A key manifestation of social communication deficits in autism spectrum disorder (ASD) is impaired perception of speech prosody. Prosody is essential to human verbal communication, conveying affective (emotions) and linguistic information (stress, intonation) through modulation of pitch and amplitude contours. A large body of behavioral research has shown that individuals with ASD are impaired in receptive and expressive prosody. While this could intuitively be attributed to deficits in processing of social information, an alternative is that it stems from deficits in auditory processing, which is also known to be impaired in ASD, for example aberrant auditory processing of simple auditory tone and speech sounds. Consistent with this hypothesis, neuroimaging work shows that individuals with ASD recruit additional cortical networks outside the core language areas when processing prosodic content. Moreover, activation of language networks when listening to speech is more bilaterally distributed in ASD than in typically developing (TD) controls. Finally, human neurophysiology studies show distortions in neural responses to auditory deviance and isolated words. However, the question whether deficits in prosody perception arise from auditory deficits in processing pitch and amplitude contours in natural continuous speech remains unanswered.

With this project we propose to close this gap, by addressing the following two questions: 1) Is the neural representation of acoustic cues to prosodic content, particularly speech envelope modulations and relative pitch, distinct between TD individuals and people with high-functioning ASD? 2) Are individual differences in neural coding of acoustic-prosodic cues across word and sentential temporal scales correlated with the behavioral ability to comprehend prosodic structures within and across groups? To achieve this, we will record magneto-encephalography data in large sample of adults with high-functioning ASD and in a matched TD control group. Participants will listen to natural speech dialogues in quiet as well as in a noisy auditory background, where comprehension is more challenging and subtle deficits might be detected more easily. We will then use state-of-the-art computational methods to analyze neural representations of prosodic patterns in the speech signal. We will extract neural markers for the neural tracking of amplitude envelope, relative pitch, and their fluctuations. Furthermore, we will separately model fluctuations in pitch and amplitude at word and phrasal levels, to assess contextual normalization of these features. Participants will also complete behavioral tasks on affective and linguistic prosody comprehension. This will allow us to identify associations between neural markers and behavioral performance, both at the group level and in analyses of individual variability.

These experiments will test the hypothesis that atypical prosody processing in ASD stems from auditory processing deficits of speech sounds. As such, they will provide insight into the neural representation of acoustic-prosodic content in speech in ASD and in TD populations. The results of this project will serve as the starting point for a larger-scale project on auditory processing of speech sounds in ASD across age groups. In the long run the results of this project can also be used to link between behavioral/systemic neural and genetic phenotypes in ASD and to provide EEG-based diagnostic metrics.