Four Verses from the Dàodéjīng by Jochen Szangolies

Dear James,

thanks for reading and rating! It seems a lot of people have saved their votes for the final day---things have moved around quite a bit today. I'll try and have a look at your essay before voting is through!

Cheers,

Jochen

Dear Noson,

I'm very happy you found the time to have a look at my essay, and even more happy you found something to like about it!

As for strong AI, it depends how you understand the thesis. I do believe that conscious machines are possible---and that, indeed, we are just such machines. The notion of dualism is superficially attractive, but so far, I simply have never found a good explanation of how two substances can causally interact, without effectively becoming unified: after all, we know the physical world only via cause and effect; whatever leaves some imprint on our measuring devices, via usually a fairly long chain of proxies unraveled by inference, we call 'physical'. But anything that can causally influence something physical is in principle detectable by a suitable measuring instrument; so what could make it non-physical?

If, however, you take strong AI to mean that conscious machines are conscious due to computation, then I would be inclined to disagree: while I believe that model-building, and consequently, the mind's explanatory capacities are indeed computational, I also think that this computation has to be grounded in something non-computational to avoid infinite regress. So consciousness is not merely the right program running on some appropriate hardware.

You're very right to point to Rorty, I think. But in a sense, my account is not wholly model-based: the connections between the model and the world, in a way the 'clay' from which the models are built, is not itself part of the model, but something more fundamental that underlies it. I think that we do build models in cognition is hard to deny: for instance, when I picture how to perform a task, say, tying my shoes, I can visualize it as a series of steps---a kind of algorithm.

But we can model the world under different aspects, yet still cut from the same cloth: for instance, when listening to somebody speak, we can pay attention to what is being said---understand the meaning of the words---or we can pay attention to how it is being said---register inflection, tone, rhythm, the uhms and uhs that one usually does not consciously perceive.

Both these views pertain to the same phenomenal experience, however: the same sounds reach our ears. We just attend to that experience in different ways---which is roughly what Ned Block calls 'access consciousness' (as opposed to 'phenomenal consciousness'). This is what my models really pertain to (although I formulated this somewhat stronger in the essay).

As for an analogy to the completeness theorem, in a sense, I already use it in the essay: in my discussion of the zombie argument, I claim that, since phenomenal experience is in some sense 'undecidable', one can imagine both that a physical system (like a brain) possesses it, or fails to (in which case it would be a zombie).

The completeness theorem, applied to a formal system subject to Gödelian incompleteness F, basically tells us that there exists a model of the system extended by its Gödel sentence G, as well as a model of the system extended by its negation, ~G. Applied to the zombie issue, this would entail a 'possible (or perhaps, imaginable) world' in which there are zombies, and a possible world in which brains possess phenomenal experience.

Of course, this talk is somewhat metaphorical at best. But it's very interesting to think about!

Thanks, again, for your comment, and your good wishes.

Cheers,

Jochen

Hello Don,

thanks for your kind words! I think it was in the discussion thread of Dean Rickles' essay that I stumbled upon a similar connection to cosmology versus particle physics as you point out, however, with ancient Greeks instead---the atomists favoring the bottom-up, particle physics style of explanation, and the Eleates considering everything to descend down from Oneness, Being, or what have you.

Neither may be any more right than the other: like with the initial values of differential equations, which you can specify at the beginning, the end, or every Cauchy surface in between, maybe one can in fact smoothly interpolate between 'levels' of fundamentality. Maybe that would have been a nice idea for this contest, too!

I'll try to sneak a brief look at your essay before voting closes.

Cheers,

Jochen

Dear Sylvia,

sorry for taking so long to reply to your comment---it's only now that I've got some free time on my hands again.

Anyway, I think you've given my essay a fair reading, and your criticism is reasonable---I've already conceded above that I put things too strongly in claiming that we interact with the world only via models.

I should have included a more careful discussion on what it is that models do for us---and here, I mostly have the notion of Ned Block's 'access consciousness' in mind. Think about the famous Necker cube: on paper, it's really a collection of seven peculiar shapes adjoining each other. This is what we see, what is phenomenally present.

But since this is a somewhat unlikely, complicated collection, our brains soon find a much simpler description: namely, as the projection of a three-dimensional cube. This puts the seven unrelated shapes into a context from which they emerge naturally.

However, the 2D mesh underdetermines the cube with respect to its orientation. Thus, there is no information to differentiate which side should be the 'front'; hence, both orientations are equally consistent, and we experience the somewhat disorienting sensation of the cube 'flipping'.

I think this 'flipping' is due to two competing models; and thus, it's here that my notion of model really ought to be introduced: as organizing phenomenal impressions 'as' something. Thus, we have phenomenal experience, which allows us to model the world as being a certain way---say, containing a cube facing a particular way. Modeling determines the structure of the world, in this sense.

Maybe my propositions make more sense to you under this interpretation. To me, it seems hard to escape---in the end, computation is just the transfer of structure, so that 'structural' and 'computational' end up picking out the same things, which then leads to the identity of computation and modeling.

Perhaps this also helps making the connection to the hard problem more clear. I don't think that experience itself is due to modeling, but rather, that modeling 'organizes' experience---the raw experience is the same in both views of the Necker cube, but we possess different models of it. But since experience underlies modeling in this sense, experience itself can't be modeled, and hence, we have the hard problem (under the---to me, plausible---assumption that we need a model to answer 'how come'-type questions).

As for the information content of sets, this is perhaps clearer if phrased in the context of computations: a computer program that produces no output has vanishing length, up to an additive constant that is generally taken to be of no importance in Kolmogorov complexity. The same goes for the program that produces every output.

As for Wittgenstein, while I didn't consciously pattern the first proposition after him, he certainly was a major influence on the thoughts that are collected in my essay (I've discussed this above for a bit, I think).

Finally, I'm glad you focused on Whitehead's insight---it's something that I think is all too often overlooked these days.

Thanks again for your reading and constructive criticism.

Cheers,

Jochen