[PS-2.11] Letting go of informative perceptual cues with distributional learning

Harmon, Z. , Idemaru, K. & Kapatsinski, V.

University of Oregon

Categories can be learned from probability distributions: a unimodal distribution suggests a single category, while a bimodal one suggests two contrasting categories (Maye & Gerken, 2000). Research on distributional learning has focused on developing a contrast through exposure to a bimodal distribution. Here, we instead aim to show that exposure to a unimodal distribution can help shift attention away from a previously informative perceptual cue, redefining a pre-existing contrast. 120 adult native English speakers were exposed to either bimodal or unimodal VOT distributions spanning the unaspirated/aspirated boundary (bear/pear). VOT is the primary cue to initial stop voicing. However, we paired stimuli with pictures of bears and pears independently of VOT in training. For each stimulus, participants were asked to guess the referent and received (random) feedback, generating an error signal suggesting that VOT is no longer informative and should be downweighed. In this design, the bimodal distribution provides a clearer error signal than the unimodal distribution. Nonetheless, participants downweighed VOT and switched attention to the secondary cue to voicing, F0, only after unimodal training. We conclude that a unimodal distribution strongly cues absence of contrast along a particular dimension, providing a novel way to effect an extradimensional attention shift.