Beyond probability gain: Information access strategies in category learning


The present study uses eye-tracking to study information access in the context of category learning. Prior research has pointed toward the importance of probability gain, the increase in the chance of getting an answer correct, as a key variable in determining what information is considered most useful to acquire before making a classification decision. We manipulate the probability gain of three features in a four-category learning task by changing the base rates of the categories to be learned. Using participants’ eye-movements to determine the order in which they acquire information after many trials of training, we find that increasing the probability gain of a feature does bias participants’ first fixation. However, participants’ strategies for acquiring feature information indicate they are more sensitive to efficiency goals: even with the low cost of eye-movements, participants direct attention to maximize efficiency, and do so without trading-off accuracy.

Back to Table of Contents