Cross-situational word learning respects mutual exclusivity

Learners are able to infer the meanings of words by observ-ing the consistent statistical association between words and their referents, but the nature of the learning mechanisms un-derlying this process are unknown. We conducted an artifi-cial cross-situational word learning experiment in which either words consistently appeared with multiple objects (extra ob-ject condition) or objects consistently appeared with multiple words (extra word condition). In both conditions, participants learned one-to-one (“mutually exclusive”) word-object map-pings. We tested whether a number of computational models of word learning learned mutually exclusive lexicons. Simple associative models learned mutually exclusive lexicons in at most one of the two conditions. In contrast, a more complex Bayesian model—which assumed that only some objects were being talked about and only some words referred—learned mutually exclusive lexicons in both conditions, consistent with the performance of human learners