Interactions between statistical aggregation and hypothesis testing mechanisms during word learning

Alexa RombergIndiana University
Chen YuIndiana University

Abstract

Adults, children and infants are all able to infer likely word meanings based on the relative frequency with which labels and referents appear together (e.g., Smith & Yu, 2007; Yu & Smith, 2008). However, the extent to which learners rely on aggregation of co-occurrence statistics vs. test specific hypotheses to infer mappings is currently a matter of significant uncertainty (Smith & Yu, 2012), exacerbated by the different experimental methods used to test learning mechanisms. Real world word learning is likely to involve a combination of statistical aggregation and active hypothesis testing. The current experiment investigates how these two learning mechanisms interact during word learning by having participants respond to a subset of items during a cross-situational word learning task. We find that hypothesis-testing is most effective when informed by statistical information and that the process of hypothesis-testing draws attention away from the remaining set of items.

Files

Interactions between statistical aggregation and hypothesis testing mechanisms during word learning (452 KB)



Back to Table of Contents