Skip to main content
. Author manuscript; available in PMC: 2007 Oct 1.
Published in final edited form as: Neural Netw. 2006 Sep 20;19(8):1013–1026. doi: 10.1016/j.neunet.2006.05.038

Figure 9.

Figure 9

A five-unit neural network that implements the decision mechanism. Mutually inhibiting accumulator units in the first layer approximate a DDM, units with high gain and bias in the second layer detect theshold crossings, and a reward rate estimator with balanced feedback and gain modulates the detector thresholds according to z = zmaxwr.