Figure 1:

Model description. (A) Decision-making network. Each circle represents a population of neurons. As the targets appear, the input population is activated in the same way on each trial. The input is fed through plastic synapses into two excitatory populations representing the two possible choices. These two populations compete through an inhibitory population and work as a winner-take-all network. The plastic synapses are modified depending on the activity of the pre- and, postsynaptic neurons and on the outcome of the choice (reward or no reward). (B) Learning rule in rewarded trials in which the population representing the left target is activated. The synapses to the chosen target are potentiated with a learning rate α_r, and those to the other target are depressed with a learning rate γα₇.. (C) Same as in panel B but in unrewarded trials. The synapses to the chosen target are depressed α_n, and those to the other target are potentiated γα_η.