Pragmatic impairments in autism spectrum disorders (ASD) are significant. The skill known as common ground (Clark, 1992), involving the ability to track what information is known by both participants in a conversation, is likely relevant to pragmatic skills, and is to date unexplored in ASD. Common ground skills could draw on Theory of Mind (ToM) as well as, potentially, working memory (WM) skills. This study assesses (1) the ability of individuals with ASD to maintain and update their representations of what knowledge is shared between conversational partners, and (2) the contributions of Theory of Mind (ToM) and WM skills to such representations.
The current study examined pragmatic language in ASD within a cooperative problem-solving task, drawing on both explicit responses and eye-tracking data to explore the relative contributions of ToM and WM load to perspective-taking abilities. We hypothesize that WM abilities contribute significant variance in task performance, over and above the contributions from ToM.
This study utilized a cooperative problem-solving task in which participants move shapes onto a computer display according to a trained partner’s spoken instructions. Some shapes are “unknown” to the partner. Studies of eye movement data in typical development suggest significant competition from such “secret” shapes. Children with ASD (n =17) and typically developing controls (n =22) matched on age (range 9-16; mean 12.9 years), IQ, and language completed the computer-based puzzle game while their eye movements were tracked. As a manipulation of WM load, the number of secret shapes varied from 1 to 4. The relationship between task performance and standardized measures of WM and ToM was also examined.
Participants were accurate in their behavioral responses (91%), with slower reaction times across groups when competing “secret” information was present during high WM load conditions, p<.02. Eye-tracking data indicated that all participants were slower to process a partner’s instruction when there were competing secret shapes, p<.02. This delay interacted with WM load, as participants were even slower to disambiguate target (correct) shapes from secret competitors under high WM loads, p<.001. The ASD group was slower to disambiguate the target, p<.001, was more influenced by the secret competing shape, p<.001, and had greater difficulty disambiguating the target in the presence of a secret competitor under high WM loads, p<.02. Across groups, performance errors were associated with language and WM, all p’s<.05, and in the ASD group errors correlated with symptom severity, p<.05. There was a trend towards an association between ToM ability and the ability to take secret shapes into account, p<.06.
Differences in performance for low versus high WM conditions suggested that WM modulates the ability to incorporate shared information, across all groups, with the ASD group particularly susceptible to WM load. Eye fixation patterns suggested that children with ASD were more influenced by “secret” information. Language and WM abilities were related to performance in this perspective taking task; TOM was weakly associated with task performance. Results are consistent with previous studies suggesting taking another person’s perspective requires additional processing capacity.