ib@apolling (Ivan N. Bach) (09/02/89)
Several people have pointed out that some neural connections in a human brain seem random and unstructured. I think that we should distinguish between the form and function. If a particular set of neural connections performs a certain function correctly, it does not matter whether axons and dendrites in that subnetwork are nicely arranged into regular patterns of parallel connections. As any hardware hacker can tell you, you can design extremely messy prototypes of new electronic circuits which perform specified functions. Once you debug a new circuit, you usually prepare a neat printed-circuit board layout with vertical and horizontal connections which meet at right angles, but these changes do not affect the basic function of the new circuit. Electronic engineers usually prepare a schematic diagram of a new circuit which does not have any physical dimensions. It just shows how different circuit components must be connected to perform the specified function. I think that physiologists should try to deduce the functions of subnetworks in the human brain, and maybe produce schematic diagrams of those subnetworks. In other words, they should try to separate the essential elements of the subnetwork and eliminate non-essential elements, such as, for example, the color of individual tissues. Once we have such diagrams, we can try to design artificial subnetworks that will perform the same functions faster, better, cheaper, etc.