Temperament and Peer Victimization as Predictors of Facial Emotion Recognition Among Adolescents with and without High-Functioning Autism

Background: According to the “modifier model” of autism, variability in social skills and behavior is the product of biological causes of autism and environmental modifiers like social relationships (Mundy et al., 2007). Temperament factors like high Negative Affectivity (NA) are associated with poorer social skills and social-emotional outcomes (Schwartz et al., 2009).  Furthermore, abused children display an anger bias in facial emotion recognition tasks, suggesting that negative social relationships may contribute to biased emotion processing (Pollak et al., 2000).

Objectives: The purpose of the present study was to examine high-functioning autism (HFA), negative affectivity (NA) and peer victimization (PV) as predictors of emotion recognition, sensitivity and bias.   It was predicted that adolescents with HFA would demonstrate poorer overall identification of and sensitivity to emotional facial expressions.  Furthermore, it was predicted that higher NA and PV would be associated with lower sensitivity and negative emotion bias in both groups.

Methods: Adolescents with (n=7; M_age=15.2Y) and without (n=17; M_age=15.1Y) HFA participated in three computerized facial emotion recognition tasks.  Prototype emotion identification was tested with prototype happy, sad, and angry faces.  Emotion sensitivity was tested via short movie-like clips of a neutral face progressing towards a happy, sad or angry expression.  Emotion bias was tested using ambiguous morphed faces along happy-sad, happy-angry and sad-angry continua. Participants were also administered the Autism Diagnostic Observation Schedule, the Kaufman Brief Intelligence Test, the Early Adolescent or Adult Temperament Questionnaire, Social Experiences Questionnaire and Olweus Bully/Victim Questionnaire.   Data collection is ongoing.

Results: Preliminary results show that adolescents with HFA were less accurate at identifying prototype emotions than controls (t (22)=2.73, p < .05), specifically happy (t (22)=2.50, p<.05) and sad (t (22)=2.07, p=.05) faces.  However, they were not impaired (minimum accuracy ≥ 83%.  With regard to temperament, high NA was associated with a bias to identify anger when presented with ambiguous happy-angry faces (r=.42, p=.05).   Finally, while adolescents who had experienced overt PV did not show a significant emotion bias, they were observed to be less sensitive to subtle facial cues.  They required a higher level of emotion strength to correctly identify the emotion of facial stimuli progressing from a neutral to an expressive face (r=.43, p=.04).

Conclusions: Adolescents with HFA were less accurate in prototype emotion identification than peers without HFA; however, they were not impaired.  Sadness was the most challenging emotion for HFA to identify.  NA was associated with a bias to identify anger given ambiguous emotion stimuli.  This suggests that temperament may play a role in adolescents’ assessment of emotional information and is consistent with Schwartz et al. (2009). Our research failed to find that children with high PV have a bias to overidentify negative emotion.  We did find, however, that children who have experienced high PV show lower sensitivity to emotional cues.  It may be that lower sensitivity to emotional information may make adolescents vulnerable to PV.  This causal explanation requires further investigation.  The interaction between HFA, NA, and PV will be explored as sample size and power increase in this ongoing study.