The learning-induced modulation of the inhibitory unitary synaptic events amplitude is dominated by a single process in which the events amplitude was doubled.

<p><b>A.</b> For each neuron an amplitude distribution curve was reconstructed from the mIPSC's amplitudes and was normalized (data taken form <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0068131#pone.0068131-Saar2" target="_blank">[17]</a>). Averaged amplitude histograms, were calculated for all neurons from the three groups. Notably, a significant portion of events in the trained group are of higher values. <b>B.</b> For each cell the weight for the first component (X-axis) was drawn against its weight for the second component (Y-axis). Only the weights of the first component are significantly different between groups. <b>C.</b> The curve calculated by subtracting the averaged pseudo amplitude distribution curve from the averaged trained amplitude distribution curve (orange) match the first component calculated by PCA (black). <b>D.</b> For each multiplication factor a different curve that describes the difference between groups assuming a multiplication model was calculated. The calculated curve assuming multiplication factor of 2 (black) matched the curve that describes the main difference between groups (orange). Inset: the correlation coefficient was calculated for each multiplication factor (for calculation of R only amplitudes >13pA were used, since at lower amplitudes, multiplication factors bigger than two requires unavailable data in amplitudes <6pA, see Methods).