January 20th 2004 04:55:34
RE: Self-Knowledge( Posts: 2; New Posts: 2 ) [ See: ComDig 2004.03 ]
An intersting issue. An analogy is with computers. Advanced computers have so called control memory, a vehicle to control computer itself. It is, though, a risky business: e.g., a change of the controls on-the-fly may easily lead to unpredictable catastrophic results. So, in such computers the direct control and control of controls is separated in time. At the same time, the modification of controls based on the accumulated experience is beneficial to the overall system, so multi-level control-of-control-of-control... systems seems to be reasonable trend for increasingly complex systems. Any comments? cheers, val
February 4th 2004 07:35:28
NEW RE: Self-Knowledge
In computer architecture, you could do this with just two levels. If the second level is "interpretive" or "open code" it can control the next level (compiled level) and also serve to control and modify itself. Reflexive architectures like this are used in "multi agent" technology - agents that interact, communicate as well as learn. Interesting question is what is the self-knowledge of these agents? They also learn both from reflection (analysis of experiential records) as well as interactions with the environment consisting of other agents as well as the physical environment.