Rule-based cues reliably modulated pupil-linked arousal.

A Pupil diameter on correct trials, aligned to choice and plotted as a function of the congruence between the choice and the prior. Lines and error bars are means and SEM across subjects. B Regression coefficients (beta values) from a linear mixed-effects model as a function of different pupil contrasts (see labels above each panel). Error bars are 95% confidence intervals of the parameter estimates. C Across-subject Spearman correlation between: 1) the degree to which each subject was biased by the prior cues, estimated from a logistic regression model of choice behavior; and 2) each subject’s pupil contrast from a linear regression model. D, E Same as A, B for incorrect trials. The time window in D and E was chosen to focus on the period in which the incongruent–congruent effect was most prominent in the correct trials (and note that incorrect trials ended earlier than correct trials; see Methods). Values in B, C, and E were computed in 20 ms-wide time bins. Green lines indicate time bins for which pcorrected < 0.05, FDR-corrected across time and contrast.