Thoughts on Conciousness Explained, by Daniel Dennett

[wildeabandon]

This is technically a reread, but the first time was around eight years ago, and I could remember very little of it. I found the first half, where he sketches out his model of consciousness - very condensed, a disparate and competing set of “content-fixation events” in the brain, some of which get retained as speech acts or memories, some of which get discarded and forgotten, but lacking any central “meaner” or “observer” co-ordinating these individual elements, and the bit of the algorithm that feels from the inside like being conscious is content-fixation events that are about other content fixation events - quite hard and slow going, and as I was reading them, wondered whether my memory of having found the book illuminating and clear was inaccurate.

Then I got onto the second half, which started looking at specific examples, and suddenly everything became much clearer, and I polished it off in about a tenth the time the first half took me. There are still bits that I don’t quite understand, or at least can’t articulate, in particular what ‘aboutness’ means as a property of a content-fixation event, but in general the theory felt quite comfortable and intuitive by the end of the book.

One thought that came out of the book that I want to follow up on is that there’s a fallacy in thought experiments, which is common to philosophical zombies, Mary the colour scientist, and the ontological argument, which goes “I can imagine this phenomenon, and using the properties of the thing I have imagined, such and such a proposition must be true (or is impossible)”. The fallacy being that you can’t actually imagine it accurately. I’m curious how much this crop up elsewhere. It almost feels as though it undermines the very concept of thought-experiments - or at least relegates them to ways of generating ideas, but not of providing any further insights.

Comments

[simont]

there’s a fallacy in thought experiments [...] which goes “I can imagine this phenomenon, and using the properties of the thing I have imagined, such and such a proposition must be true (or is impossible)”

This is making me think of mathematics. I'm put particularly in mind of an old web page by Tim Gowers, which tries to explain why it's necessary to do all those counter-intuitive 'definitions', generally by fiddly arrangements of sets, of things you thought you already knew what they were, such as defining an ordered pair as {a,{a.b}}, or defining a function from S → T as a particular kind of subset of S × T. Essentially, the point is that first you imagine a thing which has some properties you actually care about (e.g. that ordered pairs are a thing where (a,b) and (A,B) are equal if and only if a=A and b=B), and then you start using it all over the place and reasoning from those axiomatic properties, but somewhere in the middle you should ensure that at least one thing really exists with the properties you're assuming. So the point is not that everyone now needs to keep in mind forever that (a,b) really means {a,{a,b}}. It's that you check, once, that you could define it that way, and if you did then the property you really care about would hold; and now you can proceed secure in the knowledge that there is some definition that will work (but you never actually need to think about that definition again).

In a less mathematical context the ontological argument is exactly this kind of thing. My favourite way to 'correct' the argument has always been to transform it into the form

1. By definition God, if he existed, would be perfect.
2. One of the properties of perfection is existence.
3. Therefore God, if he existed, would exist.

which has always seemed to me to address a more fundamental weirdness in the argument than the assorted responses that get tied into knots over whether existence is a predicate or similarly abstruse questions.

And philosophical zombies, likewise – certainly you can kind of imagine that a thing might exist with some counterintuitive properties, and certainly you can deduce from those properties that all sorts of other counterintuitive things would have to hold if so, but none of this has any reason to affect your judgments about the real world unless you put back in the middle step of the argument in which Tim Gowers builds an example philosophical zombie out of nothing but ZFC set theory :-)

(Another mathsy analogy that this calls to my mind is the apocryphal story that there was a thriving subfield of mathematics studying a certain kind of object, until two papers came out at around the same time, one proving that all such objects had a particular property, and the other proving that none of them had that property – and on checking, both theorems were perfectly true, which put a sudden end to the study of those objects!)