Holt, L.
Learning often takes place without the support of instruction, explicit training, or overt feedback. Statistical learning, the process of becoming sensitive to statistical structure in the environment, is an influential example. Across species and development, organisms learn statistical regularities passively experienced over spoken syllables, visual shapes, tactile input, nonlinguistic tones, and even semantic categories without the benefit of explicit feedback, instruction, directed attention to the stimuli, or even an overt task. Yet, learning via passive accumulation of regularities fails under some circumstances. Troublingly, these circumstances mimic some of the complexities of learning in the natural world, such as those presented by substantial acoustic variability across talkers or continuous, fluent speech input. The core unanswered question, then, is how learning statistically-structured input proceeds when passive exposure is insufficient to drive learning and yet there is no explicit instruction or overt feedback. I will describe a program of research that examines the intermediate ground between passive exposure and instruction. This work examines whether active engagement in a multimodal perceptual environment ostensibly unrelated to learning supports statistical learning by virtue of temporal alignment of statistically-structured input with behaviorally-relevant actions, objects, and events. In this context, I will describe studies of adults learning speech and nonspeech auditory categories and present a candidate neurobiological network to support this incidental statistical learning.