Analytical investigation of network response after delay selection.

<p>Responses considered for ‘slow’ (blue), ‘medium’ (red), and ‘fast’ (green) EPSPs. (A) Amplitude of the Fourier transform of the three EPSPs as functions of frequency, as given by Equation (33) in Section 6 of Supporting <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002897#pcbi.1002897.s008" target="_blank">Text S1</a>. (B) Negative phase of the Fourier transform of the three EPSPs compared to the phase of the Fourier transform of the learning window (magenta) as functions of frequency, as given by Equation (31) and (33) in Section 6 of Supporting <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002897#pcbi.1002897.s008" target="_blank">Text S1</a>. (C) Analytically determined response of networks with delay profiles centered about the delay selected due to learning (<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002897#pcbi.1002897.e267" target="_blank">Equation (42)</a>) with input frequencies of 120 Hz (solid), 180 Hz (dashed), and 240 Hz (dot-dashed) inputs, with a profile width, , of 0.5 ms. These curves are given by <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002897#pcbi.1002897.e271" target="_blank">Equation (43)</a>. (D) Peak response frequency, , as a function of the training frequency, , of the network, for delay profiles with a width, , of 0.5 ms (solid) and 1 ms (dashed). The peak response frequency was numerically determined from the analytical formula in <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002897#pcbi.1002897.e271" target="_blank">Equation (43)</a>. The dashed line represents . Note that the dot ending some of the lines represents that, for higher training frequencies, there was no peak in the response amplitude with frequency. For plots C and D, a recurrent strength, , of 0.5 and a modulation amplitude, , of 5 spikes/s were used.