Fig. 2: Evidence integration upon probe presentation decreases as a function of prior uncertainty.

a Schematic of drift-diffusion model. Following visual encoding, evidence is successively accumulated towards either of two bounds when probed for the dominant prevalence of one of two options of a single feature (e.g., color). A button press indicates the decision once one of the bounds has been reached and motor preparation has concluded. A non-decision time parameter captures visual encoding and motor preparation, drift rate captures the amount of available information, and boundary separation captures response bias (i.e., conservative vs. liberal). b Behavioral parameter estimates for drift rate and non-decision time (NDT; discussed in Supplementary Text 4), as indicated by the hierarchical drift-diffusion model (HDDM). Data are within-subject centered for visualization (see “Methods”; drift: ***a: p = 2.9e−16, ***b: p = 6.5e−6, **c: p = 0.008, linear: b = −0.57, 95% CI = [−0.65, −0.49], t(46) = −13.94; NDT: ***a: p = 5.6e−11, ***b: p = 8.2e−12, ***c: p = 0.0001, linear: b = 0.07, 95% CI = [0.06, 0.08], t(46) = 23.27). c Modulation of the centro-parietal positivity (CPP) as a neural signature of evidence accumulation (mean ± within-subject SEM). The response-locked CPP indicates decreases in pre-response integration rate with increasing probe uncertainty. Insets show CPP slope estimates from −250 to −100 ms relative to response execution (***p = 7.2e−8, linear: b = −2e−6, 95% CI = [−2.8e−6, −1.5e−6], t(46) = −7.05), as well as the corresponding topography (CPP channel shown in yellow). Panels b and c indicate p values from two-tailed paired t-tests, pairwise comparisons were Benjamini–Hochberg-adjusted for multiple comparisons. n = 47 participants for all panels. Source data are provided as a Source Data file.