Abstract

There are an infinite number of possible word-to-world pairings in naturalistic learning environments. Previous proposals to solve this mapping problem focus on linguistic, social, representational constraints at a single moment. This paper investigates a cross-situational learning strategy based on computing distributional statistics across words, across referents, and most importantly across the co-occurrences of these two at multiple moments. We briefly exposed adults to a set of trials containing multiple spoken words and multiple pictures of individual objects with no information about wordpicture correspondences within a trial. Nonetheless, subjects learned over trials the word-picture mappings through crosstrial statistical relations. Different learning conditions compared the degree of within-trial reference uncertainty, the number of trials and the length of trials. We also propose and implement a computational model and feed it with the same training data used in different learning conditions in experimental studies, to shed light on the possible underlying mechanism of statistical learning. Overall, these results suggest that statistical cross-situational learning may be one of fundamental mechanisms to tackle the word-to-world mapping problem.

Citations