Action selection in the basal ganglia - a computational investigation

The classical explanation for how basal ganglia could perform action selection relies on the direct and indirect pathway. In this computational study we instead build on earlier studies viewing the part performing selection as consisting of the direct and hyperdirect pathways, with the indirect pathway playing the role of controlling the balance between these two [1]. The present model is a further development of a previous spiking neuron model [2], reimplemented in the software NEST [3] using conductance based integrate and fire neurons. We explore the functional consequences of including additional connections known from experimental studies, such as the collaterals from the direct pathway striatal medium spiny neurons to globus pallidus externa (GPe) as well as a dual indirect pathway through GPe. Simulation experiments suggest that this new model can perform selection with an improved sensitivity and over a larger range of dopamine modulation.