Fig. 3
From: Suboptimal but intact integration of Bayesian components during perceptual decision-making in autism
Sensitivity, decision boundary, and optimal observer analyses for Experiment 1, prior manipulation. (a) Sensitivity (d’) for each group as a function of contrast and across base rates. Note that in all experiments, the relatively low sensitivity in both groups, even when contrast is high, is due to the limit of a maximum of 80% correct in these tasks. (b, c) Decision criterion (c) as a function of contrast for the three base rate blocks for the autistic and non-autistic groups. The base rate legend gives the probability for category B to appear. (d) Difference between criterion shifts in biased (25% and 75%) base rate blocks (Dcriterion) for each group as a function of contrast on a log scale. (e) Deviation of criterion shift from optimality (cerror) as a function of contrast. Participants showed an increase in deviation from an optimal criteria adjustment as contrast decreased, with no difference between autistic and non-autistic groups in the degree to which the criterion was suboptimal. Data points show means across participants and error bars represent ± SE. The asterisks represent the group difference evaluated using unpaired t-tests, *p ≤ .05, **p ≤ .01. The sample size constituted 31 autistic and 46 non-autistic participants in (a), (b), (c) and (d), and 30 autistic and 45 non-autistic participants in (e)