The Effects of Task Demands on Self-Monitoring for Higher Functioning Children with Autism

Background: It has been proposed that perseverative responding and the inability to adapt to social situations may reflect an underlying impairment in self-monitoring in individuals diagnosed with Autism Spectrum Disorder (Mundy, 2003; Henderson et al., 2006).  Self-monitoring includes the ability to detect and correct errors.  

Objectives: In the current study, the effect of stimulus type (non-social versus social) on post-error slowing, a behavioral index of self-monitoring, was examined among a sample of children diagnosed with High Functioning Autism and an age- and IQ-matched comparison sample.

Methods: Preliminary data are presented on 46 ASD and 45 age- and Verbal IQ-matched comparison children, between the ages of 9 and 17 who completed the Eriksen Arrow Flanker task. To-date, 27 ASD and 39 comparison children have also completed the Faces Flanker task.  In the tasks participants were presented with a horizontal array of either 5 arrows (Arrow task) or 3 faces (Faces task). Participants were instructed to attend to the center stimuli, and to press a key corresponding to either the direction of the arrow or affect of the face. The dependent variables of interest for each task are (1) reaction time (RT) on correct trials following correct vs. error trials and (2) the number of correct vs. error responses on the trial following an error.

Results: On the Arrows task, a 2 (preceding trial:  correct vs. error) x 2 (diagnostic group: HFA vs. comparison) repeated measures ANOVA with RT on correct trials as the dependent variable showed that RT was unrelated to preceding trial or diagnostic group.  In contrast, on the Faces task a comparable analysis revealed an interaction between preceding trial and group, F(1, 64)=4.82, p=.03, η_p²=.07.  Specifically, children with HFA responded faster post-error than they did post-correct, t(26)=2.57, p=.02, whereas RTs did not differ based on the previous trial for children in the comparison sample. Response accuracy following error trials was analyzed with a 2 (diagnostic group: HFA vs. comparison) x 2 (accuracy: error vs. correct) repeated measures ANOVA.  On both tasks, performance following an error was predicted by an interaction between group and accuracy, F(1, 78)=6.75, p=.01, η_p²=.08 (Arrows) and F(1, 55)=4.43, p=.04, η_p²=.08 (Faces), such that HFA children were more likely to repeat errors compared to comparison children, t(78)=2.42, p<.001 (Arrows) and t(55)=1.23, p=.04 (Faces).  While comparison children made more correct than error responses following errors on both tasks, the HFA children only reliably corrected themselves on the Arrows task, t(89)=2.02, p=.04, but were unable to do so on the Faces task, t(64)=.59, p=.45.

Conclusions: Regardless of task, the HFA children seem to have more difficulty self-monitoring.  Performance seemed particularly affected when stimuli were social, where children with HFA decreased their RT, or sped up, following an error, and were more likely to make repeated errors.  These findings suggest that response monitoring may be particularly impaired in social contexts and may underlie some of the difficulties HFA children have with reciprocal social interactions.