Eyes on the past: Gaze stability differs between temporal expectation and temporal attention

Abstract

Temporal expectation and temporal attention distinctly improve performance and gaze stability, and interact at the behavioral and neural levels. Foreperiod-the interval between the preparatory signal and stimulus onset-facilitates temporal expectation. Preceding foreperiod-the foreperiod in the previous trial-modulates expectation at behavioral and oculomotor levels. Here, we investigated whether preceding foreperiod guides temporal attention. Regardless of the preceding foreperiod, temporal attention improved performance, particularly at early moments,and consistently accelerated gaze stability onset and offset by shifting microsaccade timing. However, only with preceding expected foreperiods, attention inhibited microsaccade rates. Moreover, preceding late foreperiods weakened expectation effects on microsaccade rates, but such a weakening was overridden by attention. Altogether, these findings reveal that the oculomotor system's flexibility does not translate to performance, and suggest that although selection history can be utilized as one of the sources of expectation in subsequent trials, it does not necessarily determine, strengthen, or guide attentional deployment.

Keywords: microsaccades; sequential effects; temporal attention; temporal expectation.

Figure 1.

A. Psychophysical procedure assessed visual performance at attended and unattended moments, with a variable stimulus timing. The exact timing of the stimulus onset varied across sessions, and we analyzed the data from the sessions with highest variance, “Uniform” and “Wide” temporal distributions. Expected moment is the time point when the stimuli could appear with highest probability (mean point of the temporal distributions). Stimuli could appear earlier or later than the expected time. For microsaccades, we analyzed the “prestimulus window”: the interval between the precue and the earliest possible stimulus onset. Foreperiod is defined as the interval between the precue and T1, and the preceding foreperiod is this interval in the most recent trial. B. Summary of experimental variables. There were two targets in each trial, and performance was tested separately for T1 and T2 indicated by the response cue. The precue was either valid or neutral to probe temporal attention. Temporal expectation was manipulated via target timing. Preceding trial’s target onset corresponds to preceding foreperiod, and the target onset corresponds to the current foreperiod.

Figure 2.

A. The effects of target (T1, T2), cue validity (Valid, Neutral), and current trial’s onset timing (Early_N, Expected_N, Late_N) on sensitivity (d’). B. The effect of cue validity and current trial’s onset timing on reaction times (RT). Data from Duyar et al., 2024.

Figure 3.

Effects of the attentional precue on microsaccade rates in the prestimulus window. In all graphs, behavioral accuracy increased and RT decreased with temporal attention. Error bars represent standard error of the mean. A. Effects of the attentional precue regardless of preceding foreperiod. Microsaccade rates in T1-cued trials were significantly lower than in neutral trials between 713–1200 ms. B. Effects of the attentional precue on microsaccade rates within each preceding foreperiod condition. In the Expected_N-1 condition, neutral trials showed significant differences in microsaccade rates compared to T1-cued trials between 735 and 1200 ms and T2-cued trials between 913 and 1200 ms.

Figure 4.

In all graphs, behavioral accuracy and RT plotted as a function of preceding foreperiod (no significant effect). Error bars represent standard error of the mean. A. The effect of the preceding foreperiods (Early_N-1, Expected_N-1, and Late_N-1) on microsaccade rates regardless of attentional precue. Early_N-1 and Late_N-1 conditions differed between 890 and 1090 ms. B.The effect of the preceding foreperiods on microsaccade rates within each attentional precue (Neutral, T1, and T2) condition. In the Neutral condition, a significant difference was observed in the 790–1099 ms interval, highlighting when Early_N-1 and Expected_N-1 conditions diverge. No significant effects were found in either T1 or T2 precue conditions.

Figure 5.

Microsaccades timing around stimulus onset. A. Inhibition of microsaccades was analyzed within the prestimulus window (from precue onset to 1200 ms). There was a main effect of precue, but no effect of preceding foreperiod. B. Inhibition latency shown separately for preceding foreperiod. The effect of precue on microsaccade latency was consistent across preceding foreperiod. C. Microsaccade rebound after the stimulus onset revealed a main effect of precue, and no effect of preceding foreperiod. D. The effect of precue on rebound latency was consistent across preceding foreperiod.