The Copycat Architecture

Problems with Copycat

• No explicit "awareness" of what it is doing

• Cannot remember more than one answer at a time

• Cannot compare different answers

• Cannot evaluate answers suggested to it

• many more . . .

What is "self-awareness"?

"What seems to make brains conscious is the special way they are organized---in particular, the higher-level structures and mechanisms that come into being.  I see two dimensions as being critical:  (1) the fact that brains possess concepts, allowing complex representational structures to be built that automatically come with associative links to all sorts of prior experiences, and (2) the fact that brains can self-monitor, allowing a complex internal self-model to arise, allowing the system an enormous degree of self-control and open-endedness."

        ---Douglas Hofstadter
 

To deserve to be called "self-aware", an agent must (at least) be able to:

• perceive its own behavior

• react to its own behavior

• Gives a high-level picture of the most important events that happen during a run

• 12 events versus 1,558 codelets

• Allows Metacat to explicitly represent its own behavior

• Codelets can look for patterns of events in the Temporal Trace

• Strongly influences the selection probabilities of codelets

• Selectively favors or suppresses the creation of certain kinds of structures

• Concepts in the Slipnet can be clamped in a similar fashion

• Allows the program to focus explicitly on certain ideas or structures

• Gives the program a high degree of self-control

• Answers in memory are described by collections of Slipnet concepts (called themes)

• The quality of an answer can be judged on the basis of its themes

• Answers can be compared and contrasted on the basis of their associated themes

• Metacat can be reminded of earlier answers according to the similiarity of their themes

Metacat home page