An Online Investigation of Face Training in a Large Sample of Developmental Prosopagnosics, Phase 1

Abstract

Case studies over the last 5 years have shown it is possible to improve face processing in developmental prosopagnosics (DPs) through cognitive training. Though these studies are promising, it's unclear whether face processing can be improved in a larger sample of DPs. Moreover, these studies did not sufficiently characterize the real-world significance of training-related improvements. We attempted to address these issues by recruiting a larger sample of DPs to perform an intensive face training program (45 min/day for 15 days) aimed at improving configural/holistic processing of faces. This program has shown previous evidence of training-related improvements in individual cases (for example, see DeGutis, 2007). To increase our sample we had subjects perform testing and training on the web while being in close communication with the experimenter. The assessment battery consisted of tests of face perception, memory, part and whole processing, and object processing. We compared the pre and post results of the training group with a group of DPs that were assessed before and after a 15 day wait period. The training group also completed a diary for 5 days before and after training to quantify potential real-world improvements. We found that, as a group, DPs significantly improved on accuracy and reaction time in the training task. Compared to the test-retest control group, trained DPs demonstrated significant improvements on whole trials in the part-whole task (similar to a recent intervention study with autistic children by Tanaka and colleagues (2010)). Unfortunately, group-level improvements were limited to the part-whole task and these improvements did not necessarily correspond with robust real-world improvements. In summary, it is possible to improve aspects of face processing in a larger sample of DPs through face training, but these improvements may be limited to the face processes trained and may be limited in their real-world implications.