| Action selection | in the motor cortex or basal ganglia , but do not simulate the details of action selection, Which have been addressed previously [26—30]. |
| Biological plausibility, biological detail and future work | Interestingly, [16] proposed that the basal ganglia could compute temporal difference errors by comparing activity in the indirect pathway, which might store the predicted value of the previous time-step, and the direct pathway, which could code the predicted value of the next state. |
| Biological plausibility, biological detail and future work | In addition, we also did not model the action-selection process itself, which has been suggested to take place in the basal ganglia (see [30]). |
| Comparison to previous modeling approaches | To keep the model simple, we did not specify the mechanisms causing persistent activity, which could derive from intracellular processes, local circuit reverberations or recurrent activity in larger networks spanning cortex, thalamus and basal ganglia [20—22]. |
| Comparison to previous modeling approaches | In PBMW, memory units are bistable and the model is equipped with a system to gate information in prefrontal corteX via the basal ganglia . |
| Discussion | The scheme uses units inspired by transient and sustained neurons in sensory cortices [19] , action-value coding neurons in frontal corteX, basal ganglia and midbrain [12,35,36] and neurons with mnemonic activity that integrate input in association corteX. |
| Learning | Neurons representing action values have been found in the frontal cortex, basal ganglia and midbrain [12,35,36] and some orbitofrontal neurons specifically code the chosen value, qa [37]. |