If we view the major mind function as emergent properties of the brain organization, what do we need to understand about neural complexity to explain these mental functions?
I think the major area we’re missing is at the level of neural networks. Assuming that neurons form themselves into micronets, and micronets interact with larger units, or macronets, it’s really the properties at that large level that we don’t yet understand.
With the advent of artificial neural nets we can design on computers, we are beginning to get at least a conceptual framework for making the bridge between the individual neurons on the one hand and the systems on the other. We have a long way to go.
Take the case of categorization, for example. Some regions of the brain seem to be specialized for categories of natural things and others for manmade things. And within the regions specialized for manmade things, there is further categorization - for example, tools that you use with your hands, and other kinds of things like automobiles. We don’t know how brains do this regionalization, so we don’t know how to make the nets that can help us understand it.
Another key question about brains is how they get things done in time. How can an eagle intercept its prey? How can you catch an outfield fly? How you get the timing right is a major issue for a lot of neural network theory now. What we desperately need is more understanding of neuroanatomy at the network level.
Modeling neural nets on a computer is one thing, but is the brain itself a type of computer?
It’s useful to think of the brain as a kind of computer because that allows you a framework for thinking about how individual neurons interact to achieve a certain effect. For example, if what you are trying to do is focus on a given object and your head is turning, thinking of the brain as a computer gives you a way of understanding the interaction between the neurons in the vestibular organ, which function in the sensation of movement, and the neurons that control eye movement.
But unlike a desktop computer, the very elements in the brain that process perception are also elements that store information about what you perceive. It isn’t that the memory is in one place and the processing for perception is in another place. The brain is a very different kind of computer, but I find the computer to be a useful metaphor.
Do you think the knowledge of the brain can enable us to build better computers or different kinds of computers?
Yes, I really do. I think that there is likely to be big technological payoff as we understand in more detail how neuronets function and how they solve problems. That is, how do they manage to be so flexible? How, with so few neurons, can a bee solve problems that are really quite complicated?
In general, I think the flexibility and the capacity to generalize that we see with brains and don’t see in the very brittle architecture of artificial computers are things we will understand as we get more of the story of the brain.
Do you think it’s possible to build machines that think?
I think in principle it is. Since we are machines that think, and evolution built us, then yes. Whether using the kinds of components that we are now using we can mimic the brain depends on the nature of the problem. If what we want to do is get the timing right, we may have to mimic what neurons actually do down to the level of membrane proteins.
Interview.3: Consciousness
plutonic) (charonic