Predicting Emotion Recognition Bias From Emotion Description in Adolescents with and without Autism

Background:  

Impairment in social interaction is a hallmark characteristic of autism spectrum disorders (ASD).  Two areas of impairment include accurately interpreting one’s own emotions and identifying the facial affect of others.  Previous studies have identified patterns of emotion recognition bias in high functioning autism (HFA) which are thought to cause misinterpretation of other’s emotions and harm social interactions (Macdonald et. al. 2006).  It may be possible that a person’s concept of his or her emotions influences emotion recognition bias.  This has not been examined.  

Objectives:

The current study aims to determine how adolescents with HFA compare to neurotypical (NT) peers in their open-ended descriptions of emotions and in labeling emotionally ambiguous facial expressions, and to describe the relationship between emotion description and emotion identification bias. 

Methods:  

8 HFA and 8 NT adolescents participated in a study of ambiguous emotion recognition. The groups were matched on gender and mental age (MA), with a mean MA of 15.54 (SD =2.76) for the HFA and 15.29 (SD =1.53) for the NT adolescents (t (7)=0.477, n.s.).  All adolescents were administered the Autism Diagnostic Observation Schedule (Lord et. al., 1989) to confirm diagnostic status.  During the ADOS, participants described how certain emotions make them feel.  Their responses were coded using a modified Linguistic Inquiry and Word Count system (Pennebaker & Francis, 1999). Participants were then presented images of ambiguous facially expressed emotion, which were created as morphs of two expressions, and were forced to choose between each morph’s comprising emotions, as in Beevers et. al.  (2008). Individual bias scores were calculated as the mean difference between the proportion of morphs participants labeled as each emotion and the percentage of the morph actually represented by that emotion.   Positive bias scores represent a higher tendency to label one emotion over another in these ambiguous faces.

Results:  

Results indicate that HFA adolescents were biased toward labeling an ambiguous face as happy t(6) =3.608, p <.05, and biased against labeling an ambiguous face as sad t(6) =-3.966, p<.01 when compared to NT adolescents.  No significant between group differences were observed for emotion description.  

Linear regression analysis revealed that the incidence of internal descriptors of fear (e.g. “I feel hollow inside”) rather than external descriptors (e.g. “I begin to tremble”) was a highly significant predictor of anger bias (β = .788, p < .001), accounting for 59.1% of the variance in anger bias scores.

Conclusions:  

This is the first study which investigates the association between emotion description and emotion identification bias.  Preliminary data indicate that propensity to describe fear in terms of internal feelings rather than through external signs is predictive of bias in labeling ambiguous faces as angry.  Those using internal descriptors may perceive more threat in their environment as they percieve more people with angry expressions.  That there were no significant group differences in this regard suggest that this effect is not specific to people with HFA, but may be shaped by temperament and environmental factors.