Experiment 2 Description
Spontaneous Activation for Recruitment
- This experiment also uses a 20 by 20 network of
distance-biased connected leaky integrator neurons
- The topology is rectangular
- The network is trained by seeing 2 types of patterns
- The first pattern comes from the 3 to 7 column, and the
second from the 13 to 17 column
- The pattern has each of the 100 possible neurons active with 20 precent
probability
- Each pattern is presented for 10 cycles, then network is allowed to
run for 40 cycles
- There are several conditions that differ by the
amount of spontaneous activation
- It is expected that no neurons outside of the
pattern will be in the CA with no spontaneous activation
