Twelve participants (8 female), aged between 20 and 35 years old (mean = 25, SD = 4.2), were included in the data analysis of Experiment 2. Two participants were excluded from the analysis based on floor and ceiling effects (see Experiment 2—procedure below). All participants were free from neurological and visual impairments. Experiment 2 was conducted with the approval of relevant ethics committees at the Medical Faculty of Tuebingen University and of the University of Bielefeld. Subjects provided written, informed consent, and participated either as unpaid volunteers or for financial compensation (€10 per testing session).
Sample size considerations
Experiment 2 was conducted during COVID restrictions in 2020–21, which made recruitment and testing opportunities extremely limited. We aimed to test equivalent participant numbers to Experiment 1, and managed to test 14 participants between two laboratories, though only 12 participants remained after exclusions. This was the maximum achievable sample size under the conditions that existed at the time.
Experiment 2—apparatus and stimuli
Experiment 2 was similar to that for Experiment 1 but run in a different laboratory (Physiology of Active Vision Laboratory, Werner Reichardt Centre for Integrative Neuroscience, University of Tuebingen), with minor differences in setup.Footnote 1 The monitor was a 19-inch CRT monitor with a refresh rate of 85 Hz (11.8 ms temporal resolution) positioned at a viewing distance of 57 cm. Stimuli were presented on a grey background (42.2 cd/m2). The fixation point was 0.01 degrees radius (97.3 cd/m2), and the cue (74.9 cd/m2) was presented either 1 degrees above the fixation point or at an eccentricity of 10 degrees from the fixation point and 1 degrees above the centre of the possible probe location on that side. All other stimulus parameters were identical to those in Experiment 1. Eye movements were monitored with a desktop mount EyeLink 1000 system, tracking the right eye at a sampling rate of 1000 Hz. Manual responses were recorded with two (left or right) of the five-button on a standard response pad (RESPONSEPixx, VPixx Technologies Inc.). Participants completed one testing session with one practice block and 16 blocks of experimental trials.
Experiment 2 procedure was similar to Experiment 1 except for a few critical modifications. After calibration, the experiment started with a practice block of 64 trials and no QUEST procedure was required. Instead, we used a probe stimulus with an orientation of either 84, 85, 86 degrees (right tilt) or 94, 95, 96 degrees (left tilt) chosen at random at the start of each trial. After a random interval between 800 and 1200 ms, the cue was presented for 35.2 ms either 10 degrees to the left or right of fixation or at the centre of the screen one degree above of the fixation point. The probe was presented for 90 ms after one of four randomly selected CTOAs: 47, 82, 106 and 200 ms, with equal numbers of trials at the cued location (congruent cue condition) or at the neutral location (neutral cue condition). The neutral cue condition was introduced as a control for any generalised temporal “warning effect” of the congruent cue (Coull and Nobre 1998; Hackley and Valle-Inclán 2003). On half of the trials, a black flash covering the bottom and top thirds of the screen for 11.8 ms was onset simultaneously with the probe.
Participants were asked to complete 16 blocks of experimental trials within the 75 min testing session. Within each block, there were 64 valid trials, one for each combination of cue type (congruent, incongruent) by flash (no flash, flash) by probe side (left, right) by CTOA (47, 82, 106 and 200 ms) by probe tilt (left, right). The first three factors (cue, flash, CTOA) were of theoretical interest, whilst the latter two factors (probe side and tilt) were not. Each participant completed 64 trials per CTOA for each combination of cue type (congruent, neutral) by flash (no flash, flash), except that some of the participants did not finish the last block due to time constrains (across subjects a total of 182 trials were not completed). Two of an original 14 participants were excluded from the analysis because their overall mean performance was either at floor (48.6%) or at ceiling (93.9%).
Experiment 2—data processing and analysis
We collected a total of 12,112 trials across all included participants in Experiment 2. We used identical criteria to Experiment 1 for trial exclusion. We rejected 1% of trials because of “bad data”, 0.17% trials with manual reaction times less than 200 ms and 1% of trials with manual reaction times more than 3.5 standard deviation above the participant’s average. We also removed 2.3% of trials in which critical stimulus flips were not properly timed with the monitor refresh rate. Our sample was of 11,578 trials, from which we removed 1699 trials because of microsaccades (14.7%), leaving 9879 of microsaccade free trials in an interval between − 200 ms before cue onset and 200 ms after probe onset (minimum saccade free interval equal to 250 ms and maximum saccade free interval equal to 400 ms). From the total of the excluded microsaccades, 50% belonged to the congruent and 50% in the incongruent cue condition. Similarly, 52.6% microsaccades were detected in no-flash trials and 47.3% in flash trials. We ran a mixed-effects logistic regression to test the influence of Cue type (congruent, neutral), Flash (no flash, flash), and CTOA (47, 82, 105 and 200 ms) on perceptual performance. Paired-sample t-test was used to follow up statistical analysis with an alpha level of 0.05. Microsaccade analysis was identical to Experiment 1.
All data pre-processing and statistical analyses follow the procedure described in Experiment 1 and the entire dataset is uploaded in the Open Science Framework (see Experiment 1—data processing and analysis).
In Experiment 2, we tested whether the inhibitory effect of the flash could occur within the range relevant to early orienting or if it was only effective in the later (reorienting) range. To do so, we tested orientation discrimination for probes presented at congruent cue locations in a time window between 50 and 200 ms. To control for possible effects driven only by the temporal aspects of the cue, rather than its location in space, we also tested a neutral cue condition in which a spatially non-informative cue was presented centrally. Based on the results of Experiment 1, we expected to see modulations of discrimination performance at the time of reorienting, when attention moves away from the cued location, and/or at early CTOAs, when facilitation is expected. To explore these patterns, we ran a mixed-effects logistic regression with Cue type (congruent, neutral), Flash (no flash, flash), CTOA (47 ms, 82 ms, 106 ms, 200 ms), and their interactions (same model as Eq. 1).
The first significant effect was an interaction between Flash and CTOA (Fig. 5A). In flash trials, there was an enhancement of about 4% in discrimination performance when the probe was presented 47 ms after the cue compared to later CTOAs [CTOA 82 ms: β = − 0.036, 95% CI = (− 0.32, 0.247), t = − 0.252, p = 0.801; CTOA 106 ms: β = − 0.287, 95% CI = (− 0.566, − 0.008), t = − 2.017, p = 0.044; COTA 200 ms: β = − 0.242, 95% CI = (− 0.528, 0.042), t = − 1.67, p = 0.095]. Since the interaction was irrespective of cue type, i.e. congruent or neutral, it suggests that the flash was having a general “warning effect”, increasing visual sensitivity when presented immediately after the cue (Coull and Nobre 1998; Hackley and Valle-Inclán 2003).
We also found an interaction between Cue and CTOA (Fig. 5B). While performance in the neutral cue condition was stable at about 80% (grey line), the congruent cue condition (red line) was modulated across CTOAs with a decrease in discrimination performance of about 4% taking place 200 ms after cue onset ms [β = 0.298, 95% CI = (0.013, 0.583), t = 2.049, p = 0.041]. Numerically, perceptual performance was slightly better at CTOA 82 ms, but overall we did not see a significant facilitation effect at any of the early CTOAs. This result suggested that in our paradigm, the attentional shift triggered by the cue at early CTOAs was not strong enough to boost performance significantly for this task. On the other hand, attentional allocation started reducing at cue location at around 200 ms after cue onset, compatible with previous reports (Cheal and Chastain 1999; Cheal et al. 1998; Posner and Cohen 1984; Pratt and Abrams 1999) and our results from Experiment 1.
The significant decrease in discrimination performance observed in the congruent cue condition at 200 ms after cue onset (Fig. 5B) was suggestive of attention moving away from the cue location (reorienting). This result motivated us to explore in more detail if the flash could interfere with this part of the orienting process, as observed for the incongruent cue condition in Experiment 1. To do so, we computed a measure of cue-benefit by subtracting accuracy in the neutral cue condition from that in the congruent cue condition in both no-flash and flash trials at each CTOA. Paired-sample t-test showed that only at 200 ms was there a difference between flash and no-flash conditions [t (11) = − 2.94, p = 0.013], with discrimination performance being about 6% higher in flash trials (Fig. 5C). This result supports the hypothesis that the flash has a temporal window of action in which it can inhibit attention moving away from cue location, similar to the result reported in Experiment 1.
Experiment 2—microsaccade analysis
We detected a total of 3454 microsaccades over 3133 trials in a time period of 200 ms before cue onset to 500 ms after probe onset (1.1 microsaccades per trial within this time period). Similar to Experiment 1, microsaccade rate (black line) was completely disrupted about 100 ms after cue onset (Fig. 6A). When data were aligned to probe onset (Fig. 6B) and microsaccade rate was split by flash condition (green: no flash, blue: flash), despite microsaccadic rate being close to zero for both conditions, there is indication of a slightly larger inhibition for flash trials. As in Experiment 1, microsaccadic inhibition after the probe was long-lasting, with the rebound period starting only at around 300 ms after probe onset. This result confirms that our stimuli could abolish any subsequent eye movement in a time period compatible with the covert effect recorded in absence of eye movement.