vspatch

README

VSPatch

This is a test for representing graded numerical values in terms of the overall population activity, which is essential for the striatum neurons used in the PCore and Rubicon models. The specific test case here is the VSPatch neurons that predict and discount graded rewards.

The model has opponent D1 vs D2 VSPatch layers that are trained to predict the amount of reward for N different "conditions", with T theta trial steps leading up to the reward. It needs to not respond for the first T-1 time steps, and then accurately predict the graded reward value at T. The opponent D1 vs. D2 dynamic within the VSPatch layer is critical for enabling this dichotomous behavior to be learned.

The Train/Stats plot shows all the stats for each condition after each epoch of learning, where NR = the non-reward trial activity, and otherwise it shows the final reward trial values. These conditions are also plotted in the TrainEpoch plot, after you hit Init, after an initial epoch or more of learning.

Documentation

Overview

pcore_vs simulates the inhibitory dynamics in the STN and GPe leading to integration of Go vs. NoGo signal in the basal ganglia, for the Ventral Striatum (VS) global Go vs. No case.