Abstract
Infants rapidly tune their face processing abilities during the first year of life. This developmental phenomenon has been called "perceptual narrowing" and results in a decline in ability to tell the difference between faces within less frequently encountered face groups. However, if infants are given individual-level face training (i.e., they learn the proper names for each face), they maintain their ability to discriminate between otherwise not salient exemplars (e.g., monkey faces). In contrast, infants trained with category-level labels (e.g., "monkey") exhibit the same decline seen without training. It is currently unclear whether or not the process of perceptual narrowing can be applied to object processing and whether attention or visual strategy use changes with individual-level training. The present investigation trained infants from 6- to 9-months of age with unfamiliar novel objects using individual or category level labels presented in a storybook. Before and after training, infants viewed pictures of untrained novel objects, as well as novel human and monkey faces while fixation durations were recorded with an eye-tracker (SR EyeLink). A 9-month-old untrained control group was also tested to determine general effects of training. Overall, infants fixated the objects significantly longer than the faces. After training, only infants trained at the individual-level decreased increased their looking time duration, relative to pre-training. In addition, infants trained at the individual-level looked significantly longer toward the right side of untrained exemplars within the trained novel object categories. No such looking bias was found for infants trained at the category level or infants in the control group. These results suggest that learning novel objects at the individual level leads to increased attention, indicating generalization of learning, after individual-level but not category-level training. In addition, the right-side looking bias may indicate more lateralized processing of stimuli when trained with individual-level labels.
Meeting abstract presented at VSS 2014