Processing of Self-Referential Information in High-Functioning Children with Autism

Background: Developing a strong sense of self is an important developmental precursor for understanding others and developing social skills. Preferentially remembering self-relevant information, particularly of positive valence is an adaptive process, leading to enhanced self-esteem and efficient memory storage and retrieval. In general, it appears that individuals with autism do not efficiently scaffold personally relevant information, as they do not demonstrate the typical pattern of preferentially remembering self-relevant over other-relevant information. This may hinder their ability to understand their thoughts and emotions as well as the experiences of others, leading to the observed social skill deficits. However, little is known regarding the mechanisms underlying these differences in self-representation.

Objectives: The current study seeks to examine the role of the emotional valence (positive and negative) of self-relevant trait adjectives in influencing how individuals with autism process self-referential information.

Methods: Participants with high functioning autism (HFA; N=73, 62 males, M_age=12.56, SD=2.60) and a matched comparison sample (COM; N=64, 46 males, M_age=13.56, SD=2.01) completed a self-referenced memory paradigm, where they were asked to judge whether trait adjectives were characteristic of themselves, characteristic of another person, or based on a structural feature of the word. An equal number of positive and negative adjectives were included in each condition.  After a short delay participants were asked whether they recognized the adjectives from a list of the original words interspersed among additional novel distractor adjectives.

Results: A repeated-measures ANOVA was conducted to examine the differences in rates of adjective endorsement based on valence and personal relevance. Results revealed a significant Group by Valence interaction in endorsements of self-relevant adjectives, F(1, 135)= 5.51, p=.02, but not in endorsements of other-relevant adjectives, F(1,131)=.29, p>.05 . In the self-referential condition, children with HFA (M=2.37, SD=1.72) endorsed greater rates of negative adjectives than their COM peers (M=1.54, SD =1.46), but comparable levels of positive adjective endorsement (M_HFA=5.42, SD=1.43; M_COM=5.51, SD=1.20). There were no group differences in rates of endorsement of positive or negative adjectives when children referenced another person. To examine the relationship between endorsement and recognition, a hierarchical linear regression was run, including age and verbal IQ as covariates. In the COM group, the endorsement positivity bias, or difference between number of positive and negative adjectives endorsed, predicted preferential self-referenced memory, over and above the effect of age and verbal IQ, F(3, 55)=6.17, p=.001, R²=0.25, R²_change=0.22. This was not true for the HFA participants, F(3, 64)=0.24, p>.05, R²=0.01.

Conclusions: As hypothesized, HFA participants demonstrated a reduced positivity bias when compared to their typically developing peers, which was driven by greater rates of endorsement of negative self-relevant adjectives. Interestingly, endorsement rates related to later memory performance in COM but not HFA individuals. The reduced positivity bias in individuals with autism may de-couple the associations between endorsement and recognition of self-referential information, leading to less efficient processing of incoming self-referential information. Results will be discussed in the context of the role of biased social cognition in the development of social competence, and how these concepts unfold over time.