Sensory noise increases metacognitive efficiency

Abstract

Visual metacognition is the ability to employ confidence ratings in order to predict the accuracy of one's decisions about visual stimuli. Despite years of research, it is still unclear how visual metacognitive efficiency can be manipulated. Here we show that a hierarchical model of confidence generation makes a counterintuitive prediction: Higher sensory noise should increase metacognitive efficiency. The reason is that sensory noise has a large negative influence on the decision (where it is the only corrupting influence) but a smaller negative influence on confidence (where it is one of two corrupting influences; the other one being metacognitive noise). To test this prediction, we used a perceptual learning paradigm to decrease the amount of sensory noise. In Experiment 1, seven days of training led to significant decrease in noise as well as a corresponding decrease in metacognitive efficiency. Experiment 2 showed the same effect in a brief 97---trial learning for each of two different tasks. Finally, in Experiment 3, we experimentally manipulated stimulus contrast to increase sensory noise and observed a corresponding increase in metacognitive efficiency. Our findings demonstrate the existence of a robust positive relationship between sensory noise and metacognitive efficiency. These results could not be captured by a standard model in which decision and confidence judgments are made based on the same underlying information. Thus, our study provides a novel way to directly manipulate metacognitive efficiency and suggests the existence of metacognitive noise that corrupts confidence but not the perceptual decision.

How experimental procedures influence estimates of metacognitive ability

“…In other words, when easier and harder trials are analyzed together, subjects' metacognitive abilities appear to be better than they actually are. A similar effect was previously reported for the case of meta-d 0 /d 0 and meta-d 0 -d 0 in the context of mixing equal numbers of up to three different contrasts (Bang et al 2019). Here, we show that this result holds across all popular measures of metacognitive ability and occurs even when the contrast mixture is bell-shaped, and itself a product of a staircase (Fig.…”

“…However, since metacognitive noise decreases both d 0 and meta-d 0 , it should have had the opposite effect on measures of metacognitive sensitivity (i.e. meta-d 0 , type 2 AUC, and phi) (Bang et al 2019). Thus, we do not consider metacognitive noise as the main contributor to the metacognitive inflation observed here.…”

“…The notion that mixing different contrasts leads to inflated estimates of metacognitive ability has received recent empirical support. In Experiment 3 of their paper, Bang et al (2019) found that mixing increasingly dissimilar contrast levels led to increasing estimates of metacognitive ability. However, Bang et al only tested the influence of such mixtures on the measures meta-d 0 /d 0 and meta-d 0 -d 0 , so it is possible that the effects were partly driven by the specific assumptions inherent in these measures.…”

“…Metacognitive noise is noise that affects confidence estimates but not perceptual decisions (Mueller and Weidemann 2008;Jang et al 2012;De Martino et al 2013;Maniscalco and Lau 2016;Rahnev et al 2016;van den Berg et al 2017;Shekhar and Rahnev 2018). Bang et al (2019) showed that combining several contrast levels in a single analysis amplifies the external stimulus variability and thus increases the sensory noise measured in the framework of signal detection theory. Furthermore, they demonstrated that increasing sensory noise leads to a decrease in both d 0 and meta-d 0 but that the decrease is larger for d 0 .…”

“…However, Bang et al only tested the influence of such mixtures on the measures meta-d 0 /d 0 and meta-d 0 -d 0 , so it is possible that the effects were partly driven by the specific assumptions inherent in these measures. More importantly, the mixtures in Bang et al (2019) contained equal number of trials coming from each contrast, but staircase procedures result in a bell-shaped curve of contrast prevalence such that intermediate contrasts are presented many times and very low and high contrasts are presented fewer times. It thus remains unclear whether staircase procedures lead to a similar inflation of metacognitive ability (as compared to using a single contrast), especially across a more divergent set of measures of metacognitive ability.…”