Learning

Learning is essential to the work on CAs.
The only known learning mechanism in the brain is Hebbian learning (though cell and axonal death and growth also could be considered).
Hebbian learning states that two neurons (that are connected) will tend to increase their synaptic strength if both are frequently co-active.
This leaves a lot of room for possible learning rules.
Moreover it has to include inhibitory neurons, and could contain a range of different rules for different neurons or pairs of neurons.
Increase the weight if both fire, decrease when one fires and not the other. (If you don't decrease, the weights increase indefinitely.)
The simplest rule we use is a correlatory one.
The weight becomes the value of the likelihood that the post-synaptic neuron fires when the pre-synaptic neuron fires.
This is a pre-not-post rule, and you could use the post-not-pre rule. We've mostly used pre-not-post rules.
The inhibitory correlatory values are -(1-p) where p is the likelihood of the post-synaptic neuron firing when the pre-synaptic neuron fires.