Assessing Face Processing Impairment In ASD Using the Benton Facial Recognition Test

Background: Face processing deficits in populations of individuals with autism spectrum disorders (ASD) are often reported, though findings vary according to type of task, sample characteristics, and study design. When face processing deficits are identified, conclusions regarding theoretical etiology of these deficits (social motivational vs. cognitive processing style) vary, including whether face processing deficits underlie broader social difficulties.

Objectives: The present study investigated face-processing abilities in ASD using the Benton Facial Recognition Test (BFRT), a widely used face-matching test. We hypothesized:

1.    Control group would score significantly higher than ASD group on BFRT.

2.    Classification of scores based on calculated, age-based z-scores would differ from classification based on Benton et al. (1994).     

3.    BFRT performance would significantly correlate with socialization.

Methods: Twenty-eight participants with ASD (23 male, 7.3-19.5 years old) and 29 typically-developing controls (24 male, 7.5-21.2 years old) from a larger emotion perception study completed the BFRT, the Wechsler Abbreviated Scale of Intelligence (WASI). Caregivers completed the Vineland Adaptive Behavior Scale-II (VABS-II). ADI-R and ADOS confirmed diagnoses.

Results: Based on BFRT education-corrected raw score classification (Benton et al. 1994), the control group consisted of 15 (52%) in the Normal range, six (21%) in the Borderline range, five (18%) in the Moderately Impaired range and three (10%) in the Severely Impaired range. Within the ASD group, 11 (39%) were in the Normal range, seven (25%) in the Borderline range, three (11%) in the Moderately Impaired range, and seven (25%) in the Severely Impaired range. BFRT raw scores were converted to age-based z-scores according to normative data (Benton et al., 1994; Baron, 2004) and re-classified using standard (Wechsler, 1997) cutoffs of Superior (z>1.33), Average (z= -1.33 to 1.26), Borderline (z= -1.4 to -2.0), Moderately Impaired (z= -2.067 to -2.667), and Severely Impaired (z<-2.7). Among controls, one was Superior (3.4%), 25 were Average (86.2%), two were Borderline (6.9%), and one was Moderately Impaired (3.4%). Within the ASD group, 24 were Average (85.7%), two were Borderline (7.1%), one was Mildly Impaired (3.6%), and one was Severely Impaired (3.6%). Age-based z-scores revealed significantly poorer performance by the ASD group (t(55)=1.682, p=.049), whereas raw scores did not reveal this difference (t(55)=1.595, p=.058). However, neither BFRT raw nor z-scores correlated to VABS-II Socialization (ps>.05).

Conclusions: Evidence of significant differences in face recognition abilities was not detected in this sample when using raw scores, but were detected using z-scores based on age-based norms. Conceptualization of level of impairment can be impacted greatly by the classification system used to assess performance. However, despite reclassification of age-based z-scores using standard Wechsler classification cutoffs, both control and ASD group scores spanned the Average to Impaired ranges, with the vast majority of the participants’ scores classified as Average. Furthermore, the expected relationship between face processing and socialization was not found. Given face-recognition deficits in ASD are well-established in the literature, results suggest that the BFRT may either lack sufficient sensitivity to detect differences in ASD or tap into skills other than configural, whole-face recognition abilities, for which it commonly used.