Perception during double-step saccades

Abstract

How the visual system achieves perceptual stability across saccadic eye movements is a long-standing question in neuroscience. It has been proposed that an efference copy informs vision about upcoming saccades, and this might lead to shifting spatial coordinates and suppressing image motion. Here we ask whether these two aspects of visual stability are interdependent or may be dissociated under special conditions. We study a memory-guided double-step saccade task, where two saccades are executed in quick succession. Previous studies have led to the hypothesis that in this paradigm the two saccades are planned in parallel, with a single efference copy signal generated at the start of the double-step sequence, i.e. before the first saccade. In line with this hypothesis, we find that visual stability is impaired during the second saccade, which is consistent with (accurate) efference copy information being unavailable during the second saccade. However, we find that saccadic suppression is normal during the second saccade. Thus, the second saccade of a double-step sequence instantiates a dissociation between visual stability and saccadic suppression: stability is impaired even though suppression is strong.

Figure 1

(A) Timecourse of presentations. Subjects made double-step saccades from a fixation point (FP) to the first saccadic target (ST1), and from there to a second saccade target (ST2). In experiment 1, two dots were shown around the time of the second saccade, in the top (Top Dot, TD) and bottom half of the screen (Bottom Dot, BD), both for one monitor frame. In experiment 2, a grating (G) was shown for one monitor frame at a variable time, before, during or after the double-step saccade sequence. (B) Spatial arrangement of the stimuli in experiment 1, where apparent motion was perceived between two dots shown across the second saccade of the double-step sequence. The dashed line (not part of the experimental display) indicates the direction of apparent motion that subjects typically reported in this condition, for dots that were physically aligned: motion was slightly slanted to the right, indicating under-compensation of the saccade vector. (C) Control condition of experiment 1, where the apparent motion stimulus was shown across a single saccade between ST1 and ST2. In this condition, the direction of apparent motion was typically slanted to the left, indicating over-compensation of the saccade vector, consistent with previous studies (Szinte and Cavanagh). (D–E) Spatial arrangement of the stimuli in experiment 2, with fixation and saccade targets in panel D and the contrast discrimination stimulus in (E) (the latter could be shown with this or the opposite polarity, at variable contrast levels).

Figure 2

(A) Average frequency distribution of saccade latencies for the first (blue) and second (red) saccades in the double-step sequence. Histograms are computed in individual subjects then averaged across subjects, with error bars representing SEM. (B-C) Average saccade amplitudes (B) and peak velocities (C) for the first (blue) and second saccades (red) in the double-step saccade paradigm and for the single saccades in the control condition (shown in gray). Error bars represent SEM.

Figure 3

Average apparent motion bias for stimuli presented across the second saccade of the double-step sequence (red) and across a single saccade in the control condition (gray). Error bars represent SEM. A positive/negative bias implies a bias towards seeing rightward/leftward motion (illustrated by the dashed lines in Fig. 1B/C). A negative bias is consistent with a slight overcompensation of the saccade vector.

Figure 4

Contrast sensitivity as a function of probe presentation time relative to saccade onset for a representative subject. Data are binned relative to onset of the first saccade (shown in blue) or relative to onset of the second saccade (shown in red).

Figure 5

(A) Orange points give perisaccadic sensitivity during the second (ordinate) vs. the first (abscissa) saccade in a double-step sequence, and gray points give sensitivities for the control condition, with single saccades of amplitude matched to the second or first saccade of the sequence respectively. Each point is one subject. Maximal contrast in our setup was 0.5, thus sensitivity below 2 was not measurable and is indicated in the figure by the tick mark “<2”. (B) Average perisaccadic sensitivity during the first and second saccade of the double-step sequence (red and blue respectively) and during single-saccades of matched amplitude (gray) Error bars represent SEM.