Brain and behavioural correlates of emotional voice processing in autism and its broader phenotype

Abstract

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition characterised by impairments in two core domains of social communication deficits and restricted, repetitive behaviours or interests. A milder variant of autism-like characteristics, known as the Broader Autism Phenotype (BAP), is present at higher rates in relatives of individuals with ASD than the general population, indicating a genetic liability for ASD. There is a wealth of social cognitive research documenting theory of mind, empathising, and emotional face processing deficits in ASD and the BAP, which correspond to abnormal brain function in the “social brain”. However, less research has been conducted on the brain and behavioural correlates of emotional voice processing in ASD and the BAP, despite its significance for social cognition.

This thesis aimed to profile emotional voice processing abilities in ASD and the BAP in relatives and the general population, specifically relating to non-linguistic vocalisations known as vocal affect bursts (e.g., laughter, cries, screams). This thesis also aimed to identify the neurobiological substrates of emotional voice processing in individuals with ASD and relatives with the BAP. Three studies were performed to address these aims, each using the same purpose-built, web-based and/or functional MRI (fMRI) task to assess behavioural performance and elicit brain activation. Whole-brain activation elicited by the fMRI task was assessed using random- and fixed-effects analyses (RFX and FFX), which allowed inferences to be made on the population and sample levels, respectively.

Study 1 examined vocal affect burst recognition and its neurobiological correlates in individuals with high-functioning ASD (n = 16) compared to typically-developing controls (n = 16). The ASD group demonstrated a vocal emotion recognition deficit on the web-based task, where they misclassified neutral non-linguistic voices as expressing basic emotions at a higher rate than controls. RFX analyses revealed no significant group differences in brain activation, whereas FFX analyses revealed that the ASD group demonstrated higher activation relative to controls in widely distributed regions associated with emotional voice processing, executive function, memory, motor and somatosensory processing, and visual processing. These findings enable atypical brain activation to be inferred for this specific ASD cohort, but not for the wider ASD population. Although cohort-specific, such information may facilitate hypothesis generation for future investigations of neurobiological compensation in ASD.

Study 2 addressed the same research questions as Study 1 in relatives of individuals with ASD (n = 13) who were determined to have the BAP on clinical assessment. The BAP group demonstrated no vocal emotion recognition deficit relative to controls (n = 13) on the web-based task. Under FFX analyses alone (not RFX analyses), the BAP group demonstrated significantly higher activation in the left lateral occipital cortex relative to controls. Inferences about the FFX findings are limited to the specific BAP cohort assessed here, and cannot be extended to the wider population of ASD-relatives with the BAP. Nonetheless, these findings may inform new hypotheses exploring endophenotypes (i.e., intermediate phenotypes) of ASD, characterised by a similar expression of neurobiological compensation in ASD and the BAP.

Study 3 used the web-based task to assess the association between vocal affect burst recognition and the continuous distribution of BAP traits in the general population of individuals without a family history of ASD. It was hypothesised that lower recognition accuracy for the six basic emotions would correlate with higher self-ratings of BAP traits. In contrast to expectations, higher classification accuracy (and emotional intensity ratings) for angry voices correlated significantly with higher self-ratings of rigid BAP traits. The specific anger-rigid association indicates that enhanced auditory threat detection constitutes an aspect of the BAP in the general population. Further research is recommended to examine whether this relationship is mediated by underlying personality factors like neuroticism or trait anxiety.

Overall, different behavioural profiles of emotional voice processing abilities were observed in individuals with ASD (deficit—misclassifying neutral voices as being emotional), relatives with the BAP (no deficit—intact performance), and individuals from the general population with higher levels of BAP traits (advantage—enhanced sensitivity to angry voices). The neuroimaging findings of enhanced activation in the specific ASD and BAP cohorts assessed here may have implications for future research investigating the role of neurobiological compensation for emotional voice processing in their respective populations, potentially including the exploration of endophenotypes of ASD. Such studies would ideally include larger ASD and BAP (relatives) samples that enable the assessment of more homogenous, identifiable subgroups who may be susceptible to increased cognitive demands for emotional voice processing. This thesis extends research on social cognition within the voice modality in ASD and the BAP, and may have wider implications for understanding the genetic aetiology of social communication impairments in ASD.