Delay Selection by Spike-Timing-Dependent Plasticity in Recurrent Networks of Spiking Neurons Receiving Oscillatory Inputs

doi:10.1371/journal.pcbi.1002897

Delay Selection by Spike-Timing-Dependent Plasticity in Recurrent Networks of Spiking Neurons Receiving Oscillatory Inputs

Figure 6

Simulations of peak network response after delay selection.

Response plots showing the peak in the periodic response of networks of LIF neurons plotted as a function of the modulation frequency, (‘medium’ EPSPs used unless otherwise specified). (A) Response plot for networks after 20,000s of learning with input modulation frequencies of 100 Hz (blue), 120 Hz (red), 140 Hz (green), 180 Hz (cyan), 240 Hz (magenta), and 300 Hz (yellow), as shown in Figure 4D. The simulations used ‘medium’ EPSPs (solid), except for two which used the fast EPSP (dashed and dot-dashed). The weights in the networks trained with 240 Hz and 300 Hz inputs were scaled down slightly (to about 0.99 of their original value) so that the networks were below criticality. (B) Response plot for the network trained with 100 Hz inputs in A, with the weights all scaled by 0.90 (dotted), 0.95 (dot-dashed), 0.98 (dashed), 1.00 (solid).

doi: https://doi.org/10.1371/journal.pcbi.1002897.g006