A Biologically Inspired Computational Model of Basal Ganglia in Action Selection

Figure 5

Temporal activity of the neurons in each structure of the whole network during error feedback, red dashed line punishment, blue solid line reward. The effect of reward and punishment is clearly noticeable in the activity of striatal neurons. In case of punishment (red signals), a transient dip occurs in the activity of the winning Go neuron. Conversely, the NoGo units clearly show a peak, particularly pronounced in the unit of the winning neuron. Furthermore, we can notice a transient peak in the cholinergic interneuron activity, underlying the inhibitory role of dopamine on ChI. In case of reward (blue signals), a transient peak occurs in the activity of the winning Go unit. The others Go units maintain low activities, due to the contrast enhancement effect of dopamine. All the NoGo neurons exhibit a transient dip in activity, particularly remarkable for the unit in the selected action channel. Finally, the activity of the cholinergic interneuron exhibits a transient dip, which contributes to the excitation of the Go pathway and to the inhibition of the NoGo pathway.