Theory of Mind in Children with Autism Spectrum Disorder: A “Non-Verbal” Task Highlights the Importance of Language

Background: Autism Spectrum Disorder (ASD) is characterized by social deficits in a variety of domains, including Theory of Mind (ToM; Baron-Cohen et al., 1986). ToM is the understanding that people are independent agents with mental states such as knowledge and belief that direct goals, intentions, and actions (Baron-Cohen, 2000). Prior research has established that children with ASD require a minimum threshold of verbal ability to pass classical verbal ToM tasks (Fisher et al., 2005), but few have measured children’s performance on tasks that purposefully minimize verbal demands. The present study examines the efficacy of administering a minimally verbal ToM measure to children with ASD via touchscreen computer (to collect reaction times in addition to response accuracy). Reaction time is a potentially sensitive measure of processing effort, and is especially interesting given previous research suggesting that some children with ASD succeed on ToM tasks by utilizing different mechanisms than their typical peers (which may disrupt speed of response; Bowler, 1992).

Objectives: Evaluate ToM in typically developing controls (TDC) and children with ASD by comparing response accuracy and reaction times on a task with minimal verbal demands.

Methods: Twenty-five TDC and 50 ASD children participated (mean age=10.58 years). Two ASD subgroups were matched to the TDC group on chronological age and either Verbal IQ (ASD-VIQ mean=112.68, TDC mean=112.60) or Nonverbal IQ (ASD-NVIQ mean=103.76, TDC mean=103.88), and did not differ from one another on ADOS social interaction scores (ASD-NVIQ=7.39, ASD-VIQ=6.96). In the task, participants saw 48 comic strip-style storylines and three possible conclusions, choosing the picture that best completed the story (Brunet et al., 2003). There were three types of stories: physical causality with objects (PCO), physical causality with humans (PCH), and attribution of intentions with human characters (AI). 

Results: A MANOVA revealed between-group differences in response accuracy and reaction time. Planned comparisons showed that ASD-NVIQ responded less accurately than TDC to AI questions, t(48)=1.95, p=.06, and PCH questions, t(48)=2.16, p=.04. There were no group differences in response accuracy to PCO questions, p>.10. Similarly, ASD-NVIQ responded more slowly than TDC to AI questions, t(48)=2.03, p=.05, and trended toward being slower to answer PCH questions, t(48)=1.77, p=.08. No significant differences in reaction time between ASD-NVIQ and TDC on PCO questions emerged (p>.10). Response accuracy and reaction times in the ASD-VIQ group were equivalent to TDC across all categories, even after controlling for the effect of NVIQ (ps>.10). 

Conclusions: The ToM task described here was designed with low explicit verbal demands, but our results suggest that success was nonetheless impacted by verbal ability. Participants with ASD that were matched to the TDC group on VIQ were not impaired on this test, whereas participants matched on NVIQ were slower and less accurate answering questions requiring intention attribution, or that included human characters. Our findings suggest that ToM tasks without explicit verbal demands still have implicit demands, and lend indirect support to the idea that some persons with ASD may have a different mechanism for solving ToM tasks and thus answer more slowly than TDC (Bowler, 1992).