Statistical analyses
The four stimulus intensities (i.e., none, low, medium, high) were included as predictors of pupil size in a linear mixed effects model. The trial number was also included to control for possible habituation effects.
As in Experiment 1, the time to the maximum pupil size derivative was compared between the four stimulus intensities.
To assess whether participants could discriminate the different stimulus intensities within each pair, we tested whether the number of correct responses was above chance level (i.e., 5 out of 10 correct responses) using parametric one-sample t -tests or non-parametric Wilcoxon signed rank tests. Next, we compared the performance (i.e., number of correct responses) and participants’ confidence in their responses between the three pairs of stimulus intensities using a non-parametric Friedman’s ANOVA and post-hoc Wilcoxon signed-rank tests. One participant had no data available on the discrimination task and was excluded from these analyses.