Oculomotor freezing reflects tactile temporal expectation and aids tactile perception

Abstract

The oculomotor system keeps the eyes steady in expectation of visual events. Here, recording microsaccades while people performed a tactile, frequency discrimination task enabled us to test whether the oculomotor system shows an analogous preparatory response for unrelated tactile events. We manipulated the temporal predictability of tactile targets using tactile cues, which preceded the target by either constant (high predictability) or variable (low predictability) time intervals. We find that microsaccades are inhibited prior to tactile targets and more so for constant than variable intervals, revealing a tight crossmodal link between tactile temporal expectation and oculomotor action. These findings portray oculomotor freezing as a marker of crossmodal temporal expectation. Moreover, microsaccades occurring around the tactile target presentation are associated with reduced task performance, suggesting that oculomotor freezing mitigates potential detrimental, concomitant effects of microsaccades and revealing a crossmodal coupling between tactile perception and oculomotor action.

Fig. 1. Setup, procedure, and design.

a Setup. Participants sat at a table, their head supported by a chin and forehead rest, and fixated straight ahead while their eye position was monitored. Tactile stimulators were attached to the nondominant hand; the dominant hand rested on a keyboard. b Trial timeline. Trials began contingent on 0.5 s of continuous fixation, followed by a variable time interval of 0.2–0.7 s, ensuring that the stream of tactile stimuli within any block was nonrhythmic. Tactile cue and target were separated by a foreperiod of 1, 1.5, 2, 2.5, or 3 s. The cue was a single protruding movement of the stimulator tip; the target stimulus was a 50-ms-long vibration. Participants indicated by button press whether they perceived the target frequency as faster or slower than 60 Hz. c Design. We manipulated the degree of temporal predictability by either keeping the foreperiod (cue–target interval, blue ribbons) constant—regular condition—or variable—irregular condition—within blocks.

Fig. 2. Task performance and temporal predictability.

Effects of temporal predictability condition (dark blue, regular; light blue, irregular) and foreperiod (x-axis) on a reaction times and b proportions of correct responses. Boxplots indicate the distribution of participant-level mean values per condition adjusted by their overall mean (center line, median; box limits, upper and lower quartiles; whiskers, minimum and maximum limited by 1.5× interquartile range). Circular markers show group-level mean values; the width of the ribbon around each marker equals the predictability-condition-adjusted standard error, which indicates the degree of intersubject variation in the difference between regular and irregular conditions, and therefore whether there is an effect of predictability condition on the dependent variable. All statistics are based on the full dataset (N = 30 participants, 100 repetitions per each of the 10 conditions and participant), and source data are provided as a Source Data file.

Fig. 3. Microsaccade frequency and temporal predictability.

a Group-average microsaccade rates as a function of trial time relative to the onset of the tactile target stimulus separately for each predictability condition (dark blue, regular; light blue, irregular) and foreperiod (panels). Shaded vertical bars indicate the cue and target stimulus (blackish gray), shaded rectangles the post-cue (medium gray), pretarget (light gray), and post-target (dark gray) intervals. b Group-average microsaccade rate timelines relative to the cue onset separately for each predictability condition (panels) and foreperiod (red shades). c Microsaccade rates in a comparison interval 300–500 ms after the onset of the tactile cue (left panel), in the 200-ms interval before the onset of the tactile target stimulus (center panel), and in a post-target interval 0–200 ms after the offset of the tactile stimulus (right panel), separately for each predictability condition and foreperiod (x-axis). Boxplots indicate the distribution of participant-level mean values per condition adjusted by their overall interval mean (center line, median; box limits, upper and lower quartiles; whiskers, minimum and maximum limited by 1.5× interquartile range). Circular markers show group-level mean values; the width of the ribbon matches the predictability-condition-adjusted standard error that indicates the degree of intersubject variation in the difference between regular and irregular conditions, and therefore whether there is a significant effect of predictability condition on microsaccade rates. All statistics are based on the full dataset (N = 30 participants, 100 repetitions per condition and participant), and source data are provided as a Source Data file.

Fig. 4. Microsaccades and task performance.

a, b Microsaccades by task performance. Group-average microsaccade rates as a function of time relative to target onset (vertical gray line) split by a reaction times (blue: fast, green: slow responses) and b response accuracy (blue: correct, green: incorrect responses) in regular (dark shade) and irregular (light shade) blocks. Temporal clusters with significant differences in microsaccade rates (RT: p < 0.001, accuracy: p < 0.001, two-sided cluster permutation tests) between performance categories are indicated by shaded horizontal bars. Each tile corresponds to a 200-ms bin with a significant difference (p < 0.05, two-sided permutation tests); darker shades indicate overlap between bins. c Task performance by microsaccade latencies. Gray lines indicate the 2d frequency distribution of single-trial reaction times (upper panel) and correct responses (lower panel) as a function of the latency of the last microsaccade before the target stimulus. Circular markers show group mean reaction times in regular (dark blue) and irregular (light blue) conditions for 200-ms-long time bins. Marker size represents the percentage of trials per bin. Error bars indicate standard errors corrected for between-participant variability. d Task performance by microsaccades. Reaction times (upper panel) and proportion correct (lower panel) split by the presence of microsaccades in the pretarget, target, and post-target interval (dark red, absent; light red, present). Boxplots indicate the distribution of participant-level mean values per condition adjusted for their overall mean (center line, median; box limits, upper and lower quartiles; whiskers, minimum and maximum limited by 1.5× interquartile range). Note that trials without a microsaccade in the respective interval were more frequent than those with a microsaccade resulting in a narrower distribution. Circular markers show group-level mean values. Vertical gray lines indicate the standard error of the difference between conditions; p values are based on generalized linear mixed models predicting single-trial performance from the presence of a microsaccade (see “Results” for full statistical information). All statistics are based on the full dataset (N = 30 participants, 100 repetitions per condition and participant) and weighted by the number of trials per condition. Source data are provided as a Source Data file.