Joint or Conditional Probability in Statistical Word Learning: Why decide?

Abstract

Three experiments investigated the ability of human learners to concurrently extract and track both joint and conditional probabilities in statistical word learning. In each experiment, participants were briefly trained on novel word– novel object pairs and asked to learn correct mappings by the end of training. Across a series of learning conditions, we systematically manipulated conditional and joint probabilities individually and in combination to determine whether learners are able to encode multiple statistics in various learning contexts. Our results suggest that participants acquired both joint and conditional probabilities of word-referent co-occurrences. Based on the results from these experiments, we propose that learners are capable of utilizing the most reliable statistics that they acquired in training to make correct judgments in various testing tasks. These results suggest that statistical word learning is not only powerful but also adaptive.


Back to Saturday Posters