Can I trust in what I see? – EEG evidence for reliability estimations of perceptual outcomes

Abstract

Background: During observation of ambiguous and low-visibility stimuli perception becomes unstable and may alternate between different interpretations. Tiny low-level changes can disambiguate such a stimulus and/or increase its visibility and thus stabilize its percept. Methods: We compared event related potentials (ERPs) evoked by ambiguous and low-visibility stimuli with disambiguated and high-visibility stimulus variants across different visual categories (geometry, motion) and complexity levels (up to emotional face expressions). Results: Disambiguated and highly visible stimulus variants cause stable percepts and evoke much larger amplitudes of two positive ERP components than ambiguous stimuli (d > 1). This pattern of ERP results is highly consistent both in space and time across very different categories and complexity levels. Discussion: The generality of our findings points to high-level mechanisms: We postulate that a meta-perceptual/cognitive inference unit evaluates the reliability of perceptual constructs beyond sensory details. Small ERP amplitudes reflect high small amplitudes low perceptual reliability. I will discuss our results in the context of predictive coding theories as ERP correlates of prediction success with remarkably large effect sizes.