Performance-Monitoring and Evaluative Control in High Functioning Autism

Background: Decision making difficulties in autism may arise in part from impaired awareness of feedback, including negative feedback. The error-related negativity (ERN) and post-error positivity (Pe) are event-related potential (ERP) components generated following the commission of errors. Current theories suggest the ERN reflects automatic performance- and error-monitoring while the Pe reflects error-processing and awareness. As reflections of the response monitoring system, these components have direct behavioral implications in self-monitoring and decision-making in social-emotional processes (Crowley et al., 2009). Findings from such paradigms may be especially helpful for elucidating individual differences across the autism spectrum (Henderson et al., 2006). Objectives: In light of somewhat disparate findings in the few previous studies of the ERN in autism, we included measures of anxiety, personality, and cognitive functioning to provide further information on factors that influence response monitoring. In response to suggestions by Henderson et al. (2006) we also included more trials in order to generate more errors; and we had a somewhat larger and balanced sample size (25 in each group). Methods: ASD was characterized using both the ADOS-G (total social communication score >7) and the SCQ (total score >15). High-density ERPs were acquired while 25 ASD participants and 25 matched controls performed a modified version of the Ereksen Flanker task over 400 trials. Response-locked ERPs were separately averaged for correct and error trials. Results: Behaviorally, both groups demonstrated robust response-time and error-rate interference. Similar to Groen et al. (2008), we did not find a significant group x accuracy interaction for either the ERN, nor for the Pe. However, visual inspection showed that the response to error trials but not correct trials was attenuated in the ASD group; t-test analysis confirmed this difference was significant. Unlike Henderson et al. (2006) we did not find an association between Verbal IQ and the ERN in autism; this association was significant for the control group, however. Behavioral inhibition, as reported by parent questionnaire, was significantly associated with the difference score between correct and incorrect trials in the Pe, driven by a significant positive correlation between the Pe and behavioral inhibition.
Conclusions: The ERN offers the potential for rich insight into response monitoring in ASD, but task-specific and sample-specific differences across studies have thus far preclude a clear understanding of the phenomenon. We encourage further studies that give explicit attention to important factors of diagnostic severity, cognitive function, anxiety and personality to help identify possible subtypes of autism that could be characterized by reliable physiological measures such as the ERN.