Proceedings of the Annual Meeting of the Cognitive Science Society

Tests of cross-situational word learning use a range of stimuli. How does the distinctiveness of a stimulus affect participantsability to learn its label? In two experiments, participants were presented with pairs of unfamiliar images accompanied bytwo pseudoword labels. The images were either two visually similar robots or two visually dissimilar novel objects. Bydesign, the mapping of label to image was purposefully unclear, and we further manipulated which images were displayedwith one another across trials (i.e., their presentation context). In one condition, pairs of images were randomly determined,while in the other, sets of images consistently appeared with one another throughout training. At test, participants weregiven one label and instructed to match it to one of four possible images. Participants who had been exposed to the visuallydissimilar objects outperformed those who had been exposed to the visually similar robots, regardless of presentationcontext.