You must login before you can run this tool.
This tool simulates the responses of midbrain dopamine neurons using temporal-difference learning as shown in the model below.
FIGURE 11.12 Temporal-difference learning implemented in a neural network model of the midbrain dopamine system. The input units xj project to the difference unit y over connections with weights values) vj (j = 1 , . . . , 20). The response of the difference unit y(t) is the difference between its weighted input sums at times t and t - 1. The difference unit y and the reward unit r project to the prediction error unit z over connections that both have weight 1. The response of the prediction error unit z(f) is the sum of its inputs from y and r at time t. The prediction error unit z represents a dopamine neuron.This tool is built from the MATLAB scripts for Tutorial on Neural Systems Modeling by Thomas J. Anastasio.
Cite this work
Researchers should cite this work as follows: