Social and pragmatic communication difficulties are hallmarks of Autism Spectrum Disorder (ASD). Understanding the neural mechanisms underlying these difficulties is crucial for developing effective interventions. Everyday communication involves flexible processing of various social and linguistic signals within context, including verbal utterances, facial expressions, and body language. Interpreting these nonverbal cues is vital for pragmatic inference. Real-life situations often present challenging multimodal information from multiple individuals simultaneously, requiring attentional control and integration of various cues to understand intentions and meaning. ASD is characterized by deficits in social communication, repetitive behaviors, and restricted interests. Pragmatic communication deficits are a core feature of ASD, often persisting into adulthood. However, research on multi-level contextual processing in ASD, specifically the ability to integrate and manage multiple pieces of information concurrently, remains limited. Existing literature suggests that pragmatic impairment in ASD relates to difficulties in comprehending rapidly changing social cues, cognitive inflexibility, and impaired perception of facial expressions and gestures. Further, difficulties in predicting consequences of others' movements and adopting an allocentric perspective have also been linked to ASD. Previous research suggested a right-hemisphere dominance for pragmatic communication; however, current fMRI studies indicate the involvement of large-scale brain networks, particularly the salience network (insula and anterior cingulate cortex) and the default mode network (DMN). These networks are associated with salient event detection, mental state attribution, and social inference. Atypical connectivity and activation of these networks have been linked to ASD symptoms. While research on individual aspects of social and pragmatic processing exists, understanding brain function during observation of pragmatically challenging situations remains limited. Naturalistic movie stimuli provide a more ecologically valid approach than isolated pictures or sentences, better mirroring real-life communicative challenges faced by individuals with ASD. Moreover, movies enhance arousal and minimize head motion during fMRI scanning. Prior research shows hemodynamic changes and brain activity synchronization among viewers during movie watching, with higher test-retest reliability of network connectivity compared to resting state. The relative stability of network components in visual and somatomotor networks during movie viewing makes social brain network components particularly suitable for investigating pragmatic processing differences between groups. This study hypothesized that the social communication difficulties in ASD would be reflected in brain-level processing. Therefore, the primary aim was to compare brain network component responses to complex social episodes between young adults with ASD and neurotypical controls. If group differences were found, a secondary aim was to analyze the pragmatic and social interaction aspects of the events eliciting these differences.

The literature review extensively examined existing research on pragmatic communication in ASD, highlighting the challenges in understanding the neural basis of these difficulties. The review covers studies on the role of the right brain hemisphere, the involvement of large-scale brain networks like the salience network and default mode network, and the impact of atypical connectivity on social and pragmatic processing. It also discusses previous fMRI studies using various stimuli, comparing their advantages and disadvantages in relation to ecologically valid assessment of pragmatic processing in ASD. The literature underscores the importance of naturalistic stimuli, such as movies, for better reflecting real-life communicative complexities faced by individuals with ASD and improving the reliability of fMRI data.

This study employed a video-based fMRI approach. Nineteen young adults with ASD (diagnosed in childhood based on ICD-10 criteria, using ADI-R and ADOS) and 19 neurotypical controls participated. Cognitive abilities were assessed using Wechsler's intelligence scales; no significant differences were found between groups in general ability index, verbal comprehension, or perceptual reasoning, ensuring that differences observed wouldn't be attributed to intellectual disability. However, the Autism Quotient (AQ) questionnaire revealed a significant difference between groups, confirming the autistic traits in the ASD group. Ethical approval was obtained. The stimuli consisted of seven concatenated video clips from a Finnish TV soap opera, depicting naturalistic social interactions within a family. These clips were pre-assessed by university students to identify pragmatic features and content. FMRI data were acquired using a Siemens Skyra 3T scanner with echo-planar imaging. Standard FSL preprocessing steps were applied including motion correction, brain extraction, and spatial smoothing. There was no significant difference in head motion between groups. Independent Component Analysis (ICA) was used to identify 70 independent brain network components. Three components (IC4, IC15, IC27) were selected based on their spatial patterns and assumed role in social and pragmatic understanding, largely overlapping with the salience network (insula, anterior cingulate cortex) and partially with the anterior DMN. A permutation test showed no significant difference in spatial maps between groups for the selected components. Timecourses of the selected ICs were compared between groups using a permutation test in MATLAB, focusing on identifying timepoints of significant differences in BOLD response. Fisher's combined probability test and FDR correction were applied. Qualitative analysis of video clips at timepoints with significant differences was performed by a team of three researchers in clinical pragmatics, focusing on pragmatic communication content and social interaction aspects. The analysis involved observing the temporal relationships between brain responses and events, considering the latency of the BOLD signal. This analysis determined whether there were specific communication events, such as simultaneous verbal and nonverbal exchanges, that correlated with the observed differences.

Timecourse analysis of the selected brain network components (IC4, IC15, and IC27) revealed largely similar trends between ASD and control groups across the seven video clips. However, statistically significant differences emerged at specific timepoints within two clips (clip 2 and clip 4). These differences were identified at timepoint 60 (χ²(6) = 24.10, p < 0.001) in clip 2 and timepoints 104 (χ²(6) = 20.34, p = 0.002) and 105 (χ²(6) = 20.50, p = 0.002) in clip 4 after combining probabilities across the three components and applying FDR correction. No significant differences were found in the other five clips. Qualitative analysis linked these significant differences to scenes containing two overlapping communication events: one verbal and one nonverbal, requiring attention shifting between them. Clips 1, 3, 5, 6, and 7, which did not show significant group differences, lacked such parallel communication events. In clip 2, a power struggle between two women occurred concurrently with a daughter quietly gesturing to her dog to leave the room; in clip 4, a couple attempted a private moment while a sister narrated stories. The significant differences in brain activity corresponded temporally to the onset of these parallel communication events. The figures (Figure 2, 3, and 4) illustrate average activity (±SD) of each independent component (IC4, IC15, and IC27) in both groups across the seven video clips. The asterisks (*) denote timepoints with significant differences after combining probabilities and applying FDR correction. These differences were only present in clips 2 and 4 and are consistent with the qualitative analysis linking these to simultaneous verbal and non-verbal communication.

The findings suggest that while overall processing of social events is largely similar between young adults with ASD and neurotypical controls, differences emerge under conditions of high processing load, specifically when interpreting simultaneous verbal and nonverbal communication. The similarities in the majority of the video clips might be partially attributed to the mild ASD presentation of the participants. The observed brain response differences in clips 2 and 4 might indicate a heightened processing load for individuals with ASD when dealing with these complex, multi-modal communication scenarios. This aligns with previous research showing atypical brain synchronization during social interaction in ASD and the increased mental effort required to process high-communicative-value cues in this population. The observed differences are likely linked to atypical salience network functionality in ASD, affecting the ability to shift attention between competing communication modalities and integrate multimodal information. Specifically, abnormal temporal structure in the salience network could explain difficulties in switching salience attribution between events or modalities. Moreover, atypical activation in the insula (associated with social-emotive components in gesture comprehension) and anterior cingulate cortex (linked to tracking others' motivation) might underlie these findings. The results support previous behavioral findings of difficulties with perspective-shifting and cross-modal processing in ASD. The study suggests that pragmatic comprehension in ASD is particularly vulnerable in situations involving multiple individuals and overlapping communication events. Although significant differences were localized to specific moments, the largely similar brain activation across most stimulus segments may reflect the relatively high cognitive abilities of the participants. However, these individuals still experienced challenges in everyday social situations.

This fMRI study revealed largely similar processing of complex social situations between neurotypical and ASD groups, but significant differences emerged when observing simultaneous verbal and nonverbal communication. These differences likely reflect behavioral markers of ASD related to comprehending pragmatically challenging situations. Pragmatic challenges in ASD might only manifest when the processing load exceeds a threshold due to overlapping contextual cues. Identifying specific situations causing atypical social brain network function can inform intervention programs focusing on alleviating real-world challenges.

This study had a relatively small sample size, limiting the generalizability of the findings. The ASD diagnosis was based on childhood assessments; adult confirmation was lacking, although ASD is generally considered a lifelong condition. The video clips were framed to focus on main characters; more naturalistic settings might yield different results. Future studies should address these limitations, including larger sample sizes, adult diagnostic confirmations, and more ecologically valid testing environments (e.g., VR). Furthermore, using fast fMRI to improve temporal resolution of hemodynamic response could offer insights.