Supplementary MaterialsS1 Fig: Information on ripple property estimation. a different one) in the next ripple (n+1-th ripple). If the system has memory, then a property of a ripple is predictive of the same (or a different) property in the next ripple. In a system with no memory, the cloud distribution should look like the direct product of the distributions of the two properties considered. (c) Ripple frequency does not show an obvious memory effect. Note that the distribution in n vs n+1 looks like the direct product of Fig 1B with itself. (d) Ripple duration does not show memory effect across ripples. Compare with Fig 1D times (outer product) itself. (e) Current ripple frequency does not influence next ripple duration. (f) Current ripple duration does not affect next ripple frequency.(TIF) pcbi.1004880.s002.tif (2.9M) GUID:?D9736D2E-F6F2-44F2-9402-8A5442729290 S3 Fig: Autocorrelation of firing probability of interneurons (red) and pyramidal cells (black) shows no background frequency properties in the network. (TIF) pcbi.1004880.s003.tif (230K) GUID:?86CEA779-2D44-45BF-A77E-CF29B3726F9C S4 Fig: Changing the magnitude of CA3 input affects ripple amplitude, frequency and duration. (a) Example of a simulation in which CA3-mediated input current in both pyramidal cells and interneurons of the CA1 model is decreased to 30% of its baseline magnitude, by multiplying by 0.3. Notice how big is y-axis at the top sections. The proper column shows an inferior time interval, so the ripple profile is seen. Time is within seconds in Cyclandelate every sections. Top sections: current insight (in pA) from CA3 to pyramidal cells (dark) and interneurons (reddish colored). Second Cyclandelate sections from the very best: rastergram of pyramidal cells (dark) and interneuron (reddish colored) spikes. Middle sections: possibility of spiking for pyramidal cells (dark) and interneuron (reddish Cyclandelate colored) populations, in 1ms period bins. Last two sections: wide music group (above) and filtered (100C300 Hz) LFP track (in V). (b) Identical to in (a), but also for CA3-mediated current just scaled to 80% of its baseline power. Note the way the interneuron inhabitants fires more structured, which outcomes in a filtered LFP even more organized with this complete case, set alongside the 30% scaling. (c) Overview storyline of primary ripple properties once the insight from CA3 to both pyramidal cells and interneurons can be scaled in a variety of 10C100%. Ripple amplitude increases from 5 V (undetectable) as insight size raises, and saturates between 80C100% from the insight. Ripple duration OBSCN can Cyclandelate be ill-defined at 10% insight (note the fantastic variability as several events that be eligible for ripple recognition do not display enough oscillations for the duration to be consistently estimated), and increases with increasing input amplitude. Ripple frequency is over-estimated below 30% due to the 100C300 Hz filtering in ripple detection, but once input is above 30% one can see the shift from high-gamma to ripple range, controlled by input size.(TIF) pcbi.1004880.s004.tif (3.4M) GUID:?B1A7EAAD-D27E-4D69-BC31-9DAC6F57005B S5 Fig: Network activity without I-to-I synapses. (a) Example of a typical ripple event in the network when I-to-I synapses are removed. Top: input current (in pA) from CA3 to pyramidal cells (black) and interneuron (red) population. Middle: rastergram of pyramidal cells (black) and interneurons (red) spikes during a ripple. Lower plot: wide-band (black) and filtered (100C300 Hz, red) LFPs in the network. Note that the oscillations stop much quicker than in the network with I-to-I inhibition shown in Fig 2. (b) Summary histograms for ripple frequency (Hz), duration (ms) and amplitude (V) in the case of removed I-to-I synapses. The overall properties of ripples are on average preserved (as expected), yet the filtered LFP is unable to ever generate ripples longer than 60m, compare with Fig 3B.(TIF) pcbi.1004880.s005.tif (2.5M) GUID:?44AFB365-878F-4996-AEE4-460041E1AFF6 Data Availability StatementCode is available on Model DB, https://senselab.med.yale.edu/ModelDB/ShowModel.cshtml?model=188977. Abstract Memories are stored and consolidated as a result of a dialogue between the hippocampus and cortex during sleep. Neurons active during behavior reactivate in both structures during sleep, in conjunction with characteristic brain oscillations that may form the neural substrate of memory consolidation. In the hippocampus, replay occurs within sharp wave-ripples: short bouts of high-frequency activity in area CA1 caused by excitatory activation from area CA3. In this work, we develop a computational model of ripple generation, motivated by rat data showing that ripples have a broad frequency distribution, exponential inter-arrival times and yet highly non-variable durations. Our study predicts that ripples are not persistent oscillations but result from a transient network behavior, induced by input from CA3, in which the high frequency synchronous firing of perisomatic interneurons does not depend on the time scale of synaptic inhibition. We discovered that noise-induced lack of synchrony among CA1 interneurons.