Skip to main content

Dr. Linda Smith, 4:00pm May 3rd

Dr. Linda Smith

Distinguished Professor and Chancellor's Professor of Psychological and Brain Sciences

Indiana University, Bloomington

    Tuesday, May 3rd
    Swift 107
    Reception to Follow

Rethinking referential ambiguity:  Clear cases and noisy data in statistical word-referent learning

For theorists of early word learning, the richness of the visual world is a problem.  The novice learner must figure out the referents for not known words by linking those heard names to seen things.  But the complexity of real world scenes means that there are many likely referents in any scene for a co-occurring unknown word. Researchers have sought solutions to this problem primarily outside of vision and the visual information scenes. This talk will consider how visual statistics themselves may simplify the learning problem, and at the very least should change the questions we ask and the way we do experiments. The data derive primarily from the study of the statistical structure of infant egocentric views.