Figure 1
Experimental protocol and stimulus design. a Participants were required to pass an initial period of 200 ms of fixation before we displayed eye movement instructions for 400–500 ms: Presentation of a single black dot indicated a fixation trial (in which saccades were labelled as ‘unintended’) and presentation of a black and white dot indicated a microsaccade trial (labelled ‘intended’). In the following, the stimulus was presented together with a clock for 1000 ms. To avoid sudden visual transients, (only) we modulated stimulus contrast with an initial fade-in and final fade-out period (200 ms each). Clock hands moved at a constant speed of 180°/s. After a short blank period of 50 ms, the clock reappeared, and participants were required to adjust the clock hands via rotation knob until they matched their position when the stimulus was visible with pressing the spacebar indicating stimulus absence. Procedure in Experiment 2 was identical without instruction interval. b–d Stimulus display across conditions: If an eye movement was generated in the same direction (b) and with comparable peak velocity (c, left panel, blue curve) as the grating’s rapid temporal phase shift (c, left panel, gray curve), the retinal velocity of the stimulus came close to 0, rendering the stimulus visible for a short period of time (c, left panel, tvisible of black curve). An aperture shift in the opposite direction of the phase shift (d) and with the same velocity profile of a microsaccade (c, right panel, yellow curve) similarly rendered the stimulus visible as a short distinct impression of a grating (c, right panel, tvisible of black curve). e Predictions about effect binding in different conditions: Intended microsaccades should lead to a stronger predating compared to a replay of their retinal consequences (relative to the objective time of stimulus visibility tvisible), with 0 milliseconds indicating saccade onset. Temporal estimates should not differ between stimulus conditions for unintended or spontaneous microsaccades.
Figure 2
Accurate control of and high visual sensitivity for microsaccades. a Proportion of trials of different types of microsaccades and, for intended microsaccades, across target distances ranging from 0.2 to 1 dva. Empty circles represent the average rates for each participant and condition, arranged from lowest to highest values. Squares indicate the group means, and error bars show the 95% confidence intervals. b Amplitudes of different types of microsaccades and, for intended microsaccades, across target distances. Empty dots represent the average amplitudes for each participant and target distance, while violin plots illustrate the distribution of all saccade amplitudes. c Visual sensitivity as a function of microsaccade generation, across different stimulus conditions (active vs replay) and eye movement types (intended, unintended, and spontaneous). d Visual sensitivity as a function of retinal velocity, with low-velocity trials defined by the saccade’s peak velocity being within ±30 dva/s of the grating’s velocity, and high-velocity trials where the saccade’s peak velocity falls outside this range.
Figure 3
Absence of a conscious feeling of agency for intended, unintended, and spontaneous microsaccades. a The upper and lower rows show the stacked density of time intervals (report–event) between the time of the highest retinal velocity of the stimulus (event) and the reported time of the stimulus (report) in the active (blue) and replay (yellow) conditions for different eye movement types. Different shades represent data from individual participants. The box plots in the middle panel depict the median, interquartile range, and overall spread for each condition and eye movement type, with outliers omitted for clarity. Diamond shapes indicate distribution means for individual participants (with colors corresponding between diamond and density per participant), while vertical lines connect values for each participant to illustrate the direction of the effect. b This panel presents the same measures as in a, but for intended eye movements only with separate target distances (ranging from 0.2–1.0 dva) in subpanels. All other aspects are the same as in a.