time, until eventually my brain comprises a single superchip with a massive lookup table; and, by an argument similar to that countering absent (etc.) qualia in the case of neuronal replacement, still with the same qualia. But this is the system proposed by Block that you suggest later in the chapter is only behaviorally equivalent, which plausibly has no experience, or at least a quite different experience than I do. I feel like I'm missing something here, and that there's something flawed about the superchip idea, but I can't see it. Do you have any comments? (Hmm, thinking about it, I do see multiple problems. Lookup tables aren't history-sensitive, and neurons - and brains - are. But maybe a sufficiently small granularity of the unit of physical replacement and of the state period can rescue it. More seriously, as neurons are absorbed into the superchip, I think internal timeshifting would be required to maintain functional isomorphism, and it would no longer be a pure lookup table. Maintaining functional isomorphism might end up looking the same organizationally, like just putting the existing chips in a big case. Nonetheless, I'd still welcome any comments you may have.) 2) What role has your own experience of various types of sleep experience played in your intuitions regarding consciousness, if any? More generally, I'd be very interested in anything you might care to share about your own introspective processes or methods. 3) In various places you've mentioned the idea that maintaining a system's counterfactual sensitivity is important to organizational (or informational) invariance, even if those counterfactuals are not exercised. I can see this idea's usefulness, but I can't get a good intuitive handle on it - it's very counterintuitive to me. If you were to write a short intuitional appeal for it, addressed to, say, the audience that you wrote "The Matrix as Metaphysics" for, what might you say? Or, is there a description/discussion you might refer me to? Thanks again and best regards... Edited by bemoosed on 05/13/09 - 12:18 PM

davidchalmers
Aspirant

Usergroup: Guest Speakers
Joined: May 09, 2009

Total Topics: 0
Total Posts: 27

Posted 05/15/09 - 05:30 AM:	#2
(1) Great case. I remember thinking about a case like this quite a lot when writing that article/chapter. Even setting aside your worries about history sensitivity (perhaps we're just modeling a system at a momnt), I think it's different from standard neuron-replacement in that instead of a series of small changes, it is a series of larger and larger changes (toward the end, almost the whole system changes every time). Furthermore, it is much easier to take the line that this sort of change undermines beliefs/judgments as well as qualia, because of the big changes to functional organization. So I suppose that I think that this sort of series of changes would gradually take one from a system that thinks/believes/judges to one that does not. Once that happens, the idea that qualia fade out along the way is not nearly so problematic as in the original case. (2) No real role for my sleep experiences, alas, at least that I know of. Though I don't know what sort of roles e.g. my dreams may have had without me knowing! In general my work hasn't been driven by deep personal introspection, despite the important role that I give to introspective data. I recall a reviewer criticizing my book for this property, reasonably enough. The sort of introspection that plays a role in my work is more that of certain broad structural features of the experience of vision, thinking, attending, and so on. It would be nice to have more of a role for more subtle and personal introspection, but it doesn't seem to be one of my strengths. I certainly encourage this in others! (3) I owe one simple way of conveying the idea to my advisor, Doug Hofstadter. There are boats at Disneyland (I'm told) that seem to steer downstream, but really are following a single track under the surface. If we had an AI system without counterfactual sensitivity, it would be like one of those boats. All sorts of apparently impressive behavior, but unfolding in a way that is preordinained, insensitive to the environment, and not really involving any sort of flexibility and responsiveness at all. With counterfactual sensitivity, on the other hand, then more than one track will be available -- if different inputs had come in, or if the system had gone into various different states, it still would have done appropriate intelligent things, if quite different intelligent things. That sort of sensitivity seems pretty crucial to our conceptions of cognition and intelligence. I think it is probably central to consciousness as well.

bemoosed
Initiate

Usergroup: Members
Joined: Mar 29, 2009
Location: Seattle

Total Topics: 2
Total Posts: 4

Posted 05/15/09 - 09:41 PM:	#3
(1) Ah, I think I see, thank you. Reading your response, I went through attempts to maintain a small change with each iteration, e.g., with schemes of switches and performing the process in reverse - but all seem equivalent, either with large changes, functional changes, or organizationally equivalent to start with. (2) Thank you, interesting. In part I was curious about sleep not only because you mentioned dreaming briefly in your book but because my own experiences of somnambulism (third-person reports only!) seemed to support that conceivably it could all happen in the dark (of course I might have had vivid qualia at the time, just no memory of them!) (3) Great metaphor, thank you. Evocative to consider the consciousness of the system as correlated with a field of probabilities. I recall your discussion of the Everett interpretation in "The Conscious Mind" though I can't quite suss the connection right off - like perhaps a lack of counterfactual sensitivity crams a system's experience into a cramped, too-small set of miniworlds. (No longer countered, my intuition's now just way overtaxed!) I think of two Marys in two rooms before she ever steps outside: one in a colorful room but completely color-blind, another with full color sensitivity but in a black-and-white room otherwise identical. Intuitively, to me, their experience is identical: perhaps it is, their apparent difference in counterfactual sensitivity immaterial considering the limited range of possible sensory events for each: there can be no superposed color states for b&w-room Mary by virtue of her environment, and their superposition doesn't split colorblind Mary. But if one Mary is deaf while the other can hear, perhaps their difference in counterfactual sensitivity leads to a difference in experience, even during the course of, say, (apparently) sitting in five minutes of silence(?). If a superposed sound/no-sound event occurs, deaf Mary still does not participate while hearing Mary's brain states are split... well, here I get lost looking for a justification or intuition that the experience of both hearing Marys are somehow different than that of deaf Mary, though at least one obviously is. But now I definitely have more of a sense that they could be. If you have the time and inclination for a couple more questions... (i) in evaluating the informational (and thus conscious) equivalence of two systems, is it likely reasonable to compare their counterfactual equivalence strictly in terms of the full range of inputs presented by their target environment(s) (as with the first two Marys above)? And, (ii) how might the Everett interpretation of quantum mechanics (or any other) support the centrality of counterfactual sensitivity to consciousness (or vice versa), or at least relate to it, if at all? Thanks again and best regards...

davidchalmers
Aspirant

Usergroup: Guest Speakers
Joined: May 09, 2009

Total Topics: 0
Total Posts: 27

Posted 05/17/09 - 06:20 PM:	#4
(i) Yes, full range of possible inputs. I certainly don't say that every difference in counterfactual sensitivity will make a difference to consciousness, and an answer to the question of which differences matter would probably require a theory of consciousness. But I do think that some differences in counterfactual sensitivity matter, which is what's relevant for the original point. (ii) I'm not really sure how this issue interacts with Everett. Of course, on Everett certain sorts of counterfactual sensitivity (what I would have done if the electron had come through the other slit?) are really disguised forms of actual-world sensitivity (one branch of me does one thing, another does another). So one might think one dosn't need true counterfactual sensitivity after all. But even on Everett there are plenty of counterfactuals that don't correspond to actual-world branches. And I suspect that many counterfactuals relevant to consciousness will be among them. So I'm not sure that Everett fundamentally alters the situation.

Sorry, you don't have permission to post. Log in, or register if you haven't yet.