Such changes in pyramidal cell-interneuron transmission probability developed during learning

(Figures 4B and 4C). Moreover, these learning-related weight changes did not exhibit further changes after learning: the transmission probability observed at the end see more of learning remained stable in the following postprobe session with no further changes during sleep or probe sessions (Figures 4F and 4G). The observed changes in spike transmission to p/nInt interneurons occurred during the monosynaptic delay period (0.5–2.5 ms) only, and did not affect bins outside this delay at the 5ms bins (Figure 4D) or at the 30–50 ms bins. The changes in absolute value of the transmission probability were much smaller for the 5 ms or the 30–50 ms bins as compared selleck kinase inhibitor to the monosynaptic bins (first versus fourth learning quartile; 30–50 ms bin: 0.0084 ± 0.0009, 5 ms bin: 0.0071 ± 0.0019; p = 0.623) and not correlated with those at the monosynaptic bins (0.5–2.5 ms; p = 0.549) nor with those at the 5ms bins (p = 0.626). Similar results were found with pyramidal cell-interneuron cross-correlograms by measuring the correlation coefficients of spike coincidence, which measure is independent of the firing rate of both cells (Figures S6C–S6F). Moreover, other cell pairs that did not exhibit significant

monosynaptic peaks did not show such changes in transmission probability at the 2 ms monosynaptic latency bin, even though these cells underwent similar spatial

changes in firing rate (Figure 4E; n = 14522 pairs). Had local (spatial) changes in firing rate been the cause of the correlation changes of the monosynaptic pairs, they should have equally influenced bins at 5 ms or other cell pairs at 2 ms in which monosynaptic peaks were not detected. Thus, the observed changes in spike transmission probability could not be explained by changes in place-related firing of pyramidal cell and/or interneurons or by the firing associations we measured between them. These factors would have affected joint firing across longer time delays and not solely at monosynaptic latencies, and they would have also influenced correlations in which the monosynaptic connection has not been detected. It is unlikely that learning-related changes of in spike transmission probability were caused by theta phase-related changes as pyramidal cell-interneurons cross-correlograms did not exhibit visible theta modulation (Figures 4A and S6) and changes in theta firing preference of both interneurons and pyramidal cells were not related to changes in spike transmission probability (Figures S7 and S8). Inherently these changes were linked to spatial learning as no such learning-related changes in the coupling strength were observed in the intra-maze cued task (Figure S2F).