Background: Numerous investigations have demonstrated that individuals with autism employ non-expert strategies to recognize and remember faces, compared with controls. A pilot investigation of expertise training demonstrated that individuals with autism could become experts at recognizing a set of novel faces and increase their awareness of configural information contained in the face. However, training did not improve performance on standardized measures of face recognition (Faja et al., 2008).

Objectives: The current study tested the effects of expertise training with faces or houses on adults with ASDs who were impaired on standardized measures of face recognition. Methods: Participants were randomly assigned to training group (10 Face, 9 House) and received 5-8 hours of computerized training. Behavior was examined before and after training using standardized tests of recognition and recall and experimental measures of configural awareness, holistic processing and the inversion effect. ERPs to faces and houses were also collected at baseline and after training.

Results: All but one participant met criteria for expertise (Tanaka & Taylor, 1991). Post training, there were no significant differences between groups in performance on the standardized face recognition measure. No changes were found in accuracy for recognizing manipulations of configural information, increased holistic processing or inversion effect. Performance of both groups improved on the standardized measure of face memory, t(16) = 2.14, p< .05 and for accuracy in configural recognition, t(14) = 2.22, p< .05. Results of the ERP data will also be discussed.

Conclusions: These results are consistent with previous findings of improvement resulting from expertise training, compared to a no training condition. Yet, expertise training in general appears to have an effect rather than training with faces, per se.