Studying Changes to Face Encoding After Category Learning Through Sensitivity Thresholds

Our visual sensory neurons encode objects, such as faces, into complex neural representations which can be influenced by learning. Faces are considered high-level stimuli, and their representation influences our underlying perceptions and biases. The way in which we learn face categories may contribute to how we discern new faces, leading to effects such as the own­ race bias, where we are better at distinguishing members of our own race compared to other races. To learn what mechanism of encoding change might accompany category learning. I modified several population encoding models to see, through simulation, if the changes induced would produce unique patterns of behavioral sensory thresholds. Most of the simulated models, with the exception of specific suppression with nonspecific gain, were recoverable. I then collected sensitivity thresholds from participants before and after learning of a face categorization task. Of the recoverable mechanisms of encoding change, specific gain at moving learning targets most often offered the best fit across participants. These results suggest that category learning tends to increase the responsiveness of populations of neurons that encode faces in two areas of the trained dimension, one on each side of the category boundary.

Major Professor: Fabian Soto