Modelling afferent connectivity, postsynaptic plasticity, and signal discrimination

We model a small system of hypothetical neurons having ongoing afferent input ("stimulus") short-term refraction/inhibition, and brief-latency intercellular excitatory feedback loops. We incorporate postsynaptic plasticity: synapses onto a given cell are grouped into dendritic "neighborhoods" wherein sufficiently large local postsynaptic potentials sums cause proportional changes in local synaptic efficacy. The efficacies record multiple cell-to-cell and time-to-time correlations. The efficacy patterns fall into several classes, for each of which there is a stereotypical relationship leading from "steady-state" stimuli to corresponding sets of possible "responses" (periodic firing patterns). We examine the system's ability to discriminate between stimuli via firing patterns and to retain information during change of efficacies. There is an optimal region for efficacy configurations inside of which discrimination is good and outside of which the system can be inactive, particularly forgetful, or perseverative. We find that stimulus discrimination is enhanced by divergence and convergence in the pattern of afferent input. Perseverative firing patterns, invoked or suppressed by cotransmitters, can play a role in associative activities of the network.