Abstract
Visual categorization and learning of visual categories are fundamental cognitive processes not yet well understood in infants. The main limiting factor for examining these mechanisms is the limited duration of infant cooperation (10-15 minutes), which leaves little room for multiple test trials. With its tight link to visual attention, eye tracking is a promising method for getting access to the mechanisms of category learning. But how should researchers decide which aspects of the rich eye tracking data to focus on? In the past, eye tracking variables were picked by hand, often resulting in biases or sub-optimal exploitation of the eye tracking data. Here we propose an automated method for selecting good eye tracking variables based on their usefulness to discriminate learners from non-learners of visual categories. We presented infants and adults with an unsupervised category learning task and tracked their eye movements. We then extracted an overcomplete set of eye tracking variables, encompassing durations, probabilities, latencies, and the order of fixations and saccadic eye movements. We compared three statistical techniques for identifying those variables among this large set that are useful for discriminating learners form non-learners: ANOVA ranking, Bayes ranking, and L1 regularized logistic regression. We found remarkable agreement between these methods in identifying a small set of features. These features allowed us to discriminate learners from non-learners in the adult population with 76% accuracy (chance: 50%), using a linear support vector machine. Moreover, the same eye tracking variables also allow us to classify category learners from non-learners among 6-8 month-old infants with 70% accuracy. This result suggests very similar processes underlying unsupervised category learning in infants and adults.
Meeting abstract presented at VSS 2012