Nora Bradford¹, Brian Maniscalco¹, Megan A. K. Peters¹; ¹University of California, Irvine

Research on visual metacognitive judgments has shown that stronger evidence supporting one’s choice usually leads to higher confidence judgments even in the presence of stronger evidence opposing the choice, dubbed a ‘positive evidence bias’. Thus far, this bias has typically been demonstrated in visual tasks which manipulate the objective (physical) amount of evidence available to the observer. To investigate whether and how the positive evidence bias translates to perceived (illusory) positive evidence, we presented a version of the convex hull numerosity illusion, amplified by also manipulating total surface area, in a 2AFC task accompanied by post-decisional confidence ratings to 50 healthy human participants. On each trial, observers viewed two patches of 11-32 dots, where the ratios of the two patches’ convex hulls ranged from 1:1 to 1:1.4. The ratios of total surface areas occupied by the dots in both patches covaried with the convex hull ratios. This predictably affected the perceived relative numerosity of the patches. We categorized trials as ‘congruent’ if the larger-convex hull patch was indeed the more numerous patch, and ‘incongruent’ otherwise. Participants’ visual discrimination performance on congruent trials was higher than on incongruent trials, confirming that larger convex hulls did lead to higher perceived numerosity. Confidence covaried with perceived numerosity for congruent trials, but systematically deviated from this pattern in incongruent trials. These findings provide preliminary evidence that illusory shifts in perceived numerosity can affect type 1 judgments and type 2 judgments, but may do so in different ways. Our results suggest a powerful way forward for separating out the computations underlying visual type 1 performance and subjective confidence.