Two Choice Model
- We set up a system with four subnets.
- The input was the Ss, orthogonal CAs. In the two choice model
there were just 2, but we did it with more inputs and outputs.
- The output was the As, again orthogonal CAs.
- Initially the weights were random, so the system would guess.
- The guess used the explore net, which was randomly connected to the
As.
- The system then got environmental feedback from the value net.
- If it was good, this would supress the explore net and the
correct response would be reinforced.
- If it was not good, it would not supress the explore net, and a new
response would be generated when the first CA fatigued.
