Dimension-Based Statistical Learning in Older Adults

The ability to perceptually “reweight” acoustic dimensions in response to changes in distributional statistics is known as dimension-based statistical learning (DBSL). However, it is currently unknown whether DBSL imposes a cognitive load. Older adults, who typically have age-related declines in cognitive ability, may be sensitive to this load. We examined young and older adults’ categorization of beer and pier sounds when the statistical relationship between VOT and F0 was consistent with that of American English, followed by a condition in which those statistics were reversed. Listeners made categorization decisions on each stimulus (Experiment 1), or after passive exposure to a string of stimuli (Experiment 2). In both experiments, younger and older participants demonstrated DBSL following exposure to the reversed statistics. Older adults tracked distributional statistics even when learning required accumulation of statistics over 8 sec, suggesting that rapid adaptation to regularities in speech input is robust across differing perceptual loads