Fig. 1. Human cortical backbone sequences persist across time and cognitive states.

a Conceptual schematic demonstrating how a backbone sequence may arise from a given network simply by the pre-existing synaptic connections between neurons. b Cross-correlation of the spike train of a single unit with the summed spike trains of all other units from a given epoch. This unit had a peak firing time of ~30 ms before the other units in the population. The average waveform of the unit is shown in the inset. c A representative spike raster (left) demonstrating the ordering given by the corresponding backbone sequence (right). Cooler and hotter colors in the raster indicate units that, respectively, fire earlier and later in the backbone sequence. Colors in the backbone sequence are normalized between 0 and 1 for ease of visualization, and the associated color bar is shown on the right. d Backbone sequences extracted from 2 min non-overlapping epochs from rest to 40 min into the behavioral task. Red lines indicate the bin-specific backbone sequence while black lines indicate the backbone sequence extracted from the whole session. Here we show a separate patient than in (c) with fewer units in order to allow for visualization of changes to the backbone sequence over each epoch. e The backbone sequence across participants was consistent across the behavioral task regardless of the cognitive state at any point in time. Sequence similarity over time is calculated here as the similarity between the resting backbone sequence and the backbone sequence extracted from each subsequent 2 min non-overlapping bin. Error bars represent SEM across participants. Significance of each bin is denoted as a black dot (N = 6 participants, one-sided t-test against a null of 0), and testing for multiple comparisons is corrected by FDR (q = 0.05). f Average backbone sequences calculated for each different cognitive state during the episodic memory task. Sequences are shown ordered according to the whole session backbone sequence.