Metacognition, Mindreading, and the Hypercorrection Effect in ASD

Background:  Among neurotypical adults, errors made with high confidence (i.e., errors a person strongly believed they would not make) are corrected more reliably than errors made with low confidence. This “hypercorrection effect” is thought to result from enhanced attention to information that reflects a “metacognitive mismatch” between one’s beliefs and reality. This effect is thought to be tied to metacognitive monitoring ability (awareness of one’s own thoughts/cognitive states). Theoretically, metacognitive monitoring is itself considered to be related to mentalizing/mindreading ability (awareness of others’ mental states). Although mindreading is well-known to be impaired in ASD – thus providing grounds to predict metacognitive monitoring and the hypercorrection effect will be also – studies have only recently begun to explore metacognition, and its relation to mindreading and ASD features/traits, in this disorder.

Objectives:  This study addressed three central questions, two of which were entirely novel (1 and 3) and one of which (2) provided new evidence about a phenomenon only recently-studied in ASD:

1) What is the relation between metacognitive monitoring ability and a) mindreading ability, and b) ASD traits?

2) To what extent is metacognitive monitoring is impaired in ASD?

3) To what extent do children with ASD show a hypercorrection effect?

Methods:  In Experiment 1, n = 83 neurotypical participants answered general knowledge questions and provided confidence judgements about how likely each answer was to be correct, after which feedback (i.e., the correct answer) was given. Finally, participants were retested on all questions answered incorrectly during the initial phase. Mindreading ability (Reading the Mind in the Eyes) and ASD-like traits (Autism-spectrum Quotient) were also measured. In Experiment 2, so far 11 children with ASD and 11 age- and IQ-matched comparison participants have completed the hypercorrection task.

Results:

In Experiment 1, participants made accurate confidence judgements (i.e., a close correspondence between confidence in their answers to questions and actual success on those questions, reflecting good metacognitive monitoring) and showed the hypercorrection effect (high confidence errors initially more likely to be corrected at retest than initial low confidence errors). Mindreading ability was associated significantly with metacognitive monitoring (r = .27, p < .01) and ASD-like traits (r = -.35, p <.001). However, the hypercorrection effect was non-significantly associated with mindreading (r = -.16, p = .88) and ASD-like traits (r =.08, p = .45). In Experiment 2, children with ASD children are showing a large and significant diminution of metacognitive monitoring ability, t(20) = 2.00, p = .03, d = 0.86, yet a non-significantly larger hypercorrection effect than comparison participants, t(20) = 0.57, p = .58, d = 0.25. The evidence in favour of an undiminished hypercorrection effect (null result) is moderate, according to Bayesian analysis (Bayes factor = 0.21).

Conclusions:  These results provide support for the theory that metacognitive monitoring and mindreading are linked, and confirm that both are diminished in ASD. The hypercorrection effect appears normal in ASD, however, and does not rely on the same metarepresentational resources as metacognitive monitoring and mindreading. The implications for theory and educational practice will be discussed.