ARCHITECTURE OF INTELLIGENCE - THE PROBLEMS AND CURRENT APPROACHES TO SOLUTIONS

We propose as a working hypothesis a Separability Hypothesis which posits that one can factor off an architecture for cognition from a more general architecture for mind, thus avoiding a number of philosophical objections that have been raised about the 'strong AI' hypothesis. Using a coin-sorting machine as an example, we discuss a range of positions on representations and argue that, for many purposes, the same body of matter can be interpreted as bearing different representational formalisms. We then propose that one way to understand the diversity of architectural theories is to make a distinction between deliberative and subdeliberative architectures. The search for one architectural level which will explain all the interesting phenomena of cognition is likely to be futile. There are a number of levels that interact, and this interaction makes explanation in terms of one level quite incomplete.