A multimodal advantage in statistical and artificial grammar learning

Glicksohn, A. & Cohen, A.

The Hebrew University

Our environment is richly structured with objects producing correlated information within and across sensory modalities. A prominent challenge faced by our perceptual system is to learn such regularities. Numerous studies reveal both infants? and adults? exceptional ability in learning various types of regularities. Moreover, during initial development multimodal information benefits certain forms of learning. We ask (1) whether learning occurs separately within modalities (e.g., vision, audition) or whether learning is a-modal, and (2) whether multimodal information benefits learning. We explored these questions using two prominent learning paradigms: statistical learning (SL), referring to learning of statistical regularities between elements such as transitional probabilities, and artificial grammar learning (AGL), referring to learning the underlying grammar of a set of exemplars. Adult participants were familiarized with one novel environment: (a) visual, (b) auditory, or (c) an audiovisual environment consisting of one type of regularities (transitional probabilities, or exemplars adhering to a grammar). Two main findings are noted: (a) learning occurred in all three environments and for both types of regularities, supporting the existence of an a-modal learning mechanism(s); (b) Learning for the audiovisual regularities was significantly higher than learning of the visual- or auditory- regularities, suggesting an advantage for multimodal learning. We suggest that learning of regularities is accomplished via a general, a-modal, learning mechanism which is sensitive to multimodal information. We further suggest that multimodal information plays a crucial role in learning throughout our lives.