Emotional Language Processing in Individuals with and without Autism

Background:   A fundamental aspect of successful social interactions is the ability to accurately infer others’ intent through verbal communication, often including information related to the speaker’s feelings. Autism (ASD) is characterized by atypical language and social-emotional processing, including aberrant neural responses to socially-relevant stimuli like faces, eyes, and prosody. While findings are mixed regarding the nature of emotional language processing in autism, existing evidence suggests that comprehension skills are generally stronger than expressive abilities.

Objectives:   We investigated whether the process of making emotional inferences from spoken language in autism is associated with atypical neural activity in the brain regions subserving emotion processing, including anterior cingulate cortex (aCC), medial- and ventromedial prefrontal cortex (mPFC, vmPFC), and amygdala. A secondary question involved whether autistic individuals engage compensatory activity in areas typically involved in inferential language processing, including aSTS, pCC, dorsal medial prefrontal cortex (dmPFC), inferior frontal gyrus (IFG), or the superior temporal gyrus (STG) bilaterally. 

Methods:   Functional MRI data was collected from 15 ASD individuals and 16 controls (NT; 15-30yrs) on an emotional inference task (EIT). Activity changes were estimated during two epochs: hearing positive, negative, and neutral scenarios, and during responses to a related T/F question. Accuracy and response time were recorded. We used an accelerated multiband (MB) echo planar imaging (EPI) protocol that provides both high temporal and spatial resolution, thus enabling more accurate measurement of functional responses.

Results:   For both groups, behavioral and neural results show that participants’ reactions varied depending on sentence valence and congruency, but with differential effects. Behaviorally, both groups showed overall faster and more accurate responses to emotional relative to neutral scenarios, as well as for congruent compared to incongruent judgments. However, in comparison to the typicals, the ASD group showed greater difficulty evidenced as slower and less accurate responses for incongruent judgements, especially in the negative (RT: ASD 1.57sec, NT 1.23sec; accuracy: ASD 87.9%, NT 95.3%) and neutral (RT: ASD 1.66sec, NT 1.45sec; accuracy: ASD 86.3%, NT 93.2%) valence conditions. Brain responses of the ASD group also differed from controls. Our main findings reveal similar activity patterns between groups, but with significant between-group effects including increased activity in individuals with ASD in right TPJ and precuneus for emotional but not neutral scenarios. Brain-behavior correlations showed that greater recruitment of rSTS was related to better social abilities as measured by the “Empathy Quotient” (EQ) and “Reading the Mind in the Eyes” (MinE) in controls, but this effect was not seen in ASD. In the autism group, improved social skills from EQ, MinE and the Autism Quotient (AQ) predicted greater amygdala and mPFC activity. 

Conclusions:   Our results suggest that autistic individuals are able to correctly identify others’ feelings from language that lacks overt prosodic elements or emotionally-charged words. Furthermore, like neurotypical individuals, their neural activity patterns are enhanced to positive compared to neutral valence conditions. However, in contrast to typicals, the autism group showed overall greater task-related brain activity, possibly suggesting that the task is more effortful for them.