Inhibitory spike-timing-dependent plasticity in striatal microcircuitry

GABAergic medium-sized spiny neurons within the striatum are in charge of the detection and integration of goal-oriented information during higher-level functions, such as learning and memory. The activity of these neurons enables novel and complex behaviours. The detection and selection of the information derived by the striatal activity rely upon synaptic efficacy changes. Recent works on neuroplasticity have focused on spike-timing-dependent plasticity (STDP) within the GABAergic network of the striatum, with an STDP curve resembling a Mexican hat rather than the variations of classical and asymmetric STDP
windows. Following this idea, we propose a computational model which integrates realistically connected networks of D1- and D2-dopamine-receptor type neurons and rewards triggered release of dopamine modulated by STDP. We also extend a form of synaptic plasticity in order to develop a model that can learn from distal or delayed rewards. Our results reveal that modified Hebbian learning may fail to capture all the dynamics needed for functional activity. However, STDP provides a robust mechanism for the network stability. In
the simulations where we consider STDP, the connection of any pair of neurons may disappear or reach a maximum synaptic strength. By contrast, in longer time scales, global connectivity stays stable.