Attention cueing in rivalry: insights from pupillometry

Abstract

We used pupillometry to evaluate the effects of attention cueing on perceptual bi-stability, as reported by adult human observers. Perceptual alternations and pupil diameter were measured during two forms of rivalry, generated by presenting a white and a black disk to the two eyes (binocular rivalry) or splitting the disks between eyes (interocular grouping rivalry). In line with previous studies, we found that subtle pupil size modulations (about 0.05 mm) tracked alternations between exclusive dominance phases of the black or white disk. These pupil responses were larger for perceptually stronger stimuli: presented to the dominant eye or with physically higher luminance contrast. However, cueing of endogenous attention to one of the rivaling percepts did not affect pupil modulations during exclusive dominance phases. This was observed despite the reliable effects of endogenous attention on perceptual dominance, which shifted in favor of the cued percept by about 10%. The results were comparable for binocular and interocular grouping rivalry. Cueing only had a marginal modulatory effect on pupil size during mixed percepts in binocular rivalry. This may suggest that, rather than acting by modulating perceptual strength, endogenous attention primarily acts during periods of unresolved competition, which is compatible with attention being automatically directed to the rivaling stimuli during periods of exclusive dominance and thereby sustaining perceptual alternations.Significance StatementBinocular rivalry depends on attention. When it is diverted away from the stimuli, perceptual alternations slow down; when it is preferentially directed to one stimulus, perception lingers more on it, consistent with attention enhancing the effective strength of the rivaling stimuli. Here we introduce pupillometry as a means to indirectly track changes in effective stimulus strength. We find that pupil size accurately tracks perceived luminance during two forms of rivalry: binocular rivalry and interocular grouping rivalry. Both show robust effects of attention cueing on perceptual dominance, but pupil modulations during exclusive dominance are unaffected by cueing. This suggests that endogenous attention does not affect perceptual strength during exclusive dominance, though it might do so during transition phases.

Keywords: attention; binocular rivalry; interocular grouping; pupillary light response; pupillometry; visual awareness.

Figure 1.

Dichoptic stimulation and rivalry dynamics. A, Schematics of the stimuli (white or black patches overlayed with orthogonal thin lines), presented dichoptically through a four-mirror stereoscope. B, Schematic representation of the possible stimulus configurations (thin lines omitted) and perceptual outcomes for binocular rivalry (BINriv) and interocular grouping rivalry (IOGriv). C, Example traces from a segment of the experiment, where participants used keypresses to report the dominant percept (square wave) and we recorded pupil size modulations (blue wave). D, Probability density function of the normalized phase durations for exclusive dominance of white or black disk percepts in binocular rivalry and interocular grouping rivalry.

Figure 2.

Pupil modulations track perceptual alternations. Baseline subtracted pupil size traces aligned to perceptual switches toward exclusive dominance of a white disk or a black disk percept and averaged across phases, separately for binocular rivalry (A) and interocular grouping rivalry (B). In all panels, shadings report mean ± 1 SE across participants and the blue marks on the x-axis highlight time points where pairwise comparisons between traces are significant (one tailed t test, p < 0.05 FDR corrected). Observations regarding the latency of the pupillary response and its relative magnitude are reported in Extended Data Figure 2-1, where nonbaseline subtracted traces are shown.

Figure 3.

Attention cueing affect perceptual alternations but not pupil modulations. A, B, Perceptual dominance for exclusive white or black disk percepts, without attentional cueing (dashed lines) or when the white or the black disk percept were cued (continuous lines, cueing condition indicated on the abscissa). Error bars report ±1 SE across participants. C, D, Baseline corrected average pupil size computed in a fixed temporal window (between –0.5 and 1 s from the perceptual transition) during phases of exclusive dominance of the black (black line) and the white disk (red line). Results from the no-cueing condition are reported by dashed lines. Continuous lines report the results from the trials where the white or the black percepts were cued (separated on the abscissa). Error bars report ±1 SE across participants. E, F, Individual participants’ attentional modulation indices for perceptual dominance (x-axis) and pupil size (y-axis), computed with Equations 3, 4. Dash-dot blue lines mark the x = 0 and y = 0 lines, indicating no effect of attention cueing. Each circle reports results from one participant; red dots highlight participants with a significant attentional modulation index for perceptual dominance. Red lines show the best fitting line and its 95% confidence intervals. In all panels, the left column reports results for binocular rivalry and the right for interocular grouping rivalry.

Figure 4.

Pupil time courses are comparable across cueing conditions. A, B, Pupil size traces aligned to perceptual switches toward exclusive dominance of a white disk or a black disk percept computed across phases in individual participants and then averaged for each cueing condition, separately for binocular rivalry and interocular grouping rivalry. C, D, Time course of the difference between baseline corrected pupil size during black and white percepts, computed in individual participants and then averaged for each cueing condition. The resulting traces show no effect of cueing. The same conclusions can be drawn skipping the baseline correction step or defining pupil baseline over a wider temporal interval around perceptual switch, as shown in Extended Data Figure 4-1. In all panels, shadings report mean ±1 SE across participants.

Figure 5.

Effects of attention cueing versus enhancing contrast. A, B, Perceptual dominance for exclusive white or black disk percepts, in the no cue condition (dashed lines) or when the physical contrast of the white disk was enhanced/cued (continuous lines, contrast, or cueing condition indicated on the abscissa). Error bars report ±1 SE across participants. C, Time course of the difference between baseline corrected pupil size during black and white percepts, computed in individual participants and then averaged for each condition. The blue marks on the x-axis highlight time points where pairwise comparisons between the +100% and the no cue condition traces are significant (p < 0.05 FDR corrected). Shadings report mean ±1 SE across participants.

Figure 6.

A,B, Proportion of mixed percepts in no cueing or cueing conditions (collapsed across white and black cued) for binocular rivalry (BINriv) and interocular grouping rivalry (IOGriv). Error bars report ±1 s.e. across participants. ns: not significant. C,D, Pupil traces during mixed percepts in the three cueing conditions for binocular rivalry (BINriv) and interocular grouping rivalry (IOGriv). Shaded areas show ±1 s.e. across participants and blue marks on the x-axis highlight timepoints where pairwise comparisons between the white and black cueing conditions are significant (p < 0.05 FDR corrected).