Brain Activity Underlying the Recognition of Vocal Affect Bursts in Autism Spectrum Disorder

Background: Functional MRI research has shown that individuals with Autism Spectrum Disorder (ASD) demonstrate atypical neural responses while processing emotional prosody in speech. Specifically, enhanced activation in brain regions for attention and executive function has been reported in ASD, suggesting a greater reliance on cognitive resources to process emotional cues in the voice. Previous studies have not utilised nonlinguistic emotional stimuli or vocal affect bursts (e.g., laughter, screams), although these expressions represent purer markers of affective states that do not contain the linguistic and semantic components of prosodic speech. Further, most studies have employed implicit tasks and ratings of emotional intensity or valence to elicit brain activation, instead of using tasks that explicitly assess the identification of emotions in vocal stimuli.

Objectives: We aimed to examine brain activity associated with the recognition of vocal affect bursts in ASD using a forced-choice discrimination task.

Methods: We assessed 16 participants with high-functioning Autism and 16 healthy controls (12 males per group) matched on age (M_ASD =23.00, M_Controls=23.19) and Full-Scale IQ (M_ASD=112.56, M_Controls=116.88). The in-scanner paradigm included blocks of tasks assessing emotion recognition (experimental condition) and gender recognition (baseline condition). On the emotion task, participants listened to vocal affect bursts representing the six basic emotions and identified the correct emotion from two options. On the gender task, participants listened to neutral voices and identified whether the speaker was male or female. We specified an “emotion > gender” contrast to isolate activity related to the cognitive and affective processes that underlie vocal emotion recognition.

Results: Analysis of in-scanner task performance revealed no significant difference in emotion recognition accuracy between both groups, suggesting that the ASD group reliably performed the task. Despite this, whole-brain, fixed effects analyses revealed that the ASD group demonstrated significantly higher activation across several brain regions at a cluster-corrected threshold. These included regions sensitive to emotional voices, such as the left inferior frontal gyrus and superior temporal sulcus, which belong to the mirror neuron system (MNS). The ASD group also showed more activation in structures for autobiographical memory retrieval (e.g., bilateral hippocampus, precuneus, posterior cingulate cortex) and visual processing (e.g., bilateral occipital gyri, fusiform gyri). Other regions of enhanced activation included the bilateral angular gyri and cerebellum. No region with significantly lower activation was found in the ASD group.

Conclusions: Compared to healthy controls, participants with ASD required greater activation across multiple brain regions to successfully perform the emotion task. Our findings suggest that the ASD group used additional cognitive effort to identify vocal affect bursts. It is possible that the ASD group relied more on compensatory strategies to perform the task, including cross-modal processing (e.g., mental imagery) and self-referential processing (e.g., autobiographical memory retrieval). Greater activation in the inferior frontal gyrus and superior temporal sulcus also indicates atypical MNS response to the recognition of vocal affect bursts in ASD. Our findings are consistent with fMRI research on emotional prosody perception in ASD, suggesting that enhanced brain activation to emotional voices is robust across different vocal stimuli and task demands.