Abstract
Recent studies have traced the neural correlates of confidence in perceptual choices using statistical signatures of confidence. The most widely used statistical signature is the folded X-pattern, which was derived based on a Bayesian definition of confidence as the posterior probability of being correct. The folded X-pattern entails that confidence as the subjective probability of being correct equals the probability of 0.75 if the stimulus is neutral about the choice options, increases with discriminability of the stimulus in correct trials, and decreases with discriminability in incorrect trials. We present theoretical and empirical reasons why assuming the folded X pattern a priori may lead to inaccurate conclusions about the neural correlates of confidence. On theoretical grounds, we demonstrate analytically that Bayesian confidence in incorrect choices increases, not decreases with discriminability if observers obtain reliable trial-by-trial evidence about discriminability itself. Moreover, Bayesian confidence in choices about neutral stimuli is not .75 if discriminability is varied in discrete steps within the experiment. For the empirical reasons, confidence judgments were inconsistent with the folded X pattern in three different experiments, a masked orientation discrimination task, a random-dot-motion discrimination task, and a low-contrast number discrimination task. Finally, we demonstrate how reliance on the folded X-pattern leads to identify false neural correlates of confidence. In the masked orientation discrimination task, EEG recordings revealed that the Pe component at centroparietal electrodes 200 – 500 ms after participants’ orientation judgments closely resembled the folded X-pattern. However, behavioural confidence matched neither the folded X-pattern nor the Pe component. Instead, confidence judgments were closely related to the P3 component recorded over centroparietal electrodes 300 – 500 ms after onset of the target stimulus. Overall, identifying neural correlates of confidence by presupposing a priori the folded X-pattern as the statistical signature of confidence is not legitimate.