Abstract
A Bayesian account of perceptual learning predicts that learning should occur only when stimuli are informative about statistical contingencies in the environment (e.g. Kersten, O'Toole, Sereno, Knill & Andersen, 1987). Alternatively, learning could occur in the absence of informative cues to appearance, through practice of the perceptual decision itself. We assessed these two possibilities using a perceptually ambiguous Necker cube stimulus: Cue recruitment studies have shown the perceived rotation direction can be trained to be contingent on the stimulus' retinal location (Backus & Haijiang, 2007; Harrison & Backus, 2009). One group viewed only informative presentations, with the direction of cube rotation disambiguated by disparity and occlusion depth cues. Another group viewed uninformative, ambiguous, cubes for more than 96% of presentations. The remaining, informative, trials were sufficient to prime stabilization of the percept (Brascamp, Knapen, Kanai, Noest, van Ee & van den Berg, 2008; Klink, van Ee, Nijs, Brouwer, Noest & van Wezel, 2008), such that the two groups experienced equivalent pairing of perceived rotation direction with retinal location on Day 1. The long-term influence of perceptual experience on Day 1 was assessed on Day 2 by presenting subjects with a 50:50 mix of informative and uninformative stimuli. Informative stimuli had the reverse rotation-location contingency to that experienced the previous day. Those subjects whose perceptual experience on Day 1 had been elicited by the uninformative stimuli were affected very little by the reverse-contingency informative presentations on Day 2, and instead perceived ambiguous cubes as rotating in the same direction as Day 1. In contrast, subjects whose perceptual experience on Day 1 had been elicited by informative stimuli were more likely to perceive opposite rotation on Day 2. Hence, contrary to Bayesian prediction, long-term learning of perceptual appearance was largely driven by “practice”, perhaps of the decisional process, while informative presentations played a smaller role.
NIH R01-EY-013988, HFSP RPG 3/2006, NSF BCS-0810944.