What we'd like to do with CAs
- CAs are proposed as an intermediate level of cognition
- They are the basis of symbols
- Hebb proposed them as a mechanism to separate figure from
ground
- They can be combined in higher level structures, like hierarchies,
lists, and cognitive maps
- They can be bound short term (variable binding) to form among
other things rules
- They can be used as a model for psychological phenomena
- With this power they can form the basis of a cognitive architecture,
and perhaps real intelligence
- These are things we'd like to do, but haven't actually got a
computational model working yet.
- These things are an obvious advancement on the current state of
the art in AI
- One of the problems with existing symbolic systems is that they
are brittle
- Similarly connectionist psychological models are generally
different for each phenomena
- This would give a unified architecture that would have hope
of solving the symbol grounding problem
