Fig. 6: Computational results in Experiments II and III.

a Mean sensory precision \(\log ({{\rm{\pi }}}_{{\rm{u}}})\) in the two first context-setting blocks in Experiments II and III, estimated using Bayesian Model Averaging. b, c Correlations between accuracy and \(\log ({{\rm{\pi }}}_{{\rm{u}}})\), both measured in the two first context-setting blocks, in Experiments II (b) and III (c). ***p < 0.0001. d, e Experiment II: Bayesian model selection (BMS) in the NT (d) and ASD (e) groups, comparing M₁ and M₀ in the two first context-setting blocks (12) or third context-setting and time-order effect blocks (345). Protected exceedance probability: **>0.97, ***>0.99. f Estimated precision ratio log(r) in Experiment II. g, h Experiment III: Bayesian model selection in the NT (g) and ASD (h) groups, comparing M₁ and M₀ in the two first context-setting blocks (12) or third context-setting and time-order effect blocks (345). Protected exceedance probability: ***>0.99. i Estimated precision ratio log(r) in Experiment III. Error bars correspond to the standard deviations, except for plots f and i where they represent the standard error of the mean. NT neurotypical, ASD autism spectrum disorders.