Listening in Your Shoes: Can Children with Autism Take the Perspective of Others When Interpreting Language?

Background: Individuals with Autism Spectrum Disorder (ASD) frequently fail to interpret the intent of a speaker’s utterance, apparently because they have difficulty determining the crucial aspects of common ground, which they share with the speaker. Only two studies (both with adults with ASD) have previously investigated this experimentally (Begeer et al., 2010; Sanstieban et al., 2015). Both manipulated level one visual common ground (whether the speaker can see a particular object). Both found that the participants with ASD were unimpaired relative to typical controls. However, visual perspective-taking may not align in development with social perspective-taking, which is understanding the interlocutor-specific experiences shared with the conversation partner. Objectives: To determine whether social versus visual perspective-taking have differential effects on the ability of children with ASD to interpret requests. Methods: We compared 24 eight- to eleven-year-olds with ASD with 23 typically-developing eight- to eleven-year-olds. Groups were matched on non-verbal IQ, receptive language, chronological age and gender. Children interpreted requests (e.g. ‘Can I have that ball?’) in contexts which would be ambiguous (i.e. because the child can see two balls) if perspective-taking was not utilized. There were three within-subjects conditions: Social perspective-taking, Level 1 Visual Perspective-taking (VPL1) and Level 2 Visual Perspective-taking. There were three trials per condition. For each the requester wore dark sunglasses and did not point. In the VPL1 condition one of the two objects was hidden from the viewpoint of the requester (E1). The correct choice was the object that the requester could see. In the Social Condition, the child was told that E2 had bought toys that E1 had not yet seen. E2 passed one of these (e.g. a pink ball) over to E1, who played with / discussed this with the child. Then E1 left the room and E2 showed the child another object of the same type (e.g. a yellow ball) and played with /discussed this with the child. When E1 returned to the room, both the child and E1 could see two balls as E1 excitedly verbalized the request. The correct choice was the object that was new for E1 (here: yellow ball). Results: Overall the group with ASD performed significantly worse than the typically-developing control (p = .032, ?p2 = .073). Their performance was not found to relate to affect recognition. There was also a main effect for condition (p = .033, ?p2 = .097). Children with ASD found the Social Condition significantly harder than the VPL1 Condition (p <.01). Nonetheless, Social and VPL1 conditions were strongly inter-correlated for children with ASD (r = .73, p < .001), even when non-verbal IQ, receptive language and age were partialled out (r = .73, p < .001). Conclusions: Children with ASD find it more difficult to use social than to use VPL1 to interpret language. VPL1 may be a more basic form of social perspective-taking (Harris, 1992) since the two are related. A large proportion of intellectually high-functioning children with ASD may have difficulty interpreting language if instructions or discourse require social perspective-takin