18658
Firm Evidence for Facial Identity Recognition Deficits in ASD Using a Task with No Memory Component
Impairment in facial identity recognition is a widely reported feature of autism spectrum disorders (ASD), and may be a core component of the social deficits that characterize ASD (Schultz, 2005). Recently, some have argued that these facial recognition deficits arise from face-specific memory impairments rather than perceptual processes (Weigelt, Koldewyn, & Kanwisher, 2012, 2013); however, findings of atypical visual search patterns (Chawarska & Shic, 2009) and neural activation (Dawson et al., 2005) during face perception, as well as evidence of impairments on face processing tasks without memory demands (Annaz et al., 2009; Wolf et al, 2008) contradict this claim. Moreover, some identity perception tests used in prior studies may have been too easy, such that most participants performed close to ceiling.
Objectives:
The present study aimed to resolve these inconsistencies by examining the performance of a very large sample population on the Benton Facial Recognition Test (BFRT; Benton et al., 1994), a challenging simultaneous-presentation recognition measure.
Methods:
The BFRT was administered to participants ages 5 to 50 (n = 419 ASD and 500 typically developing controls, TDC) as a part of a core phenotyping battery used across studies by the PI at two different sites (Yale and UPenn). Other core measures included a standard IQ test chosen based on age and language level (e.g. DAS, WISC) and gold standard diagnostic measures (ADOS and ADI-R). During BFRT administration, participants were prompted for a response after 12 seconds and item booklet page was turned after 16 seconds, a cutoff chosen from extensive pilot testing with TDCs. Participants over 20 years of age (nearly all TDC) or with unusually low Benton scores (< 23; chance=27) were excluded from analyses, bringing the final sample size to 410 ASD and 400 TDC. ANOVA revealed a significant main effect of gender; thus, gender was included in the final model. There were significant main and diagnostic group interaction effects of site on BFRT scores, but these effect sizes were tiny (ηp2s <.01). Regression analyses revealed significant association between BFRT, age (r=0.55) and IQ (r=0.27, with no significant difference by diagnostic group). Thus, an ANCOVA for BFRT score with diagnosis, gender, and site as explanatory variables and age and IQ as covariates was used to evaluate diagnostic group differences.
Results:
Analysis revealed a highly significant main effect of diagnosis, with participants with ASD scoring lower than TDCs, F(1, 800) = 71.18, p < .001, ηp2 =0.08, d=0.60. Regression analyses (controlling for age) showed that lower scores on the BFRT were associated with greater ADOS total scores in the ASD group (partial r=-0.25).
Conclusions:
Despite recent claims that difficulties with facial identity in ASD are restricted to tasks with a memory component, we find that youth with ASD have significant facial recognition deficits compared to TDCs (with a medium effect size), controlling for age, IQ, and gender, on a task that requires discrimination of identity from simultaneously presented face pictures, with no overt memory component.