Abstract
Social anxiety (SA) frequently co-occurs with autism spectrum disorder (ASD), but it remains unclear whether SA in ASD reflects greater symptom severity or a distinct subtype. Across two complementary studies, we characterized the behavioral and neural correlates of SA in ASD using a naturalistic social navigation task measuring affiliative and power-related behaviors. In Study 1 (ASD = 575; controls = 357), autistic individuals with SA exhibited significantly reduced task-derived power behaviors (e.g., compliance with others’ requests) compared to those without SA—a pattern not observed in non-autistic individuals. In Study 2 (ASD = 72; controls = 72), individuals with both ASD and SA showed enlarged amygdala volumes relative to both ASD without SA and non-autistic individuals. These behavioral and neural differences found in Study 1 and Study 2 were not observed in ASD with co-occurring depression or generalized anxiety disorder. Moreover, in Study 2, larger amygdala volume in the ASD group was associated with reduced task-derived power behaviors, indicating a neural signature linking social-behavioral differences and anxiety-related features. Together, these findings point to a distinct ASD phenotype marked by social anxiety, altered interpersonal dynamics, and amygdala enlargement, with implications for diagnosis and tailored intervention.
Subject terms: Human behaviour, Predictive markers, Autism spectrum disorders
Introduction
Autism spectrum disorder (ASD) is a lifelong neurodevelopmental condition that affects approximately 1–2% of the population [1, 2]. Anxiety was noted as a feature in Kanner’s 1943 description of autism, alongside core symptoms such as social communication and interaction difficulties and restricted, repetitive behaviors [3]. Research estimates that 30% to 50% of individuals with ASD (mainly cognitively unimpaired) experience social anxiety [4–6]—four times the rate in the general population [7]. Assessing social anxiety in the context of autism is challenging because certain autistic behaviors—such as reduced social approach—can resemble the behavioral criteria used to diagnose social anxiety, making differentiation difficult [5, 8–10]. Despite its prevalence and considerable phenotypic overlap [11], the social-behavioral and neural features that distinguish social anxiety in ASD remain poorly understood, and research in autistic adults is particularly limited. Social anxiety symptoms may limit the benefits of social skills interventions [12, 13], which may increase the challenges of employment and social life. Understanding social anxiety in ASD is crucial for improving diagnosis, developing effective treatments, and enhancing support for this population.
The hallmark features of social anxiety disorder are fears of negative evaluation and avoidance of anxiety-provoking social situations [14]. While social fears are associated with social anxiety in both autistic and neurotypical (NT) individuals [4, 15, 16], the nature of these fears may differ. NT individuals with social anxiety often show intact or heightened perspective-taking yet underestimate their own competence [17]. In autism, however, social fears frequently stem from core social-cognitive difficulties—such as confusion in social situations or intolerance of unpredictability—rather than prototypical fears of rejection or embarrassment [9, 10, 18]. Although autistic individuals with social anxiety display greater social avoidance than those without [5], this avoidance may arise from mechanisms distinct from those in NT populations. Many autistic individuals value peer approval and desire social connection [19–21], yet difficulties interpreting social cues can make interactions distressing and promote motivational avoidance [18]. Consequently, social anxiety in autism is often inferred from behavior rather than explicitly reported internal fears [11], and may present differently than in NT individuals, contributing to limited treatment response [22, 23]. These differences highlight the need to identify social-behavioral phenotypes that disentangle social anxiety from autism-related social difficulties and more precisely capture how social anxiety manifests within autism.
To quantify the social-behavioral profile of social anxiety in autism, we employed a naturalistic experiment for assessment of implicit social behavior [24]. Our task was designed to capture social behavior tendencies on the dimensions of power and affiliation [25–27], which define a 2D social space—a cognitive map that supports flexible social behavior across abstract domains [28]. The affiliation-power model of social space has been used to study social anxiety, social motivation and depression (e.g., [29–32]). Our naturalistic task specifically captured high-level behaviors including social motivation, reciprocity, compliance, and communication—embedded within realistic decision-making scenarios, and the items corresponding to the two social dimensions have been validated and examined across multiple samples [24, 30, 32–36].
In our previous work, we found that autistic adults showed lower affiliative behaviors and reduced social distance (a combined measure of affiliation and power in the 2D social space) compared to NT individuals [35, 36]. These results indicate that the affiliation-power model could reflect meaningful social-behavioral phenotypes of ASD. Therefore, this model is suitable for examining whether social anxiety further reduces task-derived affiliation and/or power behaviors. Studies estimated that 49% to 80% of autistic adults experience interpersonal victimization [37, 38]. Difficulties recognizing manipulation or anticipating consequences may also lead to compliance in some contexts [39], suggesting that social anxiety in autism may diminish their ability to navigate interpersonal demands rather than increasing assertive or dominant behaviors.
Beyond the limited work on social-behavioral features, neural correlates of social anxiety in autistic adults have been scarcely examined. Across species, the amygdala—particularly its medial divisions—acts as a central hub of the social behavior network, transforming diverse social cues into adaptive affiliative, defensive, or competitive responses [40–42]. Rather than merely generating emotional valence, the amygdala assigns social meaning and orchestrates downstream behavioral mechanisms, making it a core integrator of social perception and social action [41, 42]. Converging cellular, molecular, and neuroimaging evidence further highlights the amygdala as a biologically vulnerable system in autism [43, 44]. Transcriptomic work shows that numerous ASD-susceptibility genes are expressed in the developing amygdala and exhibit cell-type–specific dysregulation [45]. Neurochemical studies similarly report reduced NAA/Cr ratios—an index of neuronal integrity—in the amygdala–hippocampal region, correlating with autistic symptoms of greater social and emotional impairment [46]. Postmortem findings reveal atypical developmental trajectories, with an early surplus of neurons followed by decline into adulthood [47], and longitudinal MRI work identifies enlarged basolateral nuclei and altered subnuclear growth associated with social-communication symptoms [48].
Furthermore, at the circuit level, specific amygdala microcircuits regulate social evaluation, social-threat learning, and anxiety in social contexts [49–51], suggesting a mechanism through which amygdala abnormalities may contribute to social anxiety–related difficulties in autism. This aligns with the alternative view that amygdala dysfunction in ASD may relate more to heightened fear and anxiety than to social-cognitive deficits per se [52]. Together, these discussions point to a structurally and functionally atypical amygdala in ASD, underscoring the need to characterize amygdala structure in autistic adults and determine how it relates to individual differences in social–behavioral features and social anxiety. If amygdala structure is particularly altered in autistic individuals with social anxiety—and if social anxiety in ASD presents with distinct social-behavioral features—then the amygdala may encode social behaviors that are especially salient to social anxiety in autism.
In the existing literature, most studies examining amygdala volume and anxiety in ASD focus on children or adolescents. Several reported enlarged amygdala volumes in autistic youth with anxiety compared to autistic peers without anxiety or NT controls [53–55], although smaller volumes [56] or null findings [57] have also been documented. Enlarged amygdala volume has also been observed in NT individuals with social anxiety [58]. Yet few studies have specifically investigated amygdala structure in relation to social anxiety in ASD [59], and, to our knowledge, none have examined this question in autistic adults or its links to affiliation-power behavioral profiles—a significant gap this work seeks to address.
Here, we examined self-reported social anxiety diagnoses as an index of vulnerability and investigated their association with task-derived social behaviors and amygdala volume in autistic adults. In Study 1, using a large online sample (ASD = 575, NT = 357), we tested whether task-derived affiliative and power behaviors differed between autistic adults with and without social anxiety and NT controls. We predicted that autistic individuals with social anxiety would show distinct social-behavioral profiles compared to both other groups. In Study 2, using an in-person neuroimaging (ASD = 72, NT = 72), we compared amygdala volume across autistic adults with and without social anxiety and NT controls. We hypothesized that amygdala volume would be enlarged in autistic individuals with social anxiety. Finally, we examined whether amygdala volume was associated with autism-related social behavior phenotypes captured by our task. All comparisons were conducted using independent participant samples across studies. Figure 1 provides an overview of the study design and research questions.
Fig. 1. Overview of research questions and study design.
(A) Participants engaged in a dynamic, role-playing game simulating naturalistic social scenarios. In Study 1, participants entered a new school, navigated different settings, joined a club, and settled into campus life. In Study 2, they explored a new town while securing employment and housing. Both versions were matched in the number and type of interactions and in the social roles of the illustrated characters (e.g., an acquaintance, a peer, or a potential employer). Across the narrative, participants encountered a series of “affiliation” and “power” trials designed to elicit behaviors reflecting social closeness or dominance/control. The figure depicts a power trial. In this scene, the participant is already seated for lunch when Chris, a recurring character, walks in and initiates a brief interaction. The two response options differ in assertiveness: Option 1 (“I am good here—grab a chair and join me”) reflects a power-assertive choice and is coded as +1, whereas Option 2 (you sit where he directed you) reflects a power-yielding choice and is coded as –1. This example illustrates how choices are scored along the power dimension rather than the literal seating logistics of the scene. See Methods section for additional example trials. Across the task, participants completed 30 affiliation trials and 30 power trials (6 trials per character). In the analysis, for each participant, we averaged affiliation and power trial across characters, respectively. After the task, participants rated their perceived relationships with each character on a 2-D space based on affiliation (friendship or intimacy) and power (dominance or control). (B) Study 1 recruited an online ASD sample with SA (n = 145) and without SA (n = 430), examining how task-derived power and affiliation behaviors predicted self-reported SA diagnoses (yes or no). Sensitivity analyses tested the robustness of findings by adding covariates (e.g., perceived social relations), removing participants with cognitive delays, reanalyzing data using different methods such as bootstrapping. We also conducted specificity tests to confirm that these social behavior changes were specific to SA rather than other frequently co-occurring conditions, such as depression or generalized anxiety. Results were compared with a matched neurotypical control sample, matched on sex, age and education to the subset of ASD individuals who did not have any cognitive delays. Study 2 recruited autistic participants for both neuroimaging scans and behavioral data collection (with SA n = 15, without SA n = 57), following similar procedures to examine the relationship between amygdala volume and self-reported SA diagnoses. Finally, we examined the relationship between social behaviors and amygdala volume.
Methods
Participants
Study 1 - online behavior sample
We recruited a large sample of ASD participants through Simons Powering Autism Research (SPARK) Research Match program, a project supported by Simons Foundation. All registered SPARK participants self-attested to having received professional diagnoses of ASD. The eligibility criteria were: (1) ages 18 to 30, (2) Social Communication Questionnaire [60] total score >9. Among 956 participants who consented, 675 individuals completed all surveys and the social task. The sample size after excluding 100 participants with an “ASD validity flag” (criteria include diagnosis rescinded, or diagnosis before 15 months of age, repetitive behavior scale-revised score <10, unknown diagnosis source, etc.) identified in the SPARK phenotype dataset was 575 (409 females). IQ was measured by cognitive tests from TestMyBrain (https://www.testmybrain.org). All participants provided written informed consent and received compensation for their participation.
Neurotypical (NT) controls were from the US and recruited from Prolific (www.prolific.com), a platform that helps researchers recruit participants for their online research. The exclusion criteria were: (1) neurodevelopmental diagnoses, including ASD, dyslexia, attention deficit disorder or attention deficit hyperactivity disorder, (2) mild cognitive impairment or dementia, (3) history of neurological concerns such as epilepsy, traumatic brain injury, Parkinson’s disease, seizures, or multiple sclerosis, (4) other chronic diseases such as diabetes, heart disease or stroke, and (5) first language is not English. IQ was measured by ICAR (The international cognitive ability resource https://icar-project.com).
Because no prior studies have examined the association between task-derived power/affiliative behaviors and social anxiety in ASD, an a priori power analysis was not feasible. We therefore relied on the large online dataset we collected. Autistic individuals without intellectual impairment were shown to experience more social anxiety [11]. Among the final ASD sample of 575, 200 had self-identified cognitive or language delays (such as intellectual disability, specific language impairment). Our sample showed a similar rate of cognitive delays among ASD individuals as reported in the latest systematic review [2]. We matched NT samples to ASD samples excluding individuals with self-reported cognitive or language delays. From a total sample size of 696 NT participants, we selected 357 participants who matched our ASD sample based on sex, age, and education using the “MatchIt” package in R. Among these 357 NT individuals, 25 (7.08%) declared SA diagnoses, consistent with the general US population [7].
Study 2 - onsite imaging sample
ASD participants were recruited through physical flyers around New York City, email listserv announcements, local research registry and word of mouth. The eligibility criteria were: (1) ages 18 to 50, (2) meeting DSM-5 criteria for ASD, (3) having an IQ over 60 (assessed onsite by the Wechsler Abbreviated Scale of Intelligence and Wechsler Intelligence Scale for Adults [61, 62]), (4) having no history of neurological concerns like epilepsy or traumatic brain injury, and (5) no substance or alcohol abuse disorders and no recreational drug use. ASD screening was conducted by licensed clinicians using the Autism Diagnostic Observation Schedule, 2nd edition (ADOS-2 [63]), supplemented with developmental and clinical history as needed, to inform DSM-5 criteria. We intentionally over-recruited females to ensure better representation of this underrepresented group in autism research.
NT participants were recruited through announcements posted on physical flyers around New York City and email listservs with the eligibility criteria of (1) age between 18 and 50, (2) IQ > 60 (assessed via The Wechsler Abbreviated Scale of Intelligence; WASI-II), (3) no 1st degree relatives with idiopathic ASD, (4) no history of neurological concerns (e.g., epilepsy, TBI), and (5) no recreational drug use 24 h prior to MRI visit.
Since no prior studies in autistic adults were available to inform a power estimate, we utilized our onsite sample, which has a sample size comparable to that of earlier studies conducted in children (e.g., [53–55, 59]). One hundred and sixteen autistic individuals with onsite clinically confirmed ASD and 92 NT individuals participated in the study. Among these participants, 101 ASD participants and 77 NT participants completed the social navigation task with valid data (if they completed at least 75% of the trials and demonstrated above-chance accuracy on the post-task memory test) and have completed demographic data. Twenty-two participants from ASD and 5 participants from NT group did not complete MRI scanning, and 7 participants’ data were removed in ASD group due to unusable structural scan. After applying these criteria, the final sample consisted of 72 sex-balanced (52.8% female) adults with ASD and 72 NT adults (62.5% female) for volumetric analysis. Our current sample included all data from the previous study that examined depression in ASD [32], along with 7 additional participants whose data were collected afterward. Amygdala volume has been previously analyzed in the partially overlapping sample [32] but was examined for other measures and research questions. All participants gave written informed consent and received compensation for their participation. Our studies were not preregistered.
The proportions of self-reported SA did not differ between the Study 1 online and Study 2 onsite ASD samples (p = 0.360, Two-Proportion Z-Test). However, compared with the online adult sample, the in-person adult sample was significantly older, had higher educational attainment, a greater proportion of males, lower self-rated autism symptoms, and lower self-rated social avoidance in both the ASD and NT groups (all ps < 0.01).
Assessment of social anxiety, autism symptoms and social behaviors
Self-reported social anxiety disorder diagnosis
Social anxiety was measured by a one-item self-reported clinician diagnosis: “Have you ever been diagnosed with any of the following psychiatric disorders?” Social anxiety disorder was one of the options along with other diagnoses such as major depressive disorder (MDD) and generalized anxiety disorder (GAD). The response to this question is either “yes” or “no.”
Self-reported social avoidance
Social avoidance was measured by Liebowitz Social Anxiety Scale, which assesses how social anxiety affects participants’ lives across various situations [64]. It asks, “How often do you avoid the following situations?” Examples include speaking up at a meeting, expressing disagreement or disapproval to someone you don’t know well, or going to a party. Participants rated their avoidance on a 4-point scale: never (0), occasionally (1), often (2), and usually (3). The self-report version of this scale compares well to the clinician-administered version [65] and shows good reliability and validity [66, 67]. It has also been shown to be reliable for use in autistic adults with social anxiety disorder [68].
Self-rated autism symptoms
Self-rated autism symptoms were measured by the Broad Autism Phenotype Questionnaire (BAPQ), a self-report questionnaire for adults. This questionnaire has succeeded in meeting both sensitivity and specificity requirements for detecting the broader autism phenotype [69]. Its robust psychometric properties [69, 70] and absence of ceiling effects [71] in individuals with ASD indicate that it performs effectively in both clinical populations [72] and the general population.
Social navigation task
We used the social navigation task developed to track implicit and naturalistic social behavior during simulated social interactions [24, 30, 32–36]. This task was intentionally designed to simulate affiliation and power-based social scenarios that integrate multiple autism phenotype dimensions—including social motivation, reciprocity, compliance, and communication—within realistic decision-making contexts (Fig. 1A). The task is naturalistic in that it presents a storyline in which participants assume the role of the protagonist and experience events from a first-person perspective. Participants were given no specific instructions other than to behave naturally and make choices as they would in real life. The term “naturalistic” reflects this intentionally unstructured design, allowing behavior to unfold in a manner that approximates everyday social decision-making. By validating individual items against known autism-relevant constructs and documenting robust ASD vs. NT control group differences, we demonstrated the construct specificity and ecological validity of the task (see Supplementary Materials for details).
Neuroimaging acquisition and processing
Structural MRI data was acquired for all participants on a 7 Tesla whole body scanner (Magnetom, Siemens Healthcare, Erlangen, Germany). A SC72CD gradient coil was used with a single coil transmit and a 32-channel head coil (Nova Medical, Wilmington, MA, USA). A T1-weighted MP2RAGE sequence was performed on each participant, with a 0.7 mm × 0.7 mm × 0.7 mm voxel resolution. Field of view (FOV) was 225 × 183, orientation of scan was coronal, repetition time (TR) was 6000 ms and echo time (TE) was 3.62 ms. A coronal-oblique T2-weighted turbo spin echo (T2-TSE) sequence was also obtained for all participants, with a 0.43 mm × 0.43 mm × 2.0 mm voxel resolution. FOV was 222 × 177, orientation of scan was coronal, TR was 9000 ms and TE was 69 ms.
Results included in this manuscript come from preprocessing performed using fMRIPrep 22.0.0 (RRID:SCR_016216) [73, 74], which is based on Nipype 1.8.3 (RRID:SCR_002502) [75, 76]. One T1-weighted (T1w) image was found within the input BIDS dataset. The T1-weighted (T1w) image was corrected for intensity non-uniformity (INU) with N4BiasFieldCorrection [77], distributed with ANTs 2.3.3 (RRID:SCR_004757) [78], and used as T1w-reference throughout the workflow. The T1w-reference was then skull-stripped with a Nipype implementation of the antsBrainExtraction.sh workflow (from ANTs), using OASIS30ANTs as target template. Volume-based spatial normalization to two standard spaces (MNI152NLin2009cAsym, MNI152NLin6Asym) was performed through nonlinear registration with antsRegistration (ANTs 2.3.3), using brain-extracted versions of both T1w reference and the T1w template. The following templates were selected for spatial normalization: ICBM 152 Nonlinear Asymmetrical template version 2009c [79] [RRID:SCR_008796; TemplateFlow ID: MNI152NLin2009cAsym], FSL’s MNI ICBM 152 non-linear 6th Generation Asymmetric Average Brain Stereotaxic Registration Model [80] [RRID:SCR_002823; TemplateFlow ID: MNI152NLin6Asym]. FreeSurfer (http://surfer.nmr.mgh.harvard.edu) automated segmentation of the volumes was used to extract amygdala volume. We averaged the bilateral amygdala volumes as the amygdala volume measure for the subsequent statistical analysis.
Statistical analyses
Study 1: social behaviors in autism and social anxiety
To examine whether autistic individuals with SA exhibit distinct social behaviors, we conducted a logistic regression predicting SA diagnosis (yes vs. no) using task-derived power and affiliation behaviors as independent variables. Covariates included sex, age, SES, self-reported cognitive delays, and IQ. For sensitivity analyses, we added perceived affiliation and power, as well as self-reported MDD and GAD as additional covariates. We repeated the analysis in a subset excluding individuals with cognitive delays. To assess the robustness of our findings, we implemented a nonparametric bootstrapping procedure (1000 resamples) to obtain an empirical sampling distribution and percentile-based confidence intervals. To test specificity, we substituted SA with MDD or GAD as the dependent variable while keeping independent variables constant.
To compare findings with the NT group, we matched NT participants to ASD participants without cognitive delays on sex, age, and education. As exploratory tests, we ran two regression models predicting task-derived power and affiliation behaviors, with interaction terms for Group (ASD vs. NT) and SA (yes vs. no), controlling for sex and age. P values of the interaction were FDR-corrected for multiple comparisons. These models were repeated for perceived affiliation and power behaviors. Moreover, to explore if the results were driven by specific game character, we additionally added game character in the interaction model as a 3-way interaction.
Study 2: amygdala volume, social anxiety and social behaviors in autism
To assess whether autistic individuals with SA exhibit distinct amygdala volume differences, we conducted a logistic regression predicting SA (yes vs. no) using amygdala volume as an independent variable. Covariates included sex, age, SES, and IQ. Sensitivity analyses controlled for intracranial volume (ICV), ACC volume (since our previous research identified ACC differences in ASD with co-occurring MDD [32], we included the ACC as a control region for comparison), and self-reported MDD and GAD. A nonparametric bootstrapping procedure (1000 resamples) was applied for robustness testing. To test specificity, we substituted SA with MDD or GAD while keeping independent variables unchanged.
To compare ASD and NT groups, we used a NT sample that was not statistically different from the ASD sample on sex and age. A regression model predicted amygdala volume based on Group (ASD with SA, ASD without SA, and NT without SA), controlling for sex, age, education, and ICV. As an exploratory analysis, we examined differences in NT individuals with and without SA by including an interaction term for Group (ASD vs. TD) and SA (yes vs. no). Furthermore, in the ASD group, we tested whether amygdala volume was predicted by task-derived power and affiliation behaviors, controlling for sex, age, education, IQ.
All statistical tests used were two-sided and report test statistics, standardized beta coefficients (where applicable), and adjusted p-values. For interaction terms, estimated marginal means were computed using the “emmeans” package in R. Post-hoc pairwise comparisons were conducted using Tukey’s method to control for family-wise error. Data and code for this study are available at https://osf.io/2gzxw/?view_only=e0eda95675f84c5798e8460eb9eac01e.
Results
Study 1: is social anxiety in autism associated with unique social behaviors?
In study 1, we analyzed data from 575 young adults (71% females, Mage = 25.38, S.D. age = 3.31) who disclosed professional autism diagnoses. Of these, 145 (25.22%) also reported SA diagnoses. Demographic information and statistics of main research variables are provided in Supplementary Table S1. Individuals’ attestations of having SA diagnoses were further supported by significantly higher rates of self-reported social avoidance behavior (Supplementary Fig. S1).
Consistent with our hypothesis that autism with SA would show distinct task-derived social behaviors compared to autism without SA, our analysis found that a higher likelihood of self-reported SA was significantly associated with lower task-derived power behaviors, such as following commands or not in control of the social situation (χ2 = 4.30, β = -0.21, S.E. = 0.10, 95% CI = [-0.42, -0.01], p = 0.038). Affiliation behaviors were not related to SA (χ2 = 0.46, β = -0.07, S.E. = 0.10, 95% CI = [-0.26, 0.13], p = 0.455). These results were based on statistical model 1 shown in Table 1. We conducted a few validation analyses. The first set involved sensitivity tests. We repeated the analysis for those without cognitive delays (model 2 in Table 1) because this subgroup is more prone to anxiety [81] and we expected the effect of task-derived social behaviors to become stronger, which was confirmed. Then in the original model, we added covariates of perceived social relations with game characters that were measured right after the task (model 3 in Table 1), which would remove any effects of subjective perception of social interaction. Results showed that subjective perception was not related to SA and the effect of task-derived power behaviors remained. The last set of sensitivity test was to covary MDD and GAD which were highly correlated with SA (model 4 in Table 1). This way we could examine if task-derived social behaviors would capture any uniqueness of SA. Results showed that although MDD and GAD significantly predicted SA, the effect of task-derived power behaviors remained. The second category of validation tests were specificity tests. We repeated the analyses for predicting MDD (model 5 in Table 1) and GAD (model 6 in Table 1) using social behaviors in order to test if these social behaviors were also related to other forms of co-occurring conditions in ASD. The tests did not show any evidence that task-derived social behaviors were associated with MDD or GAD.
Table 1.
Standardized beta coefficients (standard errors) of logistic regression models predicting self-reported diagnoses (two levels, YES vs. NO) by social behaviors and demographic covariates as well as sensitivity and specificity tests in the online autism sample (n = 575).
| Original model | Sensitivity tests | Specificity tests | |||||
|---|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | ||
| Dependent variable |
SA YES n = 145 NO n = 430 |
SA YES n = 79 NO n = 278 remove those with cognitive delays |
SA YES n = 145 NO n = 430 |
SA YES n = 145 NO n = 430 |
MDD YES n = 371 NO n = 204 |
GAD YES n = 445 NO n = 130 |
|
| Independent variables | Power behavior |
-0.21 (0.10)* p = 0.038 |
-0.32 (0.14)* p = 0.020 |
-0.21 (0.11)* p = 0.044 |
-0.21 (0.11)* p = 0.046 |
-0.13 (0.10) p = 0.190 |
0.03 (0.11) p = 0.776 |
| Affiliation behavior |
-0.07 (0.10) p = 0.455 |
-0.07 (0.13) p = 0.569 |
-0.08 (0.10) p = 0.425 |
-0.06 (0.10) p = 0.582 |
0.01 (0.10) p = 0.959 |
-0.09 (0.11) p = 0.410 |
|
| Sex |
0.58 (0.24)* p = 0.012 |
0.78 (0.37)* p = 0.024 |
0.58 (0.24)* p = 0.013 |
0.19 (0.25) p = 0.448 |
1.21 (0.20)*** p < 0.001 |
1.42 (0.22)*** p < 0.001 |
|
| Age |
-0.02 (0.03) p = 0.551 |
0.02 (0.04) p = 0.588 |
-0.02 (0.03) p = 0.563 |
-0.03 (0.03) p = 0.413 |
0.05 (0.03) p = 0.124 |
0.02 (0.03) p = 0.540 |
|
| Education |
-0.12 (0.06)* p = 0.041 |
-0.20 (0.08)* p = 0.017 |
-0.12 (0.06)* p = 0.042 |
-0.06 (0.06) p = 0.335 |
-0.25 (0.06)*** p < 0.001 |
-0.20 (0.06)** p = 0.001 |
|
| IQ |
-0.03 (0.02) p = 0.076 |
-0.02 (0.03) p = 0.503 |
-0.03 (0.02) p = 0.074 |
-0.04 (0.02)* p = 0.035 |
0.03 (0.02) p = 0.125 |
0.01 (0.02) p = 0.476 |
|
| Cognitive delays |
-0.42 (0.21)* p = 0.043 |
-0.42 (0.21)* p = 0.042 |
-0.43 (0.21)* p = 0.044 |
-0.01 (0.20) p = 0.947 |
-0.18 (0.23) p = 0.439 |
||
| Perceived power |
0.07 (0.49) p = 0.886 |
||||||
| Perceived affiliation |
0.06 (0.31) p = 0.841 |
||||||
| MDD |
0.57 (0.28)* p = 0.038 |
||||||
| GAD |
1.47 (0.43) p = 0.0001 |
||||||
Cells left blank indicate variables that were not included in the corresponding statistical model. For example, compared with the original model, the first sensitivity analysis excluded the variable “cognitive delays.” Both models excluded all variables listed below “cognitive delays.”
***p < 0.001, **p < 0.01, *p < 0.05.
The last validation test was robustness tests using other models or methods (Supplementary Material 1.3). We reversed the independent variable and the dependent variable and also conducted bootstrapping analyses, which all showed consistent results with our conclusion. In addition, our results did not differ by sex since there was no significant sex interaction with power behaviors on the likelihood of self-reported SA (p = 0.825). Together, these results consistently show that autistic individuals with SA exhibited more acquiescent behaviors than those without SA, and acquiescent behaviors were specifically related to SA but not any other co-occurring and overlapping conditions such as MDD or GAD.
How do social behaviors in individuals with SA differ between individuals with and without ASD? To explore this, we compared our findings with a matched neurotypical (NT) control group (see Supplementary Table S2 for demographic comparisons). Overall, the ASD group showed significantly lower task-derived affiliation and power behaviors compared to NT (Fig. 2). Additionally, the interaction between group and SA was significant for both affiliation (p = 0.028) and power behaviors (p = 0.028). This suggests that the impact of SA on task-derived social behaviors differed between ASD and NT. Post-hoc analyses revealed that ASD individuals with SA had the lowest power behaviors among all groups (ASD without SA, NT with and without SA). By contrast, in NT group, SA did not significantly relate to both dimensions of social behaviors, possibly due to limited statistical power (small sample size in NT with SA group). There was no evidence for influence of specific game characters, as indicated by a non-significant three-way interaction between group, SA, and game characters (p > 0.33).
Fig. 2. Relation between social behaviors and self-reported SA diagnoses in ASD and NT groups.
Sample sizes and means for each group: ASD with SA n = 79, affiliation mean = 0.08, power mean = -0.19; ASD without SA n = 278, affiliation mean = 0.14, power mean = -0.12; NT with SA n = 25, affiliation mean = 0.43, power mean = 0.02; NT without SA n = 332, affiliation mean = 0.32, power mean = -0.02. (A) Affiliation Behavior Model: A regression model was used to predict task-derived affiliation behaviors, including an interaction term for group (ASD vs. NT) and SA (yes vs. no), with sex, age, and education as covariates. The ASD group showed significantly lower task-derived affiliation behaviors than the NT group (F = 96.10, β = 0.19, S.E. = 0.02, p < 0.0001). There was also a significant interaction between group and SA (p = 0.028). (B) Power Behavior Model: A regression model was used to predict task-derived power behaviors, following the same structure as the affiliation behavior model. There was a significant main effect of group, indicating that the ASD group exhibited lower task-derived power behaviors than the NT group (F = 50.09, β = -0.10, S.E. = 0.02, p < 0.0001). Additionally, a significant interaction between group and SA was observed (p = 0.028). Note: task-derived affiliation behavior scores greater than 0 indicate a preference for affiliative choices during the task; task-derived power behavior scores of 0 indicate balanced decision-making, where participants equally distributed power between themselves and game characters. Negative values reflect a tendency to give power to others more frequently. Error bars represent 95% confidence intervals. ***p < 0.001, **p < 0.01, *p < 0.05.
Unlike task-derived social behavior, perceived social relations with game characters that was measured after the task (see Fig. 1) did not significantly differ in individuals with SA compared to those without SA in both groups (Fig. S2). Together, these findings suggest that while observed social behaviors differed between SA in ASD and NT, their perception of social relationships remained similar.
Study 2: are social anxiety and social behavior in autism associated with amygdala volume?
Social anxiety
In study 2, we analyzed data from 72 individuals who completed both the behavioral task and 7 T MRI, and had clinically confirmed ASD (52.8% females, Mage = 26.85, S.D. age = 7.74, IQ = 67 to 140). Of these, 15 (21%) also disclosed SA diagnoses. Demographic information and statistics of main research variables are provided in Supplementary Table S3. Individuals’ attestation of having SA diagnoses in Study 2 was also further supported by significantly higher rates of self-reported social avoidance behavior (Supplementary Fig. S3).
Consistent with our hypothesis that amygdala volume would be linked to SA in autism, we found that self-reported SA was significantly associated with larger amygdala volume (χ2 = 5.55, β = 0.88, S.E. = 0.41, 95% CI = [0.07, 1.69], p = 0.018; Fig. 3). These results were based on statistical model 1 shown in Table 2. We conducted several validation analyses to ensure the robustness of our findings. The first set involved sensitivity tests. We began by including ICV as a covariate in the original model (model 2 in Table 2) to distinguish absolute from relative differences in amygdala volume. The results showed that the relationship between amygdala volume and SA remained significant controlling for ICV. Next, we added ACC volume as a control region (model 3 in Table 2), given our prior work showing ACC alterations in ASD with co-occurring depression [32]. This addition did not change the above finding. We then examined the potential overlap with other affective conditions by covarying for MDD and GAD (model 4 in Table 2), both of which could be highly correlated with SA, to test whether amygdala volume uniquely related to SA. Results showed that although MDD significantly predicted SA, the effect of amygdala volume remained. The second category of validation tests were specificity tests. We repeated the analyses for predicting MDD (model 5 in Table 2) and GAD (model 6 in Table 2) using amygdala volume in order to test if amygdala volume was also related to other forms of co-occurring conditions in ASD. The tests did not show any evidence that the amygdala volume was associated with MDD or GAD.
Fig. 3. Relation between amygdala volume and social anxiety and social power behaviors in autism.
(A) Amygdala atlas. (B) Relation between amygdala volume (averaged left and right) and self-reported SA diagnoses in ASD and NT groups. Sample sizes and means for each group: ASD with SA n = 15, amygdala volume mean = 1253; ASD without SA n = 57, amygdala volume mean = 1133; NT control n = 72, amygdala volume mean = 1130. A regression model was conducted predicting amygdala volume from Group (ASD with SA, ASD without SA, and NT) and other covariates including sex, age and education. Post-hoc analysis showed that larger amygdala in ASD with SA than ASD without SA (t = 2.60, p = 0.028) and NT (t = 2.86, p = 0.014). Amygdala volume did not differ between ASD without SA from NT individuals (t = 0.53, p = 0.857). (C) Relation between amygdala volume and task-derived power behaviors in the autism sample. A regression model was conducted predicting amygdala volume from task-derived power and affiliation behaviors, and other covariates including sex, age and education. It shows that larger amygdala volume corresponds to acquiescent behaviors. Note: Error bars and the shaded area represent 95% confidence intervals. ***p < 0.001, **p < 0.01, *p < 0.05.
Table 2.
Standardized beta coefficients (standard errors) of logistic regression models predicting self-reported diagnoses (two levels, YES vs. NO) by amygdala volume and demographic covariates as well as sensitivity and specificity tests in the onsite autism sample (n = 72).
| Original model | Sensitivity tests | Specificity tests | |||||
|---|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | ||
| Dependent variable |
SA YES n = 15 NO n = 57 |
SA YES n = 15 NO n = 57 |
SA YES n = 15 NO n = 57 |
SA YES n = 15 NO n = 57 |
MDD YES n = 32 NO n = 40 |
GAD YES n = 34 NO n = 38 |
|
| Independent variables | Amygdala volume |
0.88 (0.41)* p = 0.018 |
1.20 (0.49)** p = 0.005 |
1.22 (0.53)** p = 0.009 |
0.88 (0.44)* p = 0.033 |
0.26 (0.30) p = 0.369 |
0.40 (0.31) p = 0.180 |
| Sex |
0.55 (0.69) p = 0.425 |
-0.29 (0.88) p = 0.744 |
-0.30 (0.89) p = 0.735 |
-0.09 (0.81) p = 0.907 |
1.53 (0.60)** p = 0.007 |
1.56 (0.58)** p = 0.005 |
|
| Age |
-0.01 (0.05) p = 0.756 |
-0.03 (0.05) p = 0.575 |
-0.03 (0.05) p = 0.571 |
-0.04 (0.05) p = 0.475 |
0.08 (0.04) p = 0.064 |
-0.02 (0.04) p = 0.594 |
|
| Education |
-0.42 (0.40) p = 0.284 |
-0.39 (0.43) p = 0.362 |
-0.38 (0.43) p = 0.363 |
-0.21 (0.44) p = 0.636 |
-0.56 (0.34) p = 0.093 |
-0.23 (0.33) p = 0.491 |
|
| IQ |
-0.01 (0.02) p = 0.797 |
0.00 (0.02) p = 0.947 |
0.00 (0.02) p = 0.944 |
-0.02 (0.02) p = 0.420 |
0.03 (0.02) p = 0.127 |
0.04 (0.02)* p = 0.032 |
|
| ICV |
-0.87 (0.57) p = 0.101 |
-0.87 (0.57) p = 0.106 |
|||||
| ACC volume |
-0.04 (0.38) p = 0.907 |
||||||
| MDD |
1.92 (0.84)* p = 0.013 |
||||||
| GAD |
0.30 (0.81) p = 0.715 |
ICV intracranial volume.
Cells left blank indicate variables that were not included in the corresponding statistical model. For example, compared with the original model, the first sensitivity analysis includes the variable “ICV” whereas both models exclude all variables listed below ICV. Amygdala volume was calculated as the average of left and right amygdala volumes. SES was calculated by the sum of normalized scores of education and income (treated as ordinal variable). **p < 0.01, *p < 0.05.
Finally, we performed robustness checks using alternative modeling strategies (see Supplementary Material 2.3). These included reversing the independent and dependent variables and applying bootstrapping analyses. Across all approaches, the results remained consistent, further supporting the reliability of our conclusions. In addition, although self-report SA diagnoses were related to self-reported social avoidance behaviors, self-reported social avoidance was not related to amygdala volume (F = 1.31, β = -0.14, S.E. = 0.12, p = 0.245), highlighting potential distinct features of social anxiety in autism other than social avoidance. Lastly, our results did not differ by sex since there was no significant sex interaction with amygdala volume on the likelihood of self-reported SA (p = 0.794). These results consistently show that autistic individuals with SA had larger amygdala volume than those without SA, and larger amygdala volume was specifically related to SA but not any other co-occurring and overlapping conditions such as MDD or GAD.
We next asked whether amygdala volume in autistic individuals with and without SA differs from NT individuals (n = 72, 62.5% females, Mage = 25.89, S.D. age = 6.44, IQ = 76 to 135, see Table S4 for demographic comparisons with ASD group). Results showed that autistic individuals with co-occurring SA had significantly larger amygdala volumes compared to autistic individuals without SA and the NT group (F = 4.23, p = 0.016). Among these NT individuals, 9 reported SA diagnoses, and we separated them as exploratory comparison. We found that autistic individuals with co-occurring SA had significantly larger amygdala volumes compared to NT individuals without SA (t = 3.16, p = 0.010) and amygdala volume was not different for ASD and NT individuals who did not have SA (t = 1.00, p = 0.748). Amygdala volume was also not different for NT individuals with and without SA (t = 1.88, p = 0.242), likely due to limited sample size in NT with SA. All p values were adjusted by Tukey method for comparing a family of 4 estimates. Figure 3B illustrates the comparisons of the 4 groups.
Task-derived social behavior
Thus far, we have shown that self-reported SA was uniquely linked to lower task-derived power behaviors (Study 1) and larger amygdala volume (Study 2) in ASD. We therefore asked whether amygdala volume is associated with task-derived social behaviors in ASD.
Consistent with this prediction, we found that larger amygdala volume was associated with lower task-derived power behaviors (F = 6.47, β = -0.35, S.E. = 0.14, 95% CI = [-0.63, -0.08], p = 0.013), but not task-derived affiliation behaviors (F = 3.13, β = -0.23, S.E. = 0.13, 95% CI = [-0.49, 0.03], p = 0.081). These results remained significant after controlling for ICV, ACC and psychiatric medication use. Furthermore, adding self-reported SA, the beta coefficient of task-derived power behaviors in predicting amygdala volume was strengthened by 15% (F = 9.38, β = -0.41, S.E. = 0.13, 95% CI = [-0.67, -0.14], p = 0.003) but the beta coefficient of affiliation behaviors only changed by 3% (F = 3.68, β = -0.24, S.E. = 0.12, 95% CI = [-0.48, 0.01], p = 0.060). The relation between amygdala volume and social anxiety was also significant (p = 0.005). We performed a likelihood ratio test (F-test for linear models) and found that adding social anxiety in the model significantly improves the model fit (F = 8.51, p = 0.005, ΔR² = 0.09). These results suggest that lower task-derived power behaviors and SA uniquely contribute to variations in amygdala volume, indicating distinct underlying mechanisms for each.
Discussion
Social anxiety in autism has been hypothesized to be closely tied to core ASD symptoms [5, 9, 10, 82–84]. However, it’s unclear whether it manifests unique behavioral or biological correlates. Our findings show that autistic individuals with co-occurring social anxiety exhibited significantly lower task-derived power behaviors compared to both autistic individuals without social anxiety and NT individuals. Notably, this pattern was only observed as behavioral manifestations during the task and did not align with their self-perceptions of social interaction during the task. Furthermore, although social anxiety is also highly comorbid with MDD and GAD in autism, lower task-derived power behaviors were not associated with co-occurring MDD or GAD, suggesting that autism with social anxiety might present a distinct behavioral profile of social acquiescence. At the neural level, autistic individuals with social anxiety exhibited enlarged amygdalae compared to both those without social anxiety and the NT control group. Again, this enlargement was specific to social anxiety, as it was not observed in autism with MDD or GAD. Moreover, larger amygdala volume was linked to lower task-derived power behaviors in ASD, both in absolute size and relative to whole-brain volume. Overall, these findings highlight distinct behavioral and neural features of autism with social anxiety compared to autism without social anxiety as well as NT controls, emphasizing the need for more precise diagnostics and tailored interventions for this subgroup.
Autistic individuals with social anxiety tend to exhibit lower task-derived power behaviors possibly due to difficulties navigating social ambiguity or uncertainty. Specifically, they may struggle with interpreting power dynamics, consistent with prior research showing that autistic individuals process social power relationships more slowly than NT peers, suggesting a more deliberate and explicit processing style [85]. Compliance with others’ requests may stem from a desire to avoid conflict, naïveté, or fear of negative consequences [39]. Supporting this interpretation, we found that perceived social power during the interaction task did not differ between ASD and NT participants, regardless of social anxiety status, suggesting that social anxiety-related task-derived behavioral differences in power were not driven by subjective perceptions of power. In contrast, perceived affiliation was higher in the NT group than in the ASD group, indicating that autistic individuals may have reduced motivation to affiliate as NT people do (Fig. S2).
Lower task-derived social power behaviors might help explain why autistic adults were perceived as warmer but less competent than NT individuals [86]. While acquiescence may serve as a coping mechanism for forming relationships, over-reliance on it may lead to distress, increased workload, or even exploitation and betrayal. As explained by the dialectical misattunement hypothesis, autism may not be a disordered function of individual brains but a mismatch of interpersonal dynamics [87]. Over time, these accumulated negative experiences may contribute to the development or worsening of social anxiety in autism.
We also explored group differences involving the small number of NT individuals with social anxiety (percentage in our sample was consistent with population prevalence). Our findings suggest that social anxiety amplifies diminished task-derived power behaviors in autistic individuals and further widens the behavioral gap between ASD and NT participants (Fig. 2A). This pattern supports the notion that social anxiety in ASD may present differently in ASD than in NT individuals. Note that these results should be interpreted with caution, as they are based on a small subsample of participants with social anxiety.
Moreover, we found no significant differences in task-derived affiliation or power behaviors between NT individuals with and without social anxiety. While this lack of differences may be due to limited statistical power given the small NT-social anxiety subgroup, prior studies have also shown inconsistent relationships between social anxiety and submissiveness in NT populations. For example, one study found that social anxiety was associated with fewer behaviors of social cooperativeness and dominance but not escape/avoidance behaviors [88], and another study found that social anxiety was related to increased agreeable behavior [89]. The possibility that social anxiety impacts social dynamics differently in NT individuals depending on the specific behavioral measures remains open and suggests a promising area for future research.
In the brain, we discovered that enlarged amygdala volume was associated with autism with social anxiety. The amygdala is related to safety detection [90], but individuals with autism may lack effective safety detection in social contexts [91]. Research until now has primarily focused on autistic children, yielding mixed results but primarily reporting larger amygdalae related to anxiety [53–55]. Due to differing developmental changes in amygdala size in ASD and NT controls [47, 54], it is important to examine amygdala structure in social anxiety among autistic adults. Our findings contribute to filling this gap, showing that larger amygdala volume is linked to self-reported social anxiety in autism. A recent meta-analysis found increased right amygdala volume in autism who had no cognitive disabilities compared with non-autistic individuals [92]. Since we found that enlarged amygdalae in autistic adults compared to NT individuals primarily occurred in those with social anxiety, this finding suggests that prior reports of increased amygdala volume in ASD might be driven by this specific subgroup. This indicates that co-occurring social anxiety might predict a heterogeneous pattern of neuropathology in autism.
Although prior meta-analytic work reports amygdala volume reductions in GAD and MDD [93–95], these effects are typically small and detectable only in very large samples. Consistent with this, we did not observe amygdala reductions associated with GAD or MDD in either ASD or NT participants, likely reflecting limited power in these subgroups rather than disagreement with the broader literature. By contrast, we found significantly enlarged amygdala volume specifically in autistic individuals with co-occurring social anxiety, an effect not observed in ASD alone or in ASD with other co-occurring internalizing disorders. Importantly, recent transdiagnostic meta-analyses emphasize that co-occurring conditions can yield distinct neural signatures that diverge from effects seen in single disorders [96], offering a framework for why amygdala enlargement emerged only in ASD with social anxiety.
We also found that larger amygdalae were associated with lower task-derived power behaviors in autism regardless of social anxiety. Because the amygdala is deeply involved in social behavior, and prior work has linked its size to variation in social network size [97] and tolerance for social inequiality [98], this novel association with power behavior adds an important dimension to the literature, particularly given that power behavior was uniquely reduced in autistic adults with social anxiety. When accounting for social anxiety as a covariate, the association between amygdala and power behaviors was further enhanced. These results suggest that despite the link between power behaviors and social anxiety (study 1), each of these factors statistically uniquely related to variations in amygdala volume. Prior work linked amygdala reactivity to social cues with anxiety symptoms in autistic individuals [99, 100] and there was also a study suggesting that amygdala hypoactivation in ASD may be masked by co-occurring anxiety, which indicated that its function cannot be fully understood within either social or anxiety domain alone [101]. In our study, the statistically enhanced association between task-derived power behaviors and amygdala volume when controlling for social anxiety aligns with previous findings on amygdala function, suggesting that social processing, social anxiety, and ASD might involve both shared and distinct neural pathways.
In Study 2, we did not detect the effect observed in Study 1 of lower task-derived power behaviors in ASD compared to the NT group (trending, p = 0.10). Based on the effect size from Study 1 (Cohen’s d = -0.36), a total sample size of at least 248 would be needed to achieve 80% power (α = 0.05) to detect a difference in power behaviors in autistic individuals with social anxiety compared with those without. This calculation assumes that social anxiety comprises 22% of the ASD sample (excluding individuals with cognitive disabilities). Although the effect size for task-derived power behaviors related to social anxiety is small, lower power behaviors were nevertheless positively associated with amygdala volume with a medium effect size (Cohen’s f2 = 0.19), suggesting that this behavioral effect may hint at broader underlying neurobiological differences.
Taken together, our findings lead us to a central question: is autism with social anxiety simply a more severe form of autism, or does it represent a distinct phenotype? Both Studies 1 and 2 showed that the ASD group exhibited significantly lower task-derived affiliation behaviors than the NT group. These results are consistent with our previous report using a separate online sample as well as a partially overlapping onsite sample [36] (see Methods). The current study offers novel insights by differentiating task-derived affiliation behaviors in ASD versus NT, and showing that at least in those without cognitive delay who comprised our participant sample - social anxiety may further amplify these group differences (Fig. 2B). However, we also found that task-derived affiliation behaviors were not associated with amygdala volume in ASD, despite prior findings from a partially overlapping on-site neuroimaging sample showing a link between amygdala volume and clinician-rated ASD symptoms [32]. This suggests that task-derived affiliation behaviors may not be as directly tied to the neural substrates of social anxiety as task-derived power behaviors are. Moreover, while reduced task-derived power behaviors in autistic individuals with social anxiety may reflect a more severe form of autism, amygdala volume specifically tracked social anxiety–related social difficulties since amygdala volume did not differ between ASD and NT individuals without social anxiety. In addition, our exploratory analysis (Supplementary Materials section 4), did not provide conclusive evidence that autism symptom severity accounts for social anxiety. These findings suggest that autism with social anxiety may not simply represent a more severe form of autism, but rather a distinct condition.
Our studies have several limitations. First, its cross-sectional design prevents us from determining causal relationships between social anxiety, task-derived power behaviors, and amygdala volume. Longitudinal research is needed to explore whether early-life social behaviors contribute to social anxiety or if social anxiety leads to reduced power behaviors. Tracking amygdala volume changes over time could also clarify whether task-derived power behaviors influence amygdala structure or vice versa. Second, our neuroimaging study had a relatively small sample size, particularly for the subgroup of autistic adults diagnosed with social anxiety. This was the first imaging study to examine amygdala volume in relation to social anxiety specifically in autistic adults—a group that has received limited attention in neuroimaging and mental health research—so our findings provide valuable initial data for this underexplored area. Because no prior studies in autistic adults were available to guide a power estimate, we targeted a sample size comparable to earlier studies conducted in children (e.g., [53–55, 59,101]). The relative homogeneity of our sample may partly explain why significant effects emerged despite its modest size. However, this also limits generalizability, as our participants had relatively high education and IQ levels. Moreover, we excluded individuals with intellectual disability when comparing with NT participants in Study 1, and social anxiety is particularly prevalent among autistic individuals without intellectual disability [102]. Nevertheless, our use of a sex-balanced adult sample is a strength, given the underrepresentation of females in autism research. The study was guided by a priori hypothesis and a multimodal design, and therefore provides initial, theory-informed evidence for distinct behavioral and neural features of social anxiety in autism. We look forward to future studies with larger, more heterogeneous ASD samples to replicate our findings, using the present results and effect sizes as informative prior evidence.
Finally, our conclusions were based on a minimal, self-reported phenotype. Although such measures are acceptable as binary case markers in large cohorts, they are not equivalent to standardized clinical interviews and might preferentially capture individuals with more severe or treatment-seeking illness. Self-endorsement of psychiatric diagnoses was indicative of the presence of closely related conditions [103]—but our results cannot be generalized to other phenotype definitions, such as symptom cut-offs, treatment indicators, or more stringent clinical composites. In addition, internalizing symptoms—particularly different anxiety disorders—show substantial co-occurrence and overlapping symptom profiles. As a result, the present findings may not be specific to social anxiety per se, but could reflect broader internalizing or anxiety-related processes. Although we did not observe similar behavioral or neural patterns for co-occurring MDD or GAD in this sample, these null findings should be interpreted cautiously given limited power, and more rigorous diagnostic assessments will be necessary to establish disorder specificity. Moreover, although ADIS (anxiety disorders interview schedule) autism addendum for children [104] exists, no comparable structured interview is available for autistic adults. This underscores the need to develop or adapt clinical interviews tailored for autistic adults with social anxiety, to better capture autism-specific presentations of anxiety. Our finding that social anxiety was associated with reduced task-derived power behavior may provide a useful starting point for informing the development of such autism-specific diagnostic tools.
Conclusion
Social challenges in autism often intensify in adulthood, leading to negative outcomes, particularly for those with co-occurring ASD and social anxiety [4]. Understanding the causes and correlates of social anxiety in autism is essential for developing targeted interventions. Our findings highlight unique behavioral and neural features of social anxiety in autism—lower task-derived power behaviors and larger amygdala volume. Notably, amygdala volume emerged as a neural marker related to both lower task-derived power behaviors and social anxiety, suggesting a potential shared etiology. This means autistic adults with larger amygdalae are more likely to experience social anxiety and difficulties in power dynamics (dominance or control), highlighting a distinctive ASD phenotype. Further research is needed to examine how social cognition and behavior are reflected in neural substrates relevant to mental wellbeing, and to support the development of autism-specific interventions for social anxiety [22,23,105]. Integrating structural/general neuroimaging with naturalistic social paradigms could refine diagnostics and distinguish social anxiety in autism from that in neurotypical populations. Practically, these insights could inform targeted interventions, particularly social power behavior training, to enhance social competence and strengthen social connections for autistic individuals with social anxiety.
Supplementary information
Acknowledgements
We acknowledge the computational and data resources and staff expertise provided by Scientific Computing and Data at the Icahn School of Medicine at Mount Sinai and support from the Clinical and Translational Science Award (CTSA) grant UL1TR004419 from the National Center for Advancing Translational Sciences. Y.H. also thanks Dr. Linlin Yang for valuable initial discussions of this project.
Author contributions
Y.H. conceptualized the study, collected and analyzed data, and wrote the manuscript; S.M.B. provided task and coding resources and commented on the final version of the manuscript; M.S. provided task and coding resources and commented on the final version of the manuscript; E.Z. provided task and coding resources and commented on the final version of the manuscript; S.B. collected the data and commented on the final version of the manuscript; J.T. collected the data and commented on the final version of the manuscript; A.W.P. collected the data and commented on the final version of the manuscript; A.A.T. collected the data and commented on the final version of the manuscript; C.M. collected the data and commented on the final version of the manuscript; X.G. conceptualized the study and commented on the final version of the manuscript; J.H.F-F. conceptualized the study, commented on the previous versions of the manuscript and supervised the work; D.S. conceptualized the study, commented on the previous versions of the manuscript and supervised the work.
Funding
D.S. discloses support for the research of this work from Simons Foundation funding (Grant No. 877761) and National Institute of Mental Health (Grant No. R01MH122611). J.H.F-F. discloses support for the research of this work from Simons Foundation funding (Grant No. 877761) and National Institute of Mental Health (Grant No. R01MH122611). X.G. discloses support for the research of this work from Simons Foundation funding (Grant No. 877761) and National Institute of Mental Health (Grant No. R01MH122611). Y.H., S.M.B., M.S., E.Z., S.B., J.T., A.W.P., A.A.T., C.M. declare no relevant funding.
Data availability
Data and code for this study is available at https://osf.io/2gzxw/?view_only=e0eda95675f84c5798e8460eb9eac01e.
Competing interests
The authors declare no competing interests.
Ethics
The Icahn School of Medicine at Mount Sinai’s Institutional Review Board approved the study protocol (#22-00171 for study 1 and #16-01089 for study 2). All methods were performed in accordance with the relevant guidelines and regulations. All participants gave informed consent and received compensation for their participation.
Footnotes
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
These authors contributed equally: Jennifer H. Foss-Feig, Daniela Schiller.
Contributor Information
Yu Hao, Email: haoyu@nbu.edu.cn.
Jennifer H. Foss-Feig, Email: jfoss-feig@simonsfoundation.org
Daniela Schiller, Email: daniela.schiller@mssm.edu.
Supplementary information
The online version contains supplementary material available at https://doi.org/10.1038/s41398-026-04179-z.
References
- 1.Dietz PM, Rose CE, McArthur D, Maenner M. National and state estimates of adults with autism spectrum disorder. J Autism Dev Disord. 2020;50:4258–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Zeidan J, Fombonne E, Scorah J, Ibrahim A, Durkin MS, Saxena S, et al. Global prevalence of autism: A systematic review update. Autism Research. 2022;15:778–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kanner L. Autistic disturbances of affective contact. In: Nervous child. 1943;2:217–50. [PubMed] [Google Scholar]
- 4.Maddox BB, White SW. Comorbid social anxiety disorder in adults with autism spectrum disorder. J Autism Dev Disord. 2015;45:3949–60. [DOI] [PubMed] [Google Scholar]
- 5.Spain D, Sin J, Linder KB, McMahon J, Happé F. Social anxiety in autism spectrum disorder: A systematic review. Res Autism Spectr Disord. 2018;52:51–68. [Google Scholar]
- 6.Simonoff E, Pickles A, Charman T, Chandler S, Loucas T, Baird G. Psychiatric disorders in children with autism spectrum disorders: prevalence, comorbidity, and associated factors in a population-derived sample. J Am Acad Child Adolesc Psychiatry. 2008;47:921–9. [DOI] [PubMed] [Google Scholar]
- 7.Kessler RC, Petukhova M, Sampson NA, Zaslavsky AM, Wittchen H. Twelve-month and lifetime prevalence and lifetime morbid risk of anxiety and mood disorders in the United States. Int J Methods Psychiatr Res. 2012;21:169–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chevallier C, Kohls G, Troiani V, Brodkin ES, Schultz RT. The social motivation theory of autism. Trends Cogn Sci. 2012;16:231–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kerns CM, Winder-Patel B, Iosif AM, Nordahl CW, Heath B, Solomon M, et al. Clinically significant anxiety in children with autism spectrum disorder and varied intellectual functioning. Journal of Clinical Child & Adolescent Psychology. 2021;50:780–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kerns CM, Kendall PC, Berry L, Souders MC, Franklin ME, Schultz RT, et al. Traditional and atypical presentations of anxiety in youth with autism spectrum disorder. J Autism Dev Disord. 2014;44:2851–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.White SW, Schry AR, Kreiser NL. Social worries and diffi culties: autism and/or socialanxiety disorder? In: Davis TE III, White SW, Ollendick TH, editors. Handbook of Autism and Anxiety. Springer: 2014. p.121–136. 10.1007/978-3-319-06796-4_9. [DOI]
- 12.Maddox BB, Miyazaki Y, White SW. Long-Term Effects of CBT on Social Impairment in Adolescents with ASD. J Autism Dev Disord. 2017;47:3872–82. [DOI] [PubMed] [Google Scholar]
- 13.Spain D, Blainey SH, Vaillancourt K. Group cognitive behaviour therapy (CBT) for social interaction anxiety in adults with autism spectrum disorders (ASD). Res Autism Spectr Disord. 2017;41–42:20–30. [Google Scholar]
- 14.American Psychiatric Association DSMTF, American Psychiatric Association DS. Diagnostic and statistical manual of mental disorders: DSM-5. 5th ed. American psychiatric association: Washington, DC, 2013.
- 15.Capriola NN, Maddox BB, White SW. No offense intended: fear of negative evaluation in adolescents and adults with autism spectrum disorder. J Autism Dev Disord. 2017;47:3803–13. [DOI] [PubMed] [Google Scholar]
- 16.Lei J, Russell A. I have a fear of negative evaluation, get me out of here! examining latent constructs of social anxiety and autistic traits in neurotypical and autistic young people. J Autism Dev Disord. 2021;51:1729–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Pittelkow M-M, aan het Rot M, Seidel LJ, Feyel N, Roest AM. Social anxiety and empathy: a systematic review and meta-analysis. J Anxiety Disord. 2021;78:102357. [DOI] [PubMed] [Google Scholar]
- 18.Montaser J, Umeano L, Pujari HP, Nasiri SMZ, Parisapogu A, Shah A, et al. Correlations between the development of social anxiety and individuals with autism spectrum disorder: a systematic review. Cureus. 2023;15:e44841. [DOI] [PMC free article] [PubMed]
- 19.Müller E, Schuler A, Yates GB. Social challenges and supports from the perspective of individuals with Asperger syndrome and other autism spectrum disabilities. Autism. 2008;12:173–90. [DOI] [PubMed] [Google Scholar]
- 20.White SW, Roberson-Nay R. Anxiety, social deficits, and loneliness in youth with autism spectrum disorders. J Autism Dev Disord. 2009;39:1006–13. [DOI] [PubMed] [Google Scholar]
- 21.Williamson S, Craig J, Slinger R. Exploring the relationship between measures of self-esteem and psychological adjustment among adolescents with asperger syndrome. Autism. 2008;12:391–402. [DOI] [PubMed] [Google Scholar]
- 22.Wilson AC, Gullon-Scott F. ‘It’s not always textbook social anxiety’: A survey-based study investigating the nature of social anxiety and experiences of therapy in autistic people. Autism. 2024;28:2923–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wilson AC, Gullon-Scott F. Social anxiety in autistic people: does the clark and wells model fit?. J Autism Dev Disord. 2024;54:3908–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Tavares RM, Mendelsohn A, Grossman Y, Williams CH, Shapiro M, Trope Y, et al. A map for social navigation in the human brain. Neuron. 2015;87:231–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Leary T Interpersonal Diagnosis of Personality. New York: Ronald; 1957. [Google Scholar]
- 26.Wiggins JS. A psychological taxonomy of trait-descriptive terms: the interpersonal domain. J Pers Soc Psychol. 1979;37:395–412. [Google Scholar]
- 27.Fiske ST, Cuddy AJC, Glick P. Universal dimensions of social cognition: warmth and competence. Trends Cogn Sci. 2007;11:77–83. [DOI] [PubMed] [Google Scholar]
- 28.Tolman EC. Cognitive maps in rats and men. Psychol Rev. 1948;55:189–208. [DOI] [PubMed] [Google Scholar]
- 29.Blay Y, Keshet H, Friedman L, Gilboa-Schechtman E. Interpersonal motivations in social anxiety: Weakened approach and intensified avoidance motivations for affiliation and social-rank. Pers Individ Dif. 2021;170:110449. [Google Scholar]
- 30.Schafer M, Schiller D. Social avoidance can be quantified as navigation in abstract social space. Communications Psychology. 2025;3:51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Zabag R, Bar-Kalifa E, Mor N, Gilboa-Schechtman E. Social anxiety, depression and close relationship: intra and inter-personal perceptions of social-rank and affiliation. J Soc Clin Psychol. 2018;37:582–606. [Google Scholar]
- 32.Hao Y, Banker S, Trayvick J, Barkley S, Peters AW, Thinakaran A, et al. Understanding depression in autism: the role of subjective perception and anterior cingulate cortex volume. Mol Autism. 2025;16:9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Zhang L, Chen P, Schafer M, Zheng S, Chen L, Wang S, et al. A specific brain network for a social map in the human brain. Sci Rep. 2022;12:1773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Wu X, Zhang L, Liu B, Liao J, Qiu Y, Huang R. Social navigation modulates the anterior and posterior hippocampal circuits in the resting brain. Brain Struct Funct. 2023;228:799–813. [DOI] [PubMed] [Google Scholar]
- 35.Banker SM, Schafer M, Barkley S, Trayvick J, Chen A, Peters AW, et al. Neural tracking of social navigation in autism spectrum disorder. Biol Psychiatry. 2025;S0006–3223:01435–0. [DOI] [PMC free article] [PubMed]
- 36.Banker SM, Harrington M, Schafer M, Na S, Heflin M, Barkley S, et al. Phenotypic divergence between individuals with self-reported autistic traits and clinically ascertained autism. Nature Mental Health. 2025;3:286–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Griffiths S, Allison C, Kenny R, Holt R, Smith P, Baron-Cohen S. The vulnerability experiences quotient (VEQ): a study of vulnerability, mental health and life satisfaction in autistic adults. Autism Research. 2019;12:1516–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Reuben KE, Stanzione CM, Singleton JL. Interpersonal trauma and posttraumatic stress in autistic adults. Autism in Adulthood. 2021;3:247–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Pearson A, Rees J, Forster S. “This was just how this friendship worked”: experiences of interpersonal victimization among autistic adults. Autism in Adulthood. 2022;4:141–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Brothers L. The social brain: a project for integrating primate behaviour and neurophysiology in a new domain. Concepts Neuroscience. 1990;1:27–51. [Google Scholar]
- 41.Chen P, Hong W. Neural circuit mechanisms of social behavior. Neuron. 2018;98:16–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Raam T, Hong W. Organization of neural circuits underlying social behavior: A consideration of the medial amygdala. Curr Opin Neurobiol. 2021;68:124–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Sweeten TL, Posey DJ, Shekhar A, McDougle CJ. The amygdala and related structures in the pathophysiology of autism. Pharmacol Biochem Behav. 2002;71:449–55. [DOI] [PubMed] [Google Scholar]
- 44.Baron-Cohen S, Ring HA, Bullmore ET, Wheelwright S, Ashwin C, Williams SCR. The amygdala theory of autism. Neurosci Biobehav Rev. 2000;24:355–64. [DOI] [PubMed] [Google Scholar]
- 45.Herrero MJ, Velmeshev D, Hernandez-Pineda D, Sethi S, Sorrells S, Banerjee P, et al. Identification of amygdala-expressed genes associated with autism spectrum disorder. Mol Autism. 2020;11:39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Endo T, Shioiri T, Kitamura H, Kimura T, Endo S, Masuzawa N, et al. Altered chemical metabolites in the amygdala-hippocampus region contribute to autistic symptoms of autism spectrum disorders. Biol Psychiatry. 2007;62:1030–7. [DOI] [PubMed] [Google Scholar]
- 47.Avino TA, Barger N, Vargas MV, Carlson EL, Amaral DG, Bauman MD, et al. Neuron numbers increase in the human amygdala from birth to adulthood, but not in autism. Proceedings of the National Academy of Sciences. 2018;115:3710–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Seguin D, Pac S, Wang J, Nicolson R, Martinez-Trujillo J, Duerden EG. Amygdala subnuclei development in adolescents with autism spectrum disorder: Association with social communication and repetitive behaviors. Brain Behav. 2021;11:e2299. [DOI] [PMC free article] [PubMed]
- 49.Felix-Ortiz AC, Burgos-Robles A, Bhagat ND, Leppla CA, Tye KM. Bidirectional modulation of anxiety-related and social behaviors by amygdala projections to the medial prefrontal cortex. Neuroscience. 2016;321:197–209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Miller SM, Marcotulli D, Shen A, Zweifel LS. Divergent medial amygdala projections regulate approach–avoidance conflict behavior. Nat Neurosci. 2019;22:565–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Twining RC, Vantrease JE, Love S, Padival M, Rosenkranz JA. An intra-amygdala circuit specifically regulates social fear learning. Nat Neurosci. 2017;20:459–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Amaral DG, Bauman MD, Mills Schumann C. The amygdala and autism: implications from non-human primate studies. Genes Brain Behav. 2003;2:295–302. [DOI] [PubMed] [Google Scholar]
- 53.Hennessy A, Seguin D, Correa S, Wang J, Martinez-Trujillo JC, Nicolson R, et al. Anxiety in children and youth with autism spectrum disorder and the association with amygdala subnuclei structure. Autism. 2023;27:1053–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Andrews DS, Aksman L, Kerns CM, Lee JK, Winder-Patel BM, Harvey DJ, et al. Association of amygdala development with different forms of anxiety in autism spectrum disorder. Biol Psychiatry. 2022;91:977–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Juranek J, Filipek PA, Berenji GR, Modahl C, Osann K, Spence MA. Association between amygdala volume and anxiety level: magnetic resonance imaging (MRI) study in autistic children. J Child Neurol. 2006;21:1051–8. [DOI] [PubMed] [Google Scholar]
- 56.Herrington JD, Maddox BB, Kerns CM, Rump K, Worley JA, Bush JC, et al. Amygdala volume differences in autism spectrum disorder are related to anxiety. J Autism Dev Disord. 2017;47:3682–91. [DOI] [PubMed] [Google Scholar]
- 57.Yarger HA, Nordahl CW, Redcay E. Examining associations between amygdala volumes and anxiety symptoms in autism spectrum disorder. Biol Psychiatry Cogn Neurosci Neuroimaging. 2022;7:916–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Machado-de-Sousa JP, Osório F, de L, Jackowski AP, Bressan RA, Chagas MHN, et al. Increased amygdalar and hippocampal volumes in young adults with social anxiety. PLoS ONE. 2014;9:e88523. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Alberti S. Larger amygdala volume relates to social anxiety in youth with autism spectrum disorders. Doctoral dissertation. 2013.
- 60.Rutter M, Bailey A, Lord C The Social Communication Questionnaire: Manual. Los Angeles: Western Psychological Services; 2003. [Google Scholar]
- 61.Wechsler D. Wechsler Abbreviated Scale of Intelligence--Second Edition (WASI-II) [Database record]. APA PsycTests 2011.
- 62.Wechsler D. Wechsler Adult Intelligence Scale--Fourth Edition (WAIS-IV) [Database record]. APA PsycTests 2008.
- 63.Hus V, Lord C. The autism diagnostic observation schedule, module 4: revised algorithm and standardized severity scores. J Autism Dev Disord. 2014;44:1996–2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.Liebowitz MR. Social Phobia. In: Modern problems of pharmacopsychiatry 1987, pp 141-73. [DOI] [PubMed]
- 65.Heimberg RG, Horner KJ, Juster HR, Safren SA, Brown EJ, Schneier FR, et al. Psychometric properties of the liebowitz social anxiety scale. Psychol Med. 1999;29:199–212. [DOI] [PubMed] [Google Scholar]
- 66.Baker SL, Heinrichs N, Kim H-J, Hofmann SG. The Liebowitz social anxiety scale as a self-report instrument: a preliminary psychometric analysis. Behav Res Ther. 2002;40:701–15. [DOI] [PubMed] [Google Scholar]
- 67.Boulton KA, Guastella AJ. Social anxiety symptoms in autism spectrum disorder and social anxiety disorder: Considering the reliability of - instruments in adult cohorts. Autism Research. 2021;14:2383–92. [DOI] [PubMed] [Google Scholar]
- 68.Hurley RSE, Losh M, Parlier M, Reznick JS, Piven J. The broad autism phenotype questionnaire. J Autism Dev Disord. 2007;37:1679–90. [DOI] [PubMed] [Google Scholar]
- 69.Sasson NJ, Lam KSL, Childress D, Parlier M, Daniels JL, Piven J. The broad autism phenotype questionnaire: prevalence and diagnostic classification. Autism Research. 2013;6:134–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70.Hull L, Mandy W, Lai M-C, Baron-Cohen S, Allison C, Smith P, et al. Development and validation of the camouflaging autistic traits questionnaire (CAT-Q). J Autism Dev Disord. 2019;49:819–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71.Nishiyama T, Suzuki M, Adachi K, Sumi S, Okada K, Kishino H, et al. Comprehensive comparison of self-administered questionnaires for measuring quantitative autistic traits in adults. J Autism Dev Disord. 2014;44:993–1007. [DOI] [PubMed] [Google Scholar]
- 72.Esteban O, Markiewicz CJ, Blair RW, Moodie CA, Isik AI, Erramuzpe A, et al. fMRIPrep: a robust preprocessing pipeline for functional MRI. Nat Methods. 2019;16:111–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 73.Esteban O, Blair RW, Nielson DM, Varada JC, Marrett S, Thomas AG, et al. Crowdsourced MRI quality metrics and expert quality annotations for training of humans and machines. Sci Data. 2019;6:30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 74.Gorgolewski K, Burns CD, Madison C, Clark D, Halchenko YO, Waskom ML, et al. Nipype: a flexible, lightweight and extensible neuroimaging data processing framework in python. Front Neuroinform. 2011;5:13. [DOI] [PMC free article] [PubMed]
- 75.Gorgolewski KJ, Esteban O, Markiewicz CJ, Ziegler E, Ellis DG, Notter MP, et al. Nipype. Software 2018. 10.5281/zenodo.596855. [DOI]
- 76.Tustison NJ, Avants BB, Cook PA, Yuanjie Z, Egan A, Yushkevich PA, et al. N4ITK: improved N3 bias correction. IEEE Trans Med Imaging. 2010;29:1310–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 77.Avants B, Epstein C, Grossman M, Gee J. Symmetric diffeomorphic image registration with cross-correlation: Evaluating automated labeling of elderly and neurodegenerative brain. Med Image Anal. 2008;12:26–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 78.Fonov V, Evans A, McKinstry R, Almli C, Collins D. Unbiased nonlinear average age-appropriate brain templates from birth to adulthood. Neuroimage. 2009;47:S102. [Google Scholar]
- 79.Evans AC, Janke AL, Collins DL, Baillet S. Brain templates and atlases. Neuroimage. 2012;62:911–22. [DOI] [PubMed] [Google Scholar]
- 80.Mingins JE, Tarver J, Waite J, Jones C, Surtees AD. Anxiety and intellectual functioning in autistic children: A systematic review and meta-analysis. Autism. 2021;25:18–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 81.White SW, Bray BC, Ollendick TH. Examining shared and unique aspects of social anxiety disorder and autism spectrum disorder using factor analysis. J Autism Dev Disord. 2012;42:874–84. [DOI] [PubMed] [Google Scholar]
- 82.Lau BY, Leong R, Uljarevic M, Lerh JW, Rodgers J, Hollocks MJ, et al. Anxiety in young people with autism spectrum disorder: Common and autism-related anxiety experiences and their associations with individual characteristics. Autism. 2020;24:1111–26. [DOI] [PubMed] [Google Scholar]
- 83.Bearss K, Taylor CA, Aman MG, Whittemore R, Lecavalier L, Miller J, et al. Using qualitative methods to guide scale development for anxiety in youth with autism spectrum disorder. Autism. 2016;20:663–72. [DOI] [PubMed] [Google Scholar]
- 84.Kuschefski M, Falter-Wagner CM, Bente G, Vogeley K, Georgescu AL. Inferring power and dominance from dyadic nonverbal interactions in autism spectrum disorder. Autism Research. 2019;12:505–16. [DOI] [PubMed] [Google Scholar]
- 85.Neely BH. To disclose or not to disclose: Investigating the stigma and stereotypes of autism in the workplace. Master thesis, Pennsylvania State University 2016.
- 86.Bolis D, Balsters J, Wenderoth N, Becchio C, Schilbach L. Beyond Autism: Introducing the Dialectical Misattunement Hypothesis and a Bayesian Account of Intersubjectivity. Psychopathology. 2017;50:355–72. [DOI] [PubMed] [Google Scholar]
- 87.Walters KS, Hope DA. Analysis of social behavior in individuals with social phobia and nonanxious participants using a psychobiological model. Behav Ther. 1998;29:387–407. [Google Scholar]
- 88.Russell JJ, Moskowitz DS, Zuroff DC, Bleau P, Pinard G, Young SN. Anxiety, emotional security and the interpersonal behavior of individuals with social anxiety disorder. Psychol Med. 2011;41:545–54. [DOI] [PubMed] [Google Scholar]
- 89.Davis M, Whalen PJ. The amygdala: vigilance and emotion. Mol Psychiatry. 2001;6:13–34. [DOI] [PubMed] [Google Scholar]
- 90.Top DN Jr, Stephenson KG, Doxey CR, Crowley MJ, Kirwan CB, South M. Atypical amygdala response to fear conditioning in autism spectrum disorder. Biol Psychiatry Cogn Neurosci Neuroimaging. 2016;1:308–15. [DOI] [PubMed] [Google Scholar]
- 91.Kovacevic M, Macuzic IZ, Milosavljevic J, Lukovic T, Aleksic D, Gavrilovic J, et al. Amygdala volumes in autism spectrum disorders: meta-analysis of magnetic resonance imaging studies. Rev J Autism Dev Disord. 2023;10:169–83. [Google Scholar]
- 92.Kolesar TA, Bilevicius E, Wilson AD, Kornelsen J. Systematic review and meta-analyses of neural structural and functional differences in generalized anxiety disorder and healthy controls using magnetic resonance imaging. Neuroimage Clin. 2019;24:102016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 93.Townend S, Staginnus M, Gao Y, Alexander N, Arolt V, Banaschewski T, et al. Shared and distinct alterations in brain structure of youth with internalizing or externalizing disorders: findings from the ENIGMA antisocial behavior, ADHD, major depressive disorder, and anxiety working groups. Biol Psychiatry. 2025;S0006–3223:01405–2. [DOI] [PubMed]
- 94.Goodkind M, Eickhoff SB, Oathes DJ, Jiang Y, Chang A, Jones-Hagata LB, et al. Identification of a common neurobiological substrate for mental illness. JAMA Psychiatry. 2015;72:305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 95.Fortea L, Ortuño M, De Prisco M, Oliva V, Albajes-Eizagirre A, Fortea A, et al. Atlas of gray matter volume differences across psychiatric conditions: a systematic review with a novel meta-analysis that considers co-occurring disorders. Biol Psychiatry. 2025;98:76–90. [DOI] [PubMed] [Google Scholar]
- 96.Bickart KC, Wright CI, Dautoff RJ, Dickerson BC, Barrett LF. Amygdala volume and social network size in humans. Nat Neurosci. 2011;14:163–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 97.Nam HH, Jost JT, Kaggen L, Campbell-Meiklejohn D, Van Bavel JJ. Amygdala structure and the tendency to regard the social system as legitimate and desirable. Nat Hum Behav. 2017;2:133–8. [Google Scholar]
- 98.Dalton KM, Nacewicz BM, Johnstone T, Schaefer HS, Gernsbacher MA, Goldsmith HH, et al. Gaze fixation and the neural circuitry of face processing in autism. Nat Neurosci. 2005;8:519–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 99.Kleinhans NM, Richards T, Weaver K, Johnson LC, Greenson J, Dawson G, et al. Association between amygdala response to emotional faces and social anxiety in autism spectrum disorders. Neuropsychologia. 2010;48:3665–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 100.Herrington JD, Miller JS, Pandey J, Schultz RT. Anxiety and social deficits have distinct relationships with amygdala function in autism spectrum disorder. Soc Cogn Affect Neurosci. 2016;11:907–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 101.Kuusikko S, Pollock-Wurman R, Jussila K, Carter AS, Mattila M-L, Ebeling H, et al. Social anxiety in high-functioning children and adolescents with autism and asperger syndrome. J Autism Dev Disord. 2008;38:1697–709. [DOI] [PubMed] [Google Scholar]
- 102.Sordo Vieira L, Nguyen B, Nutley SK, Bertolace L, Ordway A, Simpson H, et al. Self-reporting of psychiatric illness in an online patient registry is a good indicator of the existence of psychiatric illness. J Psychiatr Res. 2022;151:34–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 103.Kerns CM, Renno P, Kendall PC, Wood JJ, Storch EA. Anxiety disorders interview schedule–autism addendum: reliability and validity in children with autism spectrum disorder. Journal of Clinical Child & Adolescent Psychology. 2017;46:88–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 104.Fresco DM, Coles ME, Heimberg RG, Liebowitz MR, Hami S, Stein MB, et al. The liebowitz social anxiety scale: a comparison of the psychometric properties of self-report and clinician-administered formats. Psychol Med. 2001;31:1025–35. [DOI] [PubMed] [Google Scholar]
- 105.Hunsche MC, Kerns CM. How social cognition relates to different dimensions of social connection among autistic individuals: A systematic review. Clin Psychol: Sci Pract. 2025;32:307–24. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data and code for this study is available at https://osf.io/2gzxw/?view_only=e0eda95675f84c5798e8460eb9eac01e.