Concurrent and Longitudinal Predictors of Theory of Mind in TD Children and Children with ASD

Background: Delayed or deficient theory of mind (TOM) development in children with ASD has been attributed as a root cause of many other patterns of social and cognitive impairment in autism. Indeed, school-age children with ASD usually perform significantly more poorly on TOM tasks than age-matched TD children (2, 3). Both linguistic and social measures have been shown to correlate with concurrent TOM performance (1, 4, 5); however, the longitudinal predictors/precursors of TOM are still not established.  Moreover, it is still unclear whether the same longitudinal predictors will be found for TD children and children with ASD.  We address these gaps with data from a longitudinal study in which both linguistic and social developmental precursors of TOM are collected. 

Objectives: We investigate both social and linguistic predictors of TOM, both concurrently and longitudinally, in 5-to 6-year-old children with TD or ASD. 

Methods: The participants were 27 TD children (MA_{visit 1}=1.7 years) and 17 children with ASD (MA_{visit 1}=2.71 years).  The TD and ASD groups were matched on Mullen expressive language at Visit1; however, by Visit 3, when the TOM tasks were administered (TD_MA=5.58 years, ASD_MA=6.54 years), the TD group had higher language scores (ps<0.05).  The Vineland Adaptive Behavior Scales, and the Mullen Scales of Early Learning were administered at Visits 1 and 2 (TD_MA=3.33 years; ASD_MA=4.5 years); the Test of Auditory Comprehension of Language-3 (TACL) was administered at V3. Two TOM tasks were administered: an unexpected contents task and an unexpected change in location task. Scores were averaged for an overall percent correct. 

Results: As expected, the TD group (Mpercent correct=89%, SD=16.69) had higher scores than the ASD group (M=43%, SD=40.43). Figure 1 shows 2 subgroups in the ASD sample; one subgroup scored in the TD range. Bivariate correlations revealed that the TACL-Q and Mullen-RL at V1 correlated positively with TOM scores in both groups (rs>.361, ps<.05). The TACL-Elaborated Sentences also correlated with TOM only in the TD group (r=.431,p<.05); the Vineland Communication at V1, MullenRL at V2, and TACL-morphology correlated with TOM only in the ASD group (rs>.663,ps<.05). Regressions revealed that for the ASD group the only significant predictors of TOM were the Vineland Communication at V1 (DR²=.546, p=.002) and Mullen-RL at V2 (DR²=.109, p=.075). For the TD group, the only significant predictor of TOM was TACL-Elaborated Sentences at V3 (DR²=.172, p = .034).

Conclusions: A subset of children with ASD appears to reach TD levels of TOM; however, their concurrent and longitudinal predictors differed from the TD group.  As predicted, complex syntax was the strongest predictor of TOM in the TD group (1, 4). In contrast, both social and more general linguistic measures were the strongest predictors of TOM in the ASD group. Thus, TOM may develop from different roots in children with ASD compared with TD children.