Increased preparation time reduces, but does not abolish, action history bias of saccadic eye movements

Abstract

The characteristics of movements are strongly history-dependent. Marinovic et al. (Marinovic W, Poh E, de Rugy A, Carroll TJ. eLife 6: e26713, 2017) showed that past experience influences the execution of limb movements through a combination of temporally stable processes that are strictly use dependent and dynamically evolving and context-dependent processes that reflect prediction of future actions. Here we tested the basis of history-dependent biases for multiple spatiotemporal features of saccadic eye movements under two preparation time conditions (long and short). Twenty people performed saccades to visual targets. To prompt context-specific expectations of most likely target locations, 1 of 12 potential target locations was specified on ~85% of the trials and each remaining target was presented on ~1% trials. In long preparation trials participants were shown the location of the next target 1 s before its presentation onset, whereas in short preparation trials each target was first specified as the cue to move. Saccade reaction times and direction were biased by recent saccade history but according to distinct spatial tuning profiles. Biases were purely expectation related for saccadic reaction times, which increased linearly as the distance from the repeated target location increased when preparation time was short but were similar to all targets when preparation time was long. By contrast, the directions of saccades were biased toward the repeated target in both preparation time conditions, although to a lesser extent when the target location was precued (long preparation). The results suggest that saccade history affects saccade dynamics via both use- and expectation-dependent mechanisms and that movement history has dissociable effects on reaction time and saccadic direction. NEW & NOTEWORTHY The characteristics of our movements are influenced not only by concurrent sensory inputs but also by how we have moved in the past. For limb movements, history effects involve both use-dependent processes due strictly to movement repetition and processes that reflect prediction of future actions. Here we show that saccade history also affects saccade dynamics via use- and expectation-dependent mechanisms but that movement history has dissociable effects on saccade reaction time and direction.

Keywords: global effect; oculomotor capture; reaction time; use-dependent plasticity.

Fig. 1.

Illustration of the trial structure and target positions. A: timeline of a single repeated target trial for conditions with long preparation time (left) and with short preparation time (right). Note that in the long preparation time condition the target is shown in advance (dotted rectangle highlights the crucial differences between conditions). B: layout of the target positions. The repeated target position was located at the 90° position, and the probe target positions were located between 30° and 330° (equally spaced with 30° angles between target positions). Note that illustrations are not drawn to scale.

Fig. 2.

From left to right, examples of individual raw saccade trajectories at baseline trials and for probe target trials (Experiment) in the long and short preparation time conditions and for trials that were classified as selection errors.

Fig. 3.

Probe target saccadic reaction time histograms of individual subjects for long preparation trials, short preparation time trials, and trials that were excluded from the analysis. Vertical dashed lines indicate the reaction time cutoff at 70 ms.

Fig. 4.

Baseline and experimental performance per target angle and preparation time conditions. A: saccadic reaction times. B: saccade amplitude. C: saccadic peak velocity. Error bars are omitted for clarity. dva, Degrees of visual angle.

Fig. 5.

Baseline and experimental performance per target angle and preparation time conditions. A: angular errors at half peak velocity. B: angular errors at peak velocity. C: errors at saccade end point. Error bars are omitted for clarity. The dashed horizontal line illustrates the zero value in each plot, to emphasize the valence of data in each condition.

Fig. 6.

Change in saccade characteristics as a function of angular distance from the repeated target for probe targets in the short preparation time and long preparation time conditions and the arithmetic differences between long and short preparation time conditions. A: saccadic reaction time. B: saccadic peak velocity. C: saccadic amplitude. Error bars represent within-subject SE (Morey 2008). The dashed horizontal line illustrates the zero value in each plot, to emphasize the valence of data in each condition. dva, Degrees of visual angle.

Fig. 7.

Angular errors at half peak velocity, peak velocity, and saccade end point as a function of target distance from the repeated target in the long preparation time (left) and short preparation time (center) conditions and the difference between short and long preparation time conditions (right). Error bars represent within-subject SE (Morey 2008). The dashed horizontal line illustrates the zero value in each plot, to emphasize the valence of data in each condition.

Fig. 8.

Angular errors at half peak velocity as a function of target distance from the repeated target in the long preparation time and short preparation time conditions at the first half (left) and the second half (right) of trials. Error bars represent within-subject SE (Morey 2008). The dashed horizontal line illustrates the zero value in each plot, to emphasize the valence of data in each condition.