Abstract
Vision is damaged by uncertainty: to make sense of the physical properties of the world the visual system needs to deal with missing information and noisy sensory mechanisms. Although we have some sensitivity to our level of uncertainty, these judgements can be systematically biased yielding over or under-confidence. Here we investigated whether observers can correctly estimate the loss of perceptual precision that characterizes peripheral vision. Fifteen naïve observers were presented with one oriented Gabor stimulus at 30 dva eccentricity and one in the centre, in random order. Observers had to categorise whether the Gabor was rotated clockwise or counter-clockwise with respect to a fixed reference, separately for the stimulus in the periphery and the one in the centre. Then, they had to choose which of the two perceptual judgments they were more confident to be correct. The rotation was fixed at 5 degrees, but the contrast of the Gabor was adaptively varied to keep both peripheral and central performance within a similar range. In order to compare peripheral and foveal confidence independently of performance, we first computed the difference in performance between the peripheral and the central judgements as a function of central and peripheral contrast. Then, we computed an index of confidence based on the agreement between the difference in performance and confidence judgements, for a given central and peripheral contrast. All the participants but one exhibited under-confidence in the periphery: their performance was higher than they thought (p=0.021 -Wilcoxon signed rank test). Although observers’ judgements were biased, they still reflected some knowledge about the level of perceptual uncertainty. In fact, when observers were more confident in either the central or the peripheral judgments, the corresponding accuracies were higher (t(14)= 2.7327,p=0.016; t(14)=7.7, p < 0.001; for central and peripheral judgments, respectively).