Skip to main content
. 2012 Jan 31;6:2. doi: 10.3389/fnins.2012.00002

Figure 6.

Figure 6

Our spiking model of reward-based learning and action selection in the cortex and basal ganglia. Dotted lines show dopaminergic connections from the substantia nigra pars compacta (SNc) which modulate the local synaptic strengths using the equations shown. Raw reward signal is input into the ventral striatum, which also computes an estimate of Q(s, a) using connections from the cortex. The difference between these values is fed to the SNc, which produces the error signal.