2006-01-19

Thinking about thinking

It's said that human thought is not "rational" or "deterministic" like a computer program. Computers receive data, run a program, and produce output. Human thought doesn't work this way, it is said, and so can't be emulated by a computer.

But human thought IS deterministic. It's an electrochemical process running on a biological machine; every step in the thought process is determined by the laws of nature. It's just that WE can't determine the course of our thoughts because the data and the programming are not independent.

Receiving the data changes our programming.

Every thought changes our programming.

Every tentative conclusion we reach changes our programming.

No system other than a human brain can follow the process of a single human thought, but the process of thinking is, nevertheless, entirely susceptible to machine emulation.

How to implement this process of emulation, of course, remains an open question, and the answer depends on details of the process of human thought that are not yet known.

What we call conscious thought is not the same as executing a computer program-- but I believe it IS comparable to the edit-compile-execute-debug cycle, so that might be the right model to use. So, how would one automate that process without a human in the loop? How DOES the brain automate this process? Perhaps by successive approximation, where the program is edited by the act of executing it, and the results are tested against standards of merit, and the process is repeated until it produces answers that are good enough.

0 Comments:

Post a Comment

<< Home