Conclusion

CAs are reverberating circuits of Neurons.

CAs are a good level to model human cognition, and few people have done modelling at this level.

It's hard to get experimental neural evidence of CA behavior.

I display an early working model of CAs named CANT.

CANT has: unidirectional connections between neurons that change via Hebbian and Anti-Hebbian learning; neurons that have activation, a firing threshold, activation decay, and fatigues; then neurons are connected in a distance-biased way.

CANT nets have CAs in the them via certain metrics.

It surpasses Hetherington's model on measurements of Reliability, Persistence, and Uniqueness.

It is an early model.

Questions still exist on:

• Learning Rules
• Topology
• Basic Model Parameters
• Metrics

As the basic CA model is solidified, questions of how CAs interact will need to be addressed. How can CAs implement hierarchies, rules, and variable binding? How can neurons participate in multiple CAs? How can a system learn with one instantiation?

These are the future directions of the model.