Schedule of Events | Search Abstracts | Symposia | Invited Symposia | Poster Sessions | Data Blitz Sessions

Symposia

Leveraging social cognitive neuroscience tools to characterize heterogeneity in Autism Spectrum Disorder

Symposium Session 12: Tuesday, April 16, 2024, 1:30 – 3:30 pm EDT, Sheraton Hall EF

Chairs: Dorit Kliemann¹, Gabriela Rosenblau²; ¹The University of Iowa, ²The George Washington University
Presenters: Caroline J. Charpentier, Gabriela Rosenblau, Dorit Kliemann, Anila M. D’Mello

Social challenges constitute a core difficulty for many psychiatric conditions, most prominently for Autism Spectrum Disorder (ASD). Despite a long line of research characterizing social differences between autistic and neurotypical individuals, there is a lack of robust neurocognitive markers of social difficulties – particularly due to the vast phenotypic heterogeneity. This session will synthesize a deeper understanding of the cognitive neuroscience of ASD. We will showcase how theory-driven experiments and state-of-the-art methods can be used to characterize the heterogeneity in ASD. We will emphasize fine-grained, objective, and complementary approaches to measure behavior and cognitive strategies using computational modeling, and its underlying neurobiology using functional neuroimaging. The first two talks will focus on the behavioral and cognitive domains. The first talk will discuss how individual differences in autistic traits are associated with reduced social goal inference using computational models of observational learning in over 1000 participants. The second talk will introduce a social learning framework that quantifies how autistic and non-autistic groups incorporate prior knowledge for learning about others’ preferences. The last two talks will tie phenotypic differences in ASD to the underlying neurobiology. The third talk will discuss amygdala functional connectivity in ASD using a preregistered Bayesian approach leveraging large datasets. The fourth talk will focus on differences in the neural representation of social and nonsocial stimuli in ASD using multivariate neuroimaging analysis. The session will culminate in a discussion with the general CNS audience on the next big challenges in precisely specifying autism phenotypes including time for Q&A.

Presentations

Individual differences in autism-like traits are associated with reduced goal emulation in a computational model of observational learning

Caroline J. Charpentier^1,2; ¹Department of Psychology, University of Maryland, College Park, USA, ²Division of Humanities and Social Sciences, California Institute of Technology, USA

The ability to infer the goals and intentions of others is crucial for social interactions, and such social cognitive abilities are broadly distributed across individuals. Autism-like traits (i.e., traits associated with autism spectrum disorder (ASD)) have been associated with reduced social inference, yet the underlying computational principles and social cognitive processes are not well characterized. Here we tackle this gap by investigating inference during social learning through computational modeling, in two large samples of adult participants from the general population (N1=943, N2=352). Autism-like traits were extracted and isolated from other associated symptom dimensions through a factor analysis of the Social Responsiveness Scale. Participants completed an observational learning task that allowed quantifying the tradeoff between two social learning strategies: imitation (repeat the observed partner’s most recent action) and emulation (infer the observed partner’s goal). Autism-like traits were associated with reduced observational learning specifically through reduced emulation (but not imitation), revealing reduced social goal inference. This association held even when controlling for other model parameters (e.g., decision noise, heuristics), and was specifically related to social difficulties in autism but not social anxiety. The findings, replicated in two independent samples, provide a powerfully specific mechanistic hypothesis for social learning challenges in ASD, employing a computational psychiatry approach that could be applied to other disorders.

Social knowledge representations for learning in autistic adolescents

Gabriela Rosenblau^1,2; ¹Psychological and Brain Sciences, the George Washington University, Washington, DC, ²The Autism and Neurodevelopmental Disorders Institute, the George Washington University, Washington, DC

Social interaction difficulties are a key aspect of autism spectrum disorder (ASD), potentially stemming from less adaptive social learning strategies that hinder an accurate understanding of the mental states of interaction partners. Here, we examined social learning in a larger and more heterogeneous group of autistic adolescents. This study aimed to uncover differences in social learning in the autistic group and link these differences to clinical profiles, in particular to cognitive flexibility. We conducted an online study with a larger sample of autistic adolescents (N=217) and sex matched non-autistic young adults (N = 194). As expected, there were significant differences in self-preferences between the non-autistic and autistic groups. We investigated whether participants relied on knowledge of their respective group when learning about non autistic and autistic adolescents. Participants learned about individuals from either a non-autistic and or autistic adolescent groups. As hypothesized, prediction errors (PEs) in the social learning task of autistic adolescents were associated with reduced cognitive flexibility. Contrary to our initial expectation, however, both non-autistic and autistic groups had lower PEs when learning about the non-autistic mean profile. They also significantly reduced PEs over time when learning about autistic profiles. Consistent with our previous findings, autistic teenagers tended to rely more on their self-preferences during the learning task. In contrast, non-autistic adults relied more on the preferences of average non-autistic adolescents rather than on their own reference group. We are currently conducting computational modeling analyses that can uncover differences and heterogeneity in learning strategies of autistic adolescents.

Functional connectivity of the amygdala: testing three leading neurobiological hypotheses of ASD

Dorit Kliemann^1,2,3; ¹Department of Psychological and Brain Sciences, The University of Iowa, Iowa City, IA, USA, ²Department of Psychiatry, The University of Iowa, Iowa City, IA, USA, ³Iowa Neuroscience Institute, The University of Iowa, Iowa City, IA, USA

Three leading neurobiological hypotheses about Autism Spectrum Disorder (ASD) propose underconnectivity in the brain, atypical amygdala function, and higher variability between autistic and neurotypical participants. Replicability and generalization of prior neuroimaging findings have been limited due to data quality issues, statistical power, and analytic bias. Addressing these limitations, the current study investigated three hypotheses in a comprehensive pre-registered study using the ABIDE datasets, the largest sample of ASD resting-state functional magnetic resonance imaging (rs-fMRI) data (N=488 after exclusions; 212 with ASD). We analyzed magnitude, pattern similarity and variability of amygdala functional connectivity from two amygdala subdivisions (basolateral, BLA; corticocentro-medial, CCM) across a range of anatomical scales from whole-brain to specific regions and networks, using a Bayesian approach for hypothesis evaluation. We found some evidence for BLA underconnectivity to the whole brain in ASD (Bayes’ Factor; BF10 = 6.9), however, the effect was weaker for normalized connectivity and disappeared when only considering a subset of regions or using a different preprocessing approach. Amygdala connectivity patterns were similar between the groups across pipelines and strategies to define amygdala subregions (BF10 > 100). We did not find robustly increased between-subject variability in the autistic group. In sum, a comprehensive and preregistered set of analyses found no robust evidence for atypical amygdala functional connectivity in ASD. Future studies would benefit from an extended set of hypotheses in deeper individual data using multiple processing pipelines to increase generalizability of findings on amygdala functional connectivity in ASD.

Autistic adults show increased variability in cortical selectivity across social and non-social domains

Anila M. D’Mello^1,2,3; ¹Department of Psychiatry, University of Texas Southwestern Medical Center, Dallas, TX, ²Peter O’Donnell Jr. Brain Institute, University of Texas Southwestern Medical Center, Dallas, TX, ³Department of Psychology, University of Texas at Dallas, Richardson, TX

Functional neuroimaging analyses rely on assumptions regarding associations between a stimulus and its neural representation. In neurotypicals (NTs), certain regions consistently represent particular stimuli (e.g., fusiform face area, FFA, for faces). These stimulus-specific responses reliably co-localize across individuals. However, studies of autism spectrum disorder (ASD) often find reduced activation in these stimulus specific brain regions. One question is whether representations of particular stimulus categories are more heterogeneous in autistic individuals, resulting in atypical activation at the group level. We acquired fMRI data in n=31 ASD and n=29 NT adults who completed classical repetition suppression tasks to measure stimulus representations across social (faces, spoken words) and non-social (text, objects) categories. We assessed the degree to which patterns of significantly activated voxels in each domain overlapped across participants in each group. Across all domains, we found reduced overlap in significantly activated voxels in ASD as compared to NT. Increased Euclidean distance between each participant’s peak activation and that of every other participant was related to greater social communication challenges across groups and within ASD. Importantly, there were no group-wise magnitude differences, suggesting that results were not simply a product of lower activations in ASD. These data suggest that neural representations of social and nonsocial stimulus categories are more heterogenous in ASD, and that this is associated with social communication difficulty. These data speak to the importance of using methods that do not rely solely on group differences in magnitude, and interpreting results within the context of increased spatial variability.