.comment-link {margin-left:.6em;}

Prior Knowledge

Tuesday, April 12, 2005

Bats, Consciousness and Simulation

I've been meaning to write about the interesting talk Doug Campbell gave on our Cass trip. I can only half-remember it now, but I'd better get something down before the other half disappears too...

Anyway, Doug was talking about Nagel's famous Bat argument. He reconstructs it as follows:
A1: Any sufficiently intelligent creature can in principle grasp any physical fact by adopting an objective perspective (e.g. through science).

A2: There are facts about what it is like to be a bat that can be grasped only by a creature that is capable of adopting a bat-like subjective perspective.

A3: It is not the case that any sufficiently intelligent creature can adopt a bat-like subjective perspective; so

A4: The physical facts do not exhaust the facts. (From A1, A2, and A3.)

His response was to highlight two distinct methods of learning facts: theory and simulation. If you want to know what each button on a VCR does, you might take it apart, carefully examine the wiring, and develop a technical theory of how it works. Alternatively, if you have another VCR of the same make and model, you might simply try pressing the buttons on this other VCR and see what happens. An assumption of similarity would allow you to carry over the results to your original VCR. In effect, you learnt about it through simulating its processes on another (similar enough) machine.

Next, suppose we want to know how other people are likely to behave. It would be very difficult to devise a fully-fledged theory of human psychology. Fortunately, we don't need a complex theory because - as with the VCR example - we have a similar 'model' at hand that we can experiment with: ourselves! Doug pointed out that consciousness allows us to learn about the behaviour of other people through simulation. We imagine how we would react in a particular situation, and - on the assumption that we are relevantly similar to the target - we conclude that they would likely react in a similar way. Moreover, Doug suggested that simulation is how we learn "what it is like" from some other perspective. I'm a bit unsure on this point, but I think he may even go so far as to suggest that the process of simulation just is "what it is like" -- i.e. facts about qualia may be identified with facts about simulation?

Now we get to the bit I had difficulty following: relating this point to the bat argument. Bats are far too different from us, so we lack the capacity to accurately simulate their behaviour. But what follows from this? If I recall correctly, I think Doug was wanting to use his point about simulation to deny premise A2 (or was it A3?), but I'm not sure how that works.

There's one crucial question in particular that I'm not sure about. We have these facts about "what it is like", that we can learn about through simulation. So far so good. My question is: are those facts only learnable through simulation? Or is it also possible, at least in principle, to devise a theory of consciousness that would yield facts about what it is like to be a bat?

If such a theory is possible, then A2 is false. But it doesn't seem that such a theory is possible -- that's the whole problem with consciousness; it seems inherently subjective. Alternatively, if qualia facts are only learnable through simulation, then it seems we should be denying A1. That is, assuming we hold such facts to still be physical facts. I'm not sure how plausible that is either though!

I've probably misunderstood the argument somewhere along the line, so if anyone can clarify things for me, please do so!