Figure 4. A) Baseline-corrected pupil response over time after tactile stimulation, expressed in arbitrary units (a.u.), plotted per stimulation intensity. Positive values indicate pupil dilation, negative values indicate pupil constriction. Error bands indicate one standard error above and below the mean. B) Pupil response derivative traces over time, averaged per stimulation intensity. Positive values indicate the change in the amount of pupil size increase, negative values the change in the amount of pupil size decrease compared to the previous time point. Error bands indicate one standard error above and below the mean. C) Linear mixed effects model for pupil response comparing t -values between stimulation intensities over time in seconds. Each line represents the t -values of the comparisons between stimulation intensities on pupil response after tactile stimulation over time, with an additive effect of trial number and random intercepts for each participant. The dotted line representst  = |1.96|, corresponding to = 0.05.D) Time to maximum pupil response in seconds, averaged per participant and split between stimulation intensities.
Participants performed above chance level in the forced-choice task discriminating between the three stimulus intensities (low vs. medium:W  = 0, p < 0.001, r = 0.51; low vs. high:W  = 1, p  < 0.001, r = 0.92; medium vs. high: t (18) = 5.22, p  < 0.001, d = 3.92; see Supplementary Figure 8A for performance per stimulus intensity pair). Performance differed between the three stimulus intensity pairs,χ2 (38) = 26.63, p  < 0.001. Participants performed worse in discriminating the medium vs. high intensities as compared to the low vs. high intensities ( = 0, = 0.002, r = -0.82), and as compared to the low vs. medium intensities (W  = 0, p  = 0.001, r = -0.85), both with large effect sizes. No differences were found between the low vs. medium intensities as compared to the low vs. high intensities (W  = 6, p  = 0.655, r = -0.10). Thus, participants were better able to discriminate the low intensity from the medium and high intensities; whereas the medium and high intensities were perceived as being more similar.
Participants differed in the certainty of their responses in the stimulus intensity discrimination task, χ2(38) = 25.62, < 0.001 (see Supplementary Figure 8B for the certainty scores per stimulus intensity pair). Participants were less confident discriminating the medium vs. high intensities as compared to the low vs. high intensities (W  = 5, p < 0.001, = 0.83), and as compared to the low vs. medium intensities (W  = 1, p< 0.001, r  = 0.87), both with large effect sizes. There were no differences between the low and medium intensities as compared to the low and high intensities (W  = 44, p  = 0.124,r­  = 0.38). Thus, the certainty scores were consistent with the accuracy in discriminating the different stimulus intensities.
As an exploratory analysis, we examined whether the ability to discriminate between the medium and high stimulus intensities, and the certainty of this judgement, were correlated with the difference in pupil size response to tactile stimulation at medium and high stimulus intensities. Only the effect for the medium vs. high stimulus intensity was evaluated, as participants showed the most variation in their discriminative performance for this stimulus pair.
We calculated difference scores for the discriminability and certainty scores for tactile stimulation of the medium vs. high stimulus intensity and for pupil responses over time. These difference scores were correlated using Spearman correlations. No consistent correlations between differences were observed over time (see Supplementary Figure 9 for details).