Saturday, May 28, 2011

What Does It Mean for a Brain to Represent Itself?

What might it mean for a brain to construct an account of itself? That’s not at all obvious. Let’s start with a simple, if somewhat odd case, and then get lost.

Damasio’s Brain Sketchs Itself

Antonio Damasio is interested in how brains work. Since his own brain falls within that class of objects, the representations he constructs of the typical brain would thereby apply to his brain. But we should not imagine Damasio in a surgical suite under local anesthesia, brain exposed, looking at an image of that brain in a mirror and making a sketch of it in the way some artist might construct a painting of Isaac Newton.

Were such a thing to happen—and for all I know it may have—it would certainly be the case that:
  1. the image of Damasio’s brain in the mirror would be physically distinct from the brain itself, that
  2. Damasio’s sketch would be physically distinct from both the mirror image and the brain itself, and that
  3. the neural representation (in Damasio’s brain) of the image in the mirror and of the sketch would be physically distinct from the mirror image itself, the sketch itself, and the brain itself and this despite the fact that it exists within that very brain.
In any event, the neural representation of brain function which Damasio created over the years would itself be physically distinct from any of the representations in 1, 2 or 3 above.

But is it Powerful Enough to Reproduce Itself?

These days, at least since John von Neuman’s work in the late 1940s, computationalists have been interested in the idea of self-reproducing machines, that is machines powerful enough, in some sense of ‘power’, to reproduce themselves. Power in this case is not a matter of energy, it’s a matter of computational power, and just what that is, well, I’ve got to leave that to the experts. The general idea is that some computing devices are more powerful than others such that a more powerful device can perform any computational that a less powerful device can perform, but not vice versa.

What is of particular interest is whether or not a given device can perform a set of computations the output of which is a fully functional copy of the device itself. Such a device would be, by that fact, self-reproducing. And, it would seem, self-representing as well, with the copy being such representation.

So, let us return to Antonio Damasio in the surgical suite making a sketch of his brain. There’s no doubt that the sketch is a representation of his brain. But it is not at all obvious that Damasio’s ability to produce such a sketch implies that his brain is a self-reproducing device in the sense that the computationalists envision. Maybe it is, maybe it isn’t. But it’s one thing for a computational device (the brain) to be able to make an image of its physical envelope, and another thing entirely to be able to create another device that is as powerful as itself.

Can Damasio’s brain do that? Well, I don’t know whether or not he has children, but if he does, then his brain already has, in some sense, reproduced itself. But that seems rather beside the point. For any animal can do that. Just as any animal can sense its own body parts and thereby represent itself.

But what other sense could there be to for a brain to be able to reproduce itself? Perhaps what we have in mind is being able to create, say, an artificial copy. Can Damasio’s brain do that, that is, create a computer that is as powerful as it is? No, nor can anyone else, so far.

And yet, we CAN create digital computers that have the power of universal Turing machines, which is as powerful as computational devices get. (We think.) We just don’t know how to harness that power to the task of mimicking a human brain.

Where Are We?

Unlike Plato, we know that the mind is in the brain. We know that the brain is, in some sense, capable of posing questions about itself – though just what sense that is, is somewhat obscure. We can talk of computational power, and imagine all the computational power that can possibly be. And we can distinguish between representing the physical envelope of a device of whatever computational power and being able to reproduce the computational capabilities of that device.

But what these various things have to do with one another, is it any more than the verbal particle ‘SELF’?

No comments:

Post a Comment