ERFs are locked in between fixation and saccade onset.

(A, B) The saccade duration binned ERF deflections followed the onset of preceding saccades. (C) Session-wise linear fit for the previously selected sensor for participant 4 between the duration of the preceding saccade and the latency of the fixation onset-locked ERF per saccade duration bin. The ERF latency is defined as the halfway point of the slope leading to the M100 (see dots in D and E; with 95% CI). (D) The fixation (left) and saccade (right) onset-locked ERF per saccade duration bin (nbins= 10). (E) Across all participants and gradiometers that show a locking at any point in time, the most optimal ERF alignment for inter- and extrapolated event onsets around saccade (α = −1) and fixation onset (α = 0) is in between the onset of these two events.

ERF alignment is best at the time point of peak saccade curvature.

(A) Conceptual overview of a real example saccade trajectory with the schematic temporal onsets of saccade-related events. (B) For each saccade, we constructed a saccade movie by extracting image crops centered on the gaze location, covering 3° of visual angle, and sampled at 1000 Hz. Motion energy was subsequently computed using the pymoten toolbox [16], resulting in a temporal motion energy profile for each saccade. (C) Onset of eye movement-related events per saccade duration bin (nbins = 10), averaged across participants. (D) Per independent component (IC; n = 80 per participant), we determined the event with the best ERF alignment across saccade duration bins and accumulated the number of ICs across participants for each event. (E) Accumulated explained variance of the neural signal based on ICs associated with each event, averaged across participants, with 95% confidence interval around the standard error of the mean. (F) Dipole locations of ICs across participants, color-coded based on the event that aligns saccade duration-binned ERFs best, and scaled based on explained variance in the neural signal. (G) IC with the best ERF alignment score per participant, locked to the onset of the optimal event.

Spatiotemporal differences between stimulus onset and fixation/peak saccade curvature ERFs.

(A) Topographies of scene, fixation and peak saccade curvature onset-locked ERFs, averaged across participants. (B) Example of Dynamic Time Warping (DTW) for one sensor and participant to compare the 3 different event onset-locked ERFs, applied between 50 and 200 ms after event onset. The dissimilarity score , where x and y are the two time series and K the length of the warping path) is obtained for each point after alignment. (C) Distribution of sensor-wise DTW dissimilarity scores per event comparison, averaged across participants. (D) Dissimilarity matrix showing the average DTW dissimilarity across participants and sensors. Trials of the same event type along the diagonal were split in half, and the number of observations was matched across splits to compute within-event dissimilarity.