Juliane T. Zimmermann ¹ T M. Ellison ² Francesco Cangemi ³ Simon Wehrle ² Kai Vogeley ^1,4 Martine Grice ^2*

• ¹ Clinic for Psychiatry and Psychotherapy, University Hospital of Cologne, Cologne, Germany
• ² IfL – Phonetics, University of Cologne, Cologne, Germany
• ³ The University of Tokyo, Bunkyo, Tōkyō, Japan
• ⁴ Cognitive Neuroscience (INM-3), Institute of Neuroscience and Medicine, Julich Research Center, Helmholtz Association of German Research Centres (HZ), Jülich, North Rhine-Westphalia, Germany

Although mentalizing abilities in autistic adults without intelligence deficits are similar to those of control participants in tasks relying on verbal information, they are dissimilar in tasks relying on non-verbal information. The current study aims to investigate mentalizing behavior in autism in a paradigm involving two important nonverbal means to communicate mental states: eye gaze and speech intonation. In an eye-tracking experiment, participants with ASD and a control group watched videos showing a virtual character gazing at objects while an utterance was presented auditorily. We varied the virtual character’s gaze duration towards the object (600 ms or 1800 ms) and the height of the pitch peak on the accented syllable of the word denoting the object. Pitch height on the accented syllable was varied by 45 Hz, leading to high or low prosodic emphasis. Participants were asked to rate the importance of the given object for the virtual character. At the end of the experiment, we assessed how well participants recognized the objects they were presented with in a recognition task. Both longer gaze duration and higher pitch height increased the importance ratings of the object for the virtual character overall. Compared to the control group, ratings of the autistic group were lower for short gaze, but higher when gaze was long but pitch was low. Regardless of an ASD diagnosis, participants clustered into three behaviorally different subgroups, representing individuals whose ratings were influenced 1) predominantly by gaze duration, 2) predominantly by pitch height, or 3) by neither, accordingly labelled “Lookers”, “Listeners” and “Neithers” in our study. “Lookers” spent more time fixating the virtual character’s eye region than “Listeners”, while both “Listeners” and “Neithers” spent more time fixating the object than “Lookers”. Object recognition was independent of the virtual character’s gaze duration towards the object and pitch height. It was also independent of an ASD diagnosis. Our results show that gaze duration and intonation are effectively used by autistic persons for inferring the importance of an object for a virtual character. Notably, compared to the control group, autistic participants were influenced more strongly by gaze duration than by pitch height.