Fig. 3: Trial-by-trial stimulus-dependent repulsive bias.

a Probability of categorizing stimulus n as “strong” as a function of trial n Δspeed, with curves grouped by Δspeed of trial n − 1. Darker curves correspond to higher n − 1 Δspeed. Blue squares denote PSE. b Trial n psychometric curves, grouped by trial n − 1 choice, “strong” (red) and “weak” (blue). Transparent shading represents 95% confidence interval of bootstrapped probabilities. c Probability of categorizing stimulus n as “strong” as a function of the Δspeed of trial n − 1. All Δspeed values of trial n are pooled. The green line corresponds to n − 1 correct trials, the violet line to n − 1 incorrect trials. Transparent shading represents 95% confidence interval of bootstrapped probabilities. d Bias of the trial n psychometric curve, depending on Δspeed in trial n − 1 (far left plot) to trial n − 4 (far right plot). Shading inside the squares denotes Δspeed in trial n − 1. Squares correspond to individual rats. e Slope of the regressions shown in panel d, in the eight preceding trials. White denotes slopes different from 0 (p < 0.05, two-tailed t-test). Transparent shading represents standard deviation between rats. f Coefficient values (β) of the GLM including stimuli of up to 10 preceding trials as predictors. White denotes coefficients different from 0 (p < 0.05, two-tailed t-test). Discontinuous ordinate shows large positive coefficient associated with stimulus n. Transparent shading represents standard error of the mean. Source data are provided as a Source Data file.