Warren, D. , Roembke, T. , McMurray, B. & Duff, M.
University of Iowa
Cross-situational word learning, in which mappings between arbitrarily paired words and objects are learned through statistical co-occurrence, promotes rapid acquisition of new words, but its neural bases are not well understood. Neuropsychological evidence suggests that learning arbitrary relations requires medial temporal lobe structures, specifically hippocampus. We investigated whether hippocampus is necessary for cross-situational word learning by testing a severely amnesic patient with focal bilateral hippocampal damage (N=1). During the task, each trial presented pairs of unfamiliar objects and one novel word. Participants selected the object that was associated with the word. Learning was supported by the reliable co-occurrence of one word with each object (45 presentations/word×3 blocks); each trial?s second, competitor object was randomly selected. Patient performance improved from chance to >75% correct for 5 of 8 items, and patient recognition of studied words and objects was good (Mwords=83%, Mobjects=100%). Notably, patient learning was slower than that of healthy comparison participants, and further studies will be needed to fully characterize hippocampal influences on the rate and durability of cross-situational word learning. However, these preliminary results suggest that the hippocampus is not necessary for cross-situational word learning, a finding with important implications for memory theory and strong translational potential.