Camouflaging involves adjusting autism-related behaviors to appear neurotypical (Petrolini et al., 2023). Intellectual ability and theory of mind (ToM) may contribute to camouflaging variability (Livingston et al., 2019; Hull et al., 2021). Our group studies individuals with a history of autism who have lost the diagnosis (LAD); they no longer meet DSM-V criteria (Fein et al., 2013, Eigsti & Fein, 2022). We expect their ToM and intellectual abilities to be similar to neurotypical peers. Anecdotal evidence suggests that autism symptoms minimally affect their experiences. This study tested associations between ToM, verbal ability, and camouflaging in LAD, neurotypical, and autistic individuals. We hypothesized that verbal ability, but not ToM, would predict camouflaging in autistic people, given theories that some autistic people camouflage in spite of poor ToM and that IQ may help compensate for symptoms (Livingston et al., 2019). We hypothesized that both ToM and verbal ability would predict camouflaging in NT and LAD groups.

Participants ages 12-39 years completed interviews and surveys. Groups included ASD n=25, LAD n=26, NT n=34. Groups did not differ on age, race/ethnicity, or income; the NT group included more females. The ASD group had more autism symptoms and lower adaptive skills. The Camouflaging Autistic Traits Questionnaire (CAT-Q; Hull et al., 2019) measured camouflaging. The Social Attribution Task (SAT; Bell et al., 2010) measured ToM. The Penn Verbal Reasoning Task (PVRT; Gur et al., 2001) measured verbal cognitive ability. Multiple imputation replaced missing data; this dataset produced F and R² ranges. T-tests and chi-square tests assessed group differences. Relations among PVRT, SAT, and CAT-Q scores were tested in group-wise linear regression models. Post-hoc analyses explored correlations between CAT-Q scores and autism symptoms (ADOS-2; Lord et al., 2012).

The ASD group had higher CAT-Q scores than the NT group (t = -2.6; p = .01). Groups did not differ on PVRT or SAT. Neither overall models nor predictors were significant for LAD and NT groups. Only the ASD model significantly predicted camouflaging (F range = 3.66-4.91, p = .02-.04; PVRT and SAT R² range = 0.25-0.31). Within the ASD model, lower verbal efficiency corresponded to greater camouflaging (B = -4.4, SE B = 1.9, t = -2.4, p = .02). Higher ToM scores corresponded to greater camouflaging (B = 2.9, SE B = 1.4, t = 2.1, p = .04). Camouflaging and autism symptoms were not correlated in any group.

Camouflaging in LAD people fell between neurotypical and autistic levels, and the autism group endorsed significantly more camouflaging. Verbal cognitive ability uniquely influenced camouflaging in autistic people. Unexpectedly, the PVRT, which captures verbal problem-solving speed and accuracy, was negatively associated with camouflaging. Perhaps better verbal abilities allow people to quickly and flexibly adapt behavior and thus master camouflaging behavior earlier in development. Those with lower verbal efficiency may put more energy into compensating and thus endorse more camouflaging. Greater ToM may enable autistic individuals to camouflage more. ToM captured by the SAT may not tap social cognition skills used by more neurotypical people.