The phase of ongoing EEG oscillations predicts the amplitude of peri-saccadic mislocalization

Abstract

Our constant eye movements mean that updating processes, such as saccadic remapping, are essential for the maintenance of a stable spatial representation of the world around us. It has been proposed that, rather than continually update a full spatiotopic map, only the location of a few key objects is updated, suggesting that the process is linked to attention. At the same time, mounting evidence links attention to oscillatory neuronal processes. We therefore hypothesized that updating processes should themselves show oscillatory characteristics, inherited from underlying attentional processes. To test this, we carried out a combined psychophysics and EEG experiment in human participants, using a saccadic mislocalization task as a behaviourally measureable proxy for spatial updating, and simultaneously recording 64-channel EEG. We then used a time-frequency analysis to test for a correlation between oscillation phase and perceptual outcome. We found a significant phase-dependence of mislocalization in a time-frequency region from around 400 ms prior to saccade initiation and peaking at around 7 Hz, principally apparent over occipital electrodes. Thus the degree of perceived mislocalization is correlated with the phase of a theta-frequency oscillation prior to saccade onset. We conclude that spatial updating processes are indeed linked to rhythmic processes in the brain.

Figure 1

(A) Stimulus paradigm. The screen background was not white as shown here, but rather a dark red. Trials began when participants fixated the point at the left of the screen. The mislocalization probe appeared (all other elements were already onscreen), and fell towards the bottom of the screen. Participants were instructed to make a saccade to the saccade target point from around the time when the mislocalization probe passed through the demarcated zone. The mislocalization probe subsequently disappeared, and participants had to report the maximum horizontal extent (in either direction, but typically towards the saccade target) of any perceived deviation of the probe (which we take as a measure of mislocalization) at the time of the saccade. (B) Example sketch of one subject’s typical perceptual experience of the probe trajectory. The measured variable reported by subjects on each trial was the peak horizontal displacement (indicated by a star on the plot). (C) Histogram of saccade latency (all subjects, n = 12), relative to the time at which the mislocalization probe entered the saccade initiation zone. (D) Histogram of reported mislocalization (all subjects, n = 12), negative numbers indicating a perceived leftwards deviation (i.e. towards the saccade target, against the direction of the saccade).

Figure 2

(A) Time-frequency plot of pre-saccadic phase-opposition between trials with different perceptual outcome, averaged across all participants and electrodes. Significance (p) values are plotted (fixed effects analysis, see methods). Empirical values were compared directly to a large set of surrogate distributions (10⁹ repeats), setting a limit on the minimum p-value that can be measured. For equivalent parametrically estimated values, see Supplementary Figure S1. The transparent red region around time-zero indicates the window of influence of the wavelet used for the time-frequency analysis, and thus susceptible to contamination from the saccade event at time zero. For the remainder of the plot, a transparent blue layer has been set at the 0.05 significance level (Bonferroni-corrected for the number of time and frequency comparisons), to emphasize higher values. The small area plot at the left hand side shows the profile of significance values across frequencies, measured at the time of the highest value (−356 ms). (B) Time-frequency plot, showing the outcome of a random-effects style analysis to test for regions of significant phase locking, corrected for multiple comparisons using a cluster-based technique (see Methods). The plot shows a count of the number of electrodes (out of 64 total) having a significant cluster incorporating the relevant time-frequency point. We focus on a smaller frequency range for greater detail. No significant clusters were apparent outside of this range. (C) Topographic plots showing the distribution of phase opposition hotspots for the region of interest identified above (for each electrode, phase-opposition z-scores averaged across all time-frequency pixels that showed significance over more than two electrodes).

Figure 3. Modulation of mislocalization as a function of EEG phase, calculated for the individual time-frequency point showing highest phase opposition (electrode POz, −358 ms, 6.6 Hz).

The plot shows the average (±S.E.M) of smoothed curves fit to the data for each subject, extended to two oscillation cycles for clarity. The inset shows the power spectrum of this average curve, with a peak at the fundamental frequency of 6.6 Hz as expected, and a second peak at the first harmonic frequency (13.2 Hz) emphasizing the non-sinusoidal nature of the curve.