From the kinetic information obtained for isolated PSCs, the former are expected to be steeper even when several time-shifted PSCs overlap. The values of the derivative at such extremal points are shown as cumulative histograms in Figures 4B and 6D. We found that for our data, FFT does selleckchem not allow quantitative statements on the fast structure of cPSCs because of their short duration and the disproportionate weight assigned to events with higher-amplitude oscillations. Also, the temporal fine structure differs from event to event. This variability

is particularly problematic for analyses in the frequency domain, as for high frequencies f a small temporal jitter Δt is transformed to a large frequency jitter Δf = −f2/Δt PFT�� order owing to the reciprocal relation f = 1/t. To quantify the rhythmicity of fast network input during cPSCs, we instead analyzed the intervals between strong slopes within cPSCs ( Figure 4C: 10% strongest interdownward slope intervals in red, 25% strongest in gray; 0.5–400 Hz trace; see also Figures 6E and

S4B). The ripple band peak is robust for a wide range of low-pass frequencies (400–600 Hz) and fractions of selected strongest slopes (1%–25%). Filtering out higher frequencies is required to avoid too many local extrema of the derivatives being detected for what is in practice

the same onset. EPSCs outside SWRs (“spontaneous PSCs”) were detected among the cells Ketanserin recorded at −66 mV as strong downward slopes (top 5% of all maximal slopes on the 0.5–400 Hz filtered intracellular trace of each cell). A total of 1,000 events from 5 cells were postselected by eye to exclude those that are too small to be distinguished from noise and also to avoid multiple events where a second PSC arrives during the tail of the first. The fit was performed with an alpha function α(t;A,t0,τd,τr)=AN(τd,τr)Θ(t−t0)[e−(t−t0)τd−e−(t−t0)τr],where N(τd,τr) is a normalization factor so that the fitted amplitude A corresponds to the maximum value attained by the function, and Θ(t-t0) is the Heaviside step function defined as 1 for positive arguments (t later than onset t0) and 0 elsewhere. The average time constants obtained from the fits were of τr = 1.70 ± 0.04 ms and τd = 4.04 ± 0.08 ms. Here our PSC detection algorithm is based on slopes that are influenced by both time constants. To highlight the separation of timescales used by the algorithm, we plot histograms of durations of rises and decays (20%–80% and 80%–20% of maximum amplitude) instead of time constants ( Figure 4A).