Figure 1.

Single population scenario: network architecture and activity, connectivity and STP parameters adaptation in the output population with (U, τ_rec) learning scheme (Part 1). (A) Architecture. The learning network (green) is divided into an input region (blue) and an output region (red). Connections (black arrows) are all-to-all and obey both Spike-Timing Dependent Plasticity and rate-dependent Short-Term Plasticity. Input neurons receive an external wave-like stimulus (blue dashed arrows). (B) Mean firing rate of the output population. Shaded area represents standard deviation, horizontal dotted gray lines show the two target firing rates (high = 30 Hz, low = 5 Hz) and vertical black arrows mark the onset of the four dynamic phases alternating the target according to the sequence low-high-low-high. (C) Symmetry measure applied on the connectivity of the output population. In accordance to the target, connectivity switches between unidirectionality (low values) and bidirectionality (high values). (D,E) Mean values of the recovery time constant τ_rec and synaptic utilization U for the synapses projecting onto the output neurons. We observe depression (high values) at low firing rates and facilitation (low values) at high firing rates.