Abstract
Can unconscious knowledge influence what we see? We present evidence for effects of rapid implicit learning of arbitrary crossmodal associations on visual perceptual selection. We measured perceptual selection with binocular rivalry, in which incompatible images presented separately to the two eyes result in a perceptual alternation between the images. Although there is evidence that sounds that are temporally (Kang & Blake, 2005), directionally (Conrad et al., 2010), or semantically (Chen et al., 2011) congruent with one of a pair of rivalrous visual percepts can promote dominance of the congruent percept, these effects could be attributed to explicit knowledge and attentional control. Here, we show that implicit audio-visual statistical learning (Fiser & Aslin, 2001, Seitz et al., 2006) influences perceptual selection. During a brief (10-minute) passive exposure phase, subjects were presented with rapid streams of images (gratings and simple patterns) and sounds (pure tones and chords). For half of the stimuli, a given sound was consistently paired in time with a given image, facilitating an association between the sound and image. The remaining sounds and images were randomly paired throughout the exposure phase. Audio-visual pairings were randomly assigned and counterbalanced across subjects. In a subsequent rivalry test, subjects were presented with combinations of sounds and rivalrous images, all from the exposure phase. At the onset of rivalry, subjects were more likely to perceive an image when it was presented with its auditory match than when it was presented with a non-matching sound. Importantly, these effects of implicit learning on rivalry did not correlate with subjects’ conscious knowledge of the audio-visual pairings, as assessed in a separate explicit learning test for each subject. Our results indicate that our prior knowledge—even that which is recently acquired and unconscious—helps us resolve ambiguity in the visual world.
Meeting abstract presented at VSS 2012