Fig. 3: Entorhinal cortex layer III input inhibition weakens theta phase precession but enhances theta sequences.

a Representative examples of perturbed phase precession, with decreased phase range and slope, during periods of optogenetic manipulation (laser ON). Top, theta phase-position raster plot. Bottom, phase-position density plot. 0 = peak of theta oscillation in the pyramidal layer. Field-phase regression line in red. b Theta-phase firing probability distribution during medium gamma (MG)-network states for phase locking (n = 103) and and phase precessing (n = 199) populations, during laser OFF and laser ON periods (p < 10e-3, t-test, Multiple Comparison Correction). c Top row, global average of the in-field spike probability during medium gamma network states, in laser OFF (left) and laser ON (right), for phase locking (n = 103) and phase precessing (n = 199) over two theta cycles. Phase 0 = peak of theta oscillation in the pyramidal layer. Bottom row, contour of regions with highest density of spikes for phase locking and phase precessing fields, curves correspond to 90 % and 85 % of the peak value of spike density (same data as top row panels). Two-Dimensional Kolmogorov-Smirnov test for spike densities (p = 10e-3). d Phase-position slope values plotted as a function of the Theta Score for each individual field, for periods of slow gamma (SG, blue) and medium gamma (MG, red) network states. Left, laser OFF. Right, laser ON. Spearman correlation of Slope vs. Gamma Ratio for each Theta Score quantile. N = 6 gamma quantiles, p = 0.0167. e Place field distance vs phase distance regression for periods of elevated SG (blue) or MG (red) and laser conditions (OFF and ON). All comparisons are statistically significant. Confidence intervals, 95th percentiles, N = 2218 cell pairs, p = 10e-3. f Examples of encoded position (y-axis) reconstruction over a theta cycle (x-axis, 1 bin = 10 ms). Black-white color scale: posterior probability from Bayesian decoding. Dotted gray line: real position of the animal, used to compute the amount of local coding as the probability density around the line. Red line: best positive-slope fit of probability distribution, used as an estimation of sequential non-local cell activation. g Comparison of mean positional and sequential scores from Bayesian decoding between different gamma conditions (SG in blue, MG in red, all periods in black). Paired t-test, N = 11 sessions. P values position: p = 0.826, p = 0.002, p = 0.013. P values sequence: p = 0.009, p = 0.130, p = 0.018. n.s. non-significant; *p < 0.05; **p < 0.01.