Attention was successfully shifted and maintained.

(A) Schematic showing the progression of a single example trial. A colour change of the central fixation dot cued which of two visual stimuli was most likely to change orientation (the visual target). After a variable delay, the target event (the orientation change) occurred in either the cued (80% of trials, valid) or other (20%, invalid) stimulus and participants reported whether it was clockwise or anticlockwise. Colours shown serve only as an example, for the experimental colours, see the methods section. Supplementary Figure S1 shows the stimulus screen and stimulus sizes and distances to scale. (B) Behavioural performance (reaction times above, accuracy scores below) as function of cue-target interval for validly vs. invalidly cued targets. Circular markers indicate mean values, with shaded areas indicating the standard error of the mean. Bars reflect the mean behavioural performance, with whiskers indicating the standard error of the mean. Grey lines indicate individual participants’ data. Throughout the entire figure, the following significance levels were used: *: p< 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Attention was accompanied by a bias in microsaccade direction.

(A) Time courses of saccade rates for saccades towards the cued and the other item. This analysis only includes downward saccades, as stimuli were positioned at a 45° angle below fixation, allowing us to distinguish saccades towards the item from those returning to central fixation. The time courses indicate mean values, with shaded areas indicating the standard error of the mean. (B) Time courses of the attentional modulation of saccade direction (toward cued – toward other). The time courses indicate mean values, with shaded areas indicating the standard error of the mean. Black horizontal lines indicate significant temporal clusters (as determined by a cluster-based permutation test). (C) Time courses of attentional modulation (as shown in B) as a function of saccade size. For reference, dashed horizontal lines indicate 1 degree visual angle, as well as the locations of the centre and closest border of the visual stimulus. Throughout the entire figure, the following significance levels were used: *: p< 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Microsaccade directions are biased predominantly during attentional shifting.

Bars represent the mean saccade towardness rate in the denoted timeframes, with whiskers indicating the standard error of the mean. Distribution and scatter plots show individual participants’ data, with dark lines indicating the mean. Note that the y-axes have different scales between Experiment 1 and Experiment 2 and between bar graphs and distribution plots, for visualisation purposes. Throughout the entire figure, the following significance levels were used: *: p< 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Cue-validity effects at each of the tested cue-target intervals.

This table splits the cue-validity effects from Figure 1 into the ten possible cue-target intervals. The top half of the table shows the effects of cue-validity on reaction time; the bottom half of the table shows the effects of cue-validity on accuracy. Results from Experiment 1 and 2 are shown, respectively, on the left and right sides of the table. The table shows both uncorrected and Bonferroni corrected p-values, along with their significance levels: *: p< 0.05, **: p < 0.01, ***: p < 0.001.

Average number of trials per participant per condition and average number of saccades per participant per condition.

This table reflects the amount of data included in the main analyses shown in Figure 1, Figure 2 and Figure 3.

Size of the experimental screen, stimuli, and distances to scale.

Full time courses of saccade data.

This figure replicates the results shown in Figure 2 using the complementary analysis described in the Methods section. (A) Time courses of saccade rates for saccades towards the cued and the other item. This analysis only includes downward saccades, as stimuli were positioned at a 45° angle below fixation, allowing us to distinguish saccades toward the item from those returning to central fixation. The time courses indicate mean values, with shaded areas indicating the standard error of the mean. (B) Time courses of the attentional modulation of saccade direction (toward cued – toward other). The time courses indicate mean values, with shaded areas indicating the standard error of the mean. Black horizontal lines indicate significant temporal clusters (as determined by a cluster-based permutation test). (C) Time courses of attentional modulation (as shown in B) as a function of saccade size. For reference, dashed horizontal lines indicate 1 degree visual angle, as well as the locations of the centre and closest border of the visual stimulus. Throughout the entire figure, the following significance levels were used: *: p< 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Polar histograms of all saccades included in the main analyses.

This figure shows the percentage of detected saccades going in a specific direction for both experiments. This visualisation is repeated for the three main timeframes of interest: (1) the whole trial (from -100 to 1400 ms after cue onset), (2) the ‘shift’ period (from 200 to 600 ms after cue onset) and (3) the ‘maintain’ period (from 600 to 1400 ms after cue onset). The coloured saccades are those classified as ‘downwards’ saccades in the analyses of Figure 2. The radial axis shows the percentage of detected saccades going in one of twenty binned possible directions, which is calculated by weighting each participant equally (i.e., averaging the proportion of saccades detected for each participant per polar angle bin).

Attentional bias in upwards saccades.

This figure repeats the analysis shown in Figure 2, but for upwards saccades instead of downwards saccades. Note that the visual items were always downward in our task. (A) Time courses of saccade rates for saccades away from the cued and the other item. This analysis only includes upwards saccades. As stimuli were positioned at a 45° angle below fixation, this mainly reflects a return to central fixation. The time courses indicate mean values, with shaded areas indicating the standard error of the mean. (B) Time courses of the attentional modulation of saccade direction (away cued – away other). The time courses indicate mean values, with shaded areas indicating the standard error of the mean. Black horizontal lines indicate significant temporal clusters (as determined by a cluster-based permutation test). Throughout the entire figure, the following significance levels were used: *: p< 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Relationship between the magnitude of the cue-related attentional modulation and the magnitude of the spatial saccade bias.

This figure shows the relationship between the cue-related attentional modulation (referred to in the figure as a ‘benefit’ for simplicity) and the average saccade bias during the ‘shift’ period (from 200 to 600 ms after cue onset), for both reaction time and accuracy. Each dot represents one participant. Throughout the entire figure, the following significance levels were used: *: p< 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Gaze density during the ‘shift’ period.

This figure shows the gaze density during the ‘shift’ period (from 200 to 600 ms after cue onset) for both experiments, separately for left cued trials and right cued trials. The very right column shows the difference in gaze density between left and right cued trials.

Relationship between amplitude and peak velocity of all detected saccades.

This figure shows the relationship between the amplitude (in degrees visual angle) and peak velocity (in degrees visual angle / millisecond), for Experiment 1 and Experiment 2 separately. A visual inspection of both plots shows that the detected saccade generally follow the main sequence.

Re-analysis of saccade data with narrower angular definitions of ‘toward cued’ and ‘toward other’.

This figure repeats the analysis shown in Figure 2, but uses an angular definition of 45° around the direction of the visual stimulus to define the ‘toward cued’ and ‘toward other’ directions, instead of the original 90°. (A) Time courses of saccade rates for saccades towards the cued and the other item. This analysis only includes downward saccades, as stimuli were positioned at a 45° angle below fixation, allowing us to distinguish saccades towards the item from those returning to central fixation. The time courses indicate mean values, with shaded areas indicating the standard error of the mean. (B) Time courses of the attentional modulation of saccade direction (toward cued – toward other). The time courses indicate mean values, with shaded areas indicating the standard error of the mean. Black horizontal lines indicate significant temporal clusters (as determined by a cluster-based permutation test). Throughout the entire figure, the following significance levels were used: *: p< 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.