Hierarchical gradient of neuronal timescales.

Timescales for lorazepam (a), placebo (b), and D-cycloserine (c). Top row: Distribution of neuronal timescales in milliseconds (averaged across participants) across the cortex. Bottom row: Timescales of all 68 cortical areas (averaged across participants) as a function of T1w/T2w ratio (z-scored), an index of the hierarchical position of a cortical area. Shades around the regression line represent the 95% confidence interval. rS = Spearman correlation coefficient. Spatial autocorrelation was preserved for the computation of p-values.

Drug effects on neuronal timescales and their hierarchical gradient.

a Left panel: percent change of timescales under lorazepam compared to placebo. Right panel: t-statistic on neuronal timescales, permutation test of the effect of lorazepam compared to placebo, projected onto the cortical surface. b t-statistic on neuronal timescales, permutation test of the effect of D-cycloserine on neuronal timescales compared to placebo, projected onto the cortical surface. D-cycloserine did not significantly modulate neuronal timescales. c Cortical hierarchy (operationalized via the z-scored T1w/T2w ratio) of neuronal timescales as a function of drug condition (lorazepam or placebo). Shades depict the 95%-confidence intervals.

Spectral characterization of dynamic large-scale cortical networks.

For each state, three representations are presented, averaged across all participants in the placebo condition. Left: Cortical maps of state-specific 3-30 Hz power changes relative to the time-averaged power across states. Top right: Whole-brain average power spectrum for each state (black) compared with the whole-brain average spectrum across all states (grey). Bottom right: State-specific coherence networks in the 2–30 Hz range, thresholded at the 98th percentile to highlight the strongest functional connections.

Timescales vary across cortical networks and are differentially modulated by lorazepam.

a Percent change in neuronal timescales during network occurrences relative to the time-averaged timescale across states. b Percent change in neuronal timescales under lorazepam relative to placebo. Only parcels with significant effects are shown (p < 0.05, FDR-corrected). Significance was assessed for each parcel and state using within-subject GLMs with permutation testing, comparing state-specific timescales with the time-averaged timescale (a) or between lorazepam and placebo (b). p-values were FDR-corrected across parcels and states.

Lorazepam modulates large-scale cortical network dynamics.

a Lorazepam–placebo contrasts of fractional occupancy across all states. Each dot represents an individual participant. Lorazepam reduced the fractional occupancy of State 1 (dorsal attention network, DAN) and increased that of State 3 (frontal default mode network, DMN). b-d Lorazepam-placebo contrasts of state lifetimes (b), interval times (c), and state rates (d) for the DAN and frontal DMN. Dots indicate individual participants. e Lorazepam-placebo contrasts of state transition probabilities. Heatmap colours represent t-statistics from within-participant GLMs. Rows denote the current state and columns the subsequent state. Lorazepam reduced DAN self-transitions, increased transitions from the DAN and other networks into the frontal DMN, and reduced transitions out of the frontal DMN. f Schematic summary of significant lorazepam-induced transition changes involving the frontal DMN. Values indicate percentage change relative to placebo, with arrows denoting transition direction. Statistical significance in a-e was assessed using maximum t-statistic permutation tests, correcting for multiple comparisons across states or transitions. Asterisks indicate significance: p < 0.050 = *, p < 0.010 = **.