Abstract
Perception is regularly discussed by analogy with higher-level cognition (e.g. Helmholtz’s ‘perception as inference’). But rarely do such analogies go beyond the metaphorical to permit an explicit comparison between the ways in which the perceptual brain and the executive brain deal with comparable problems. This study aims to do this. In a comparison between models of decision making and of perceptual bi-stability, we identified formal similarities between, first, how deliberate two-choice decisions are made and, second, how a percept emerges at the appearance of a bi-stable stimulus. We then experimentally evaluated this parallel. Specifically, properties of deliberate decisions are commonly quantified using a variable termed ‘response proportion’: the proportion of trials in a condition that yield a given choice. When analyzed in these terms, deliberate choices with a high response proportion (e.g. correct choices in an easy condition) involve both shorter response times and weaker decision-related pupil dilations than choices with a lower response proportion (e.g. choices in a harder condition). To test whether the same applies to non-volitional percept choices, we recorded pupils while intermittently presenting an ambiguous structure-from-motion sphere to observers who reported, at each stimulus onset, either leftward or rightward rotation: the perceptual outcome of a non-volitional decision. Because perception at such onsets depends on prior history (priming), we could use prior history to assign each onset a probability that either direction would be perceived: a quantity analogous to response proportion. We found that, indeed, manual responses were fastest, and pupil dilations smallest, for percept reports that were strongly predicted by prior history (i.e. percept choices with a high response proportion). These results show how analytical tools from the decision literature can be applied to a perceptual process, and suggest similarities between the ways in which the perceptual brain and the executive brain arbitrate between options.