July 04, 2006: Does the Brain have State?

One consequence of trying to draw upon our understanding of digital computers to enlighten our understanding of the brain is that the former is grounded in the concept of state. The consequence is that we assume that we can apply this concept to the brain, but is this a valid or even desirable assumption? The usual argument is that, however, large the number of neurons involved may be, the concept of state, itself, can scale to any magnitude. So, while an accurate state description may be unwieldy, it is still (at least in theory) possible.

However, the real question we should be asking is whether or not it make sense to describe the brain in terms of a static object. Where many processes are involved, we can still make analytic progress by "freezing time" and examining the "state" of those processes at such a "frozen instant." However, that abstraction is only useful if we can think in terms of how the state we are examining is the result of a transition and what similar transitions are possible to proceed from that state. This is where we may run into trouble by applying such thoughts to the brain. The intimidating truth is that every single neuron is a dynamic processes whose nature we are still trying to understand. We talk about these cells firing as if that activity can be abstracted into a binary state, but that abstraction may be too coarse to explain brain behavior when matters such a memory are involved.

I have invested a fair amount of my own mental effort in trying to make sense out of a hypothesis that Gerald Edelman made about memory. His hypothesis rejects the idea that biological memory is based on a store-and-retrieve model. Instead, he argues that what we remember arises from ongoing processes of refreshment that take place in the brain. Edelman is thus talking about a dynamic process that, by virtue of its complexity, cannot be readily abstracted down to transitions across static states. I have no idea with Edelman's hypothesis will ever be settled in my lifetime; but I do know that, however we may choose to think about the brain using the digital computer as a model, in just about any practical setting we have not done very well by thinking about human memory in terms of a database! This is likely to be yet another situation in which it is more important for us to develop better skills for "thinking in time."