Abstract
Object recognition and discrimination can be significantly improved by perceptual learning. However, the neural mechanism of object perceptual learning remains illusive. In this study, we searched for the functional and structural correlates of face view discrimination learning in human brain. Subjects were trained to discriminate face views around an in-depth face orientation of 30°over eight daily sessions, which resulted in a significant improvement in sensitivity to the face view orientation. This improved sensitivity was highly specific to the trained orientation and persisted up to one month. Before, immediately after and one-month after training, subjects underwent MRI scans to obtain functional and structural brain images. We first performed univariate amplitude analysis and multivariate pattern analysis of BOLD signals responding to the trained and untrained face views in six face selective cortical areas (OFA, STS and FFA in both hemispheres). We found that, relative to the untrained views, the mean amplitude of BOLD signal in the left and right FFA increased for the trained view immediately after training. But the increase was short-lived and it disappeared one month later. On the other hand, training improved the stability of the spatial activity pattern for the trained view in the left FFA. The improvement persisted even one month later and was correlated with the behavioral improvement. Then, we performed an ROI analysis of the face selective areas and a whole cortical surface analysis to measure cortical thickness before and after training. Although little cortical thickness change was detected after training, we found that the cortical thickness of the left FFA before training was inversely correlated with the behavioral improvement. That is, the thinner the cortex in the left FFA, the greater the learning effect. Taken together, these findings provide converging evidence that the left FFA plays a pivotal role in adaptive face processing.
Meeting abstract presented at VSS 2013