Autistic and non-autistic adults use discourse context to determine a speaker's intention to request.

The current study focuses on how autistic adults utilize context to determine whether ambiguous utterances (e.g., "I'm thirsty") are intended as indirect requests or as literal comment/questions. Two questions are addressed: (1) How do autistic adults compare to neurotypical adults in using context to interpret an utterance's intention as either literal or a request? (2) What cognitive mechanisms correlate with indirect request interpretation, and are these different for participants in each group? Twenty-six autistic and 26 neurotypical college students participated, engaging in an online experiment where they read narratives that ended with utterances open to literal or request interpretations, based on context. After each narrative, participants selected the best paraphrase of the utterance from two options, literal versus request. Following this task, participants completed two mentalizing measures (a false belief and emotion-identification task) and several executive functioning tests. The best model for predicting paraphrase choice included scores on the emotion-identification task and context as main effects, along with the interaction between both. Participants with higher emotion-identification test scores were more likely to provide correct paraphrases. Models including group as a main effect and/or interaction were not better at fitting the data, nor were any models that included executive functioning measures as main effects or interactions. Emotion-identification test scores, but not autism diagnosis, predict how adults infer whether an utterance is a request. Findings suggest that autistic adults use context similarly to neurotypical adults when interpreting requests, and that similar processes underlie performance for each group.