(a) Overall schema: (1) Cortical DS convergence leads to thalamic DSs; (2) thalamic DSs produce thalamic spindles; and (3) thalamic spindles drive cortical spindles. (b) Single trial example of a cortical DS leading a thalamic DS. (c) Cortical recording sites. Asterisks indicate cortical locations whose DSs significantly lead DSs in one particular thalamic site, indicated by label, paired to each subject's cortical channels (colour-coded). (d) Thalamic recording sites. (e) Example transcortical bipolar pair. (f) Thalamic DS peaks are locked at 0 ms (vertical black line) and the number of cortical DS peaks occurring within ±2 s are plotted in red 50 ms bins. Cortical DSs occur significantly (*) more often before thalamic DSs than after (). Representative corticothalamic pair; additional examples in . Similar results were obtained in N2 and N3 (). (g) Thalamic DS probability significantly increases when two cortical locations, compared with one, produce a DS. Boxes along the diagonal show the probability of a medial pulvinar DS peak occurring within 500 ms after a cortical DS peak in the cortical channel listed to the left. The remaining boxes indicate the probability of a medial pulvinar DS peak occurring within 500 ms after a cortical DS peak in the cortical channel listed to the left, given a DS in the cortical channel listed on top also occurs within 500 ms after the cortical channel listed to the left (*P<0.05, χ², Bonferroni corrected).

Spindles start, on average, at the thalamic DS peak and end near the following upstate peak. (a) Representative single trace. (b) Histograms of thalamic spindle onsets (first column) or spindle terminations (second column) in relation to the thalamic DS peak at 0 ms (vertical orange line) for each thalamic channel; counts in 50 ms bins. Waveforms show the averaged local field potential in each channel. Spectral power from 5 to 100 Hz (third column), or from 5 to 20 Hz (fourth column), averaged on the DS peaks at 0 ms, baseline corrected over entire epoch, thresholded at P<0.01, uncorrected. Representative channels; additional channels in . Similar results were obtained in N2 and N3 () and with stricter DS detection parameters ().

(a) Cortical sites. Asterisks indicate cortical locations where thalamic spindles significantly lead cortical spindles, in a particular thalamic site/subject. (b) Thalamic spindle onsets are locked at 0 ms (vertical black line) and the number of corresponding cortical spindle onsets occurring within ±2 s are plotted in red 50 ms bins. Cortical spindles occur significantly (*) more often after thalamic spindles than before (). Representative corticothalamic pair; additional examples in . Similar results were obtained in N2 and N3 (). (c) Single trial of overlapping medial/lateral pulvinar and annectant gyrus spindles. (d) Bandpassed (10–16 Hz) average of the 81% overlapping detected spindles, locked to the pulvinar spindle onsets (vertical black line), indicating synchronous spindle occurrence. (e) In the same pair, locked to individual pulvinar spindle troughs (black trace) in the local field potential (LFP), the averaged cortical LFP spindle troughs are synchronous to the pulvinar (left); however, the corresponding high gamma amplitude in the pulvinar leads the cortex by 14.7 ms (right).

(a) Histograms of cortical spindle onsets in relation to the cortical DS peak at 0 ms (vertical orange line) for each cortical channel in 50 ms bins. Waveforms show the averaged local field potential in each channel. Representative channels; additional channels in . Cortical spindle onset occurs during the down-to-upstate transition in the cortex. Similar results were obtained in N2 and N3 () and with stricter DS detection parameters (). (b) Both slow and fast spindles occur on the down-to-upstate transition in the cortex. Results are z-scored on 17 channels in which at least 100 slow (>10 and ≤12 Hz) spindles occurred within±500 ms of a DS and on 12 channels in which at least 100 fast (>12 and <16 Hz) spindles occurred within±500 ms of a DS. The red arrow indicates the point, around 250 ms after the cortical DS, when both slow and fast cortical spindles show a peak in start times.

Convergence of cortical DSs produces thalamic hyperpolarization (that is, a DS) by removing depolarizing input from the cortex (that is, disfacilitation, open red arrows; see ). This thalamic hyperpolarization de-inactivates I_T and activates I_h, evoking a thalamic spindle that is reinforced by rhythmic feedback inhibition from thalamic reticular neurons (see ). The thalamic spindle, in turn, is focally projected to the cortex (see ), arriving during the cortical down-to-upstate transition (see ). PY, pyramidal; RE, thalamic reticular; TC, thalamocortical.

DSs detected in the medial pulvinar/lateral pulvinar channel of Subject 3 were sorted based on spindle-band amplitudes. The spindle-band amplitude was measured by taking the maximum of the Hilbert amplitude between 0 and 500 ms (DS peak at 0 ms) calculated after applying a 10–16 Hz bandpass filter. (a) The DSs corresponding to the top 10% of spindle-band amplitudes show a significant increase in both spindle-band and high gamma power. (b) The DSs corresponding to the bottom 10% of spindle-band amplitudes do not show a significant spindle-band increase, but do continue to show a significant increase in high gamma power. For both (a,b) DSs were epoched from −2.3 to 2.3 s around the DS peaks at 0 ms. Time frequency plots are from 5 to 500 Hz and baseline corrected over the entire epoch. Significance level was set to P<0.01, uncorrected. Waveforms show the averaged local field potential for the same trials.

(a) Spindles are manually marked (red lines) on 10 min of N2 and 10 min of N3 raw data, per channel. (b) The amplitude of peaks in the amplitude envelope corresponding to manually marked spindles are calculated (manual amplitude, black circles). (c) A per channel threshold (purple line) is applied to the amplitude envelope based on the lowest manual amplitude for that channel. Peaks, marked by red asterisks, which exceed that threshold are kept as candidate epochs. (d) The closest peak in the edge envelope to the selected peak in the amplitude envelope is detected as the edge amplitude (blue vertical lines). The bounds of the onset and offset of the spindle (pink traces) are calculated at 45% or greater of this edge amplitude. (e) The final candidate epochs highlighted in red include epochs where spindles had (red lines) and had not (black arrow) been manually marked.