Abstract
Social interaction difficulties are a defining feature of autism. Historically, research has concentrated on comparing the social behaviors of autistic and neurotypical individuals in cross-neurotype dyads, often attributing challenges in communication to autistic social deficits. However, recent studies on interactions between two autistic individuals provide new insights, suggesting that some social difficulties may at least partially arise from mutual misunderstandings. This narrative review analyzes 25 studies on autistic, mixed, and neurotypical interactions, highlighting consistent differences (e.g., reduced mutual gaze and backchanneling in autistic interactions) and similarities between autistic and neurotypical social interactions (e.g., turn-taking), as well as measurements yielding mixed results (e.g., social motor synchrony). We discuss the variability in social behaviors, interaction outcomes, and study designs in the literature and interpret the existing findings in the light of individual deficits and relational difficulties. We emphasize the need for future research to consider both same- and mixed-neurotype interactions and to include perspectives from neurotypical as well as autistic individuals to enhance our understanding of social interaction in autism.
Community Brief
Why is this topic important?
Social interaction difficulties are a core feature of autism, but traditional research has often focused on comparing autistic individuals with neurotypical individuals, attributing communication difficulties to deficits within autistic individuals. Recent studies, however, suggest that social difficulties may arise not only from individual differences but also from mismatches in interaction styles between autistic and neurotypical individuals. This shift could improve how we approach and support social communication in autism.
What was the purpose of this article?
The purpose of this narrative review was to examine and synthesize findings from 25 studies that explore social interactions in autistic, mixed-neurotype, and neurotypical dyads. The review aimed to identify patterns in social behaviors, understand the differences and similarities across these groups, and offer a more nuanced view of the factors contributing to social difficulties in autism.
What were the results of this study?
The review found both consistent and variable patterns in social interactions. Autistic interactions often show reduced mutual gaze and backchanneling (e.g., nodding or verbal affirmations), but similarities to neurotypical interactions were found in turn-taking. The results on social motor synchrony (the coordination of movements between individuals) were mixed, highlighting the complexity of social behaviors. The study also noted considerable variability in how social behaviors were measured across studies.
What do the authors conclude?
The authors concluded that social difficulties in autism cannot be fully explained by deficits in autistic individuals alone. Instead, these challenges often stem from relational difficulties—misunderstandings or mismatches in how autistic and neurotypical individuals interact. By considering both individual differences and relational dynamics, the review advocates for a shift away from the fully deficit-based models of autism and toward a more relational and context-sensitive understanding of social interaction.
What do the authors recommend for future research on this topic?
The authors recommend that future research should focus on examining both same- and mixed-neurotype interactions. They also emphasize the need for studies to include perspectives from both autistic and neurotypical individuals to gain a deeper understanding of social communication. More systematic studies are needed that look at how both parties in an interaction contribute to social difficulties and successes.
How will this analysis help autistic adults now and in the future?
This review provides new insights into the nature of social difficulties faced by autistic individuals and emphasizes the importance of relational dynamics rather than solely focusing on deficits. For autistic adults, this perspective could help improve social support, educational programs, and clinical practices by recognizing that social misunderstandings are mutual rather than one-sided. In the future, this research could contribute to developing more effective interventions that focus on improving communication and understanding in both same- and mixed-neurotype interactions, ultimately promoting more inclusive social environments.
Introduction
Autism and social interactions
Social interactions are complex, dynamic processes involving multiple participants and influenced by various intra- and interpersonal factors. Each participant contributes their social abilities—cognition, motivation, and behaviors—creating unique interaction dynamics and outcomes (see Fig. 1). Difficulties in social interaction are considered core characteristics of autism, yet the causes remain debated. This review examines social behaviors in interactions between autistic1,2,a and/or neurotypical b individuals, critically evaluating current knowledge.
Model of social interaction. Social ability of each interactant includes social cognition (e.g., theory of mind, perspective taking, empathy, emotion recognition), social motivation (e.g., social orienting, willingness to initiate and maintain relationships) and other contextual factors (e.g., cultural knowledge, language proficiency, physical well-being). The expression of this ability is social behaviors, spanning intrapersonal behaviors (i.e., individual behavioral skills and tendencies that can be measured in a single-person setting), and interpersonal behaviors (i.e., those that emerge and can only be measured in an interaction with another social agent). Social interaction outcomes are measurable results of social interactions, assessing the quality and effectiveness of the interaction according to participants’ perceptions (e.g., ratings of rapport) or observable prosocial behaviors (e.g., information sharing, helping).
The clinical model recognizes autism as a disorder with persistent social deficits. 3 Accounts such as the Theory of Mind 4 or the Social Motivation Hypothesis 5 support this perspective by proposing that autistic people have limited ability to interpret other people’s minds and that they do not want to or cannot engage in and maintain social relations. These accounts receive criticism for being overly pathologizing6–8 and having limited power to predict the “real-world” social behaviors of individuals in the autistic constellation.9–11,c For example, despite difficulties, many autistic people are as motivated to have social relationships as their neurotypical peers. 12 The neurodiversity movement conceptualizes autism as a natural neurodevelopmental variation, with interaction difficulties rooted in differences in their social interactions rather than being deficient. As the minority in the society (∼1% worldwide 13 ), autistic people are part of a society where structures, systems, and communication styles are defined by neurotypical norms. Thus, autistic communication styles are not always accommodated, which may lead to communication difficulties, particularly when interacting with neurotypical people.
Regardless of the theoretical stance, autistic people experience significant social interactions difficulties, especially, but not exclusively, with neurotypical peers. For example, both neurotypes form unfavorable impressions of autistic peers after brief interactions or video observation.14–18 Thus, social behavior carries signals that lead to instantaneous social judgments. At the same time, one’s “communication style” seems to be stable across contexts.19,d For example, people are consistent in what and how they communicate (feelings, needs, thoughts) in different social settings (e.g., with their partner, friend, coworker). 20 Because of that, researchers have investigated individual social behaviors such as facial expressivity 21 or use of co-speech gestures 22 in single-person designs. However, it is unclear whether differences between autistic and neurotypical people in such designs could explain the interaction problems in cross-neurotype dyads.
Social interaction style
Social interactions are inherently dynamic, bi- (or multi-) directional, back-and-forth processes, in which one person influences the other. 17 Hence, instead of focusing on individual communication styles, we propose examining interaction styles, encompassing both intrapersonal social abilities (e.g., facial expressivity or perspective taking) and interpersonal relational factors emerging between interlocutors (e.g., turn-taking). Autistic and neurotypical social interaction styles likely differ, leading to misunderstandings. The key question is whether these difficulties stem from autistic social skill deficits, mutual biases, or both.
The idea of an individual having a consistent social interaction style is not new. Previous research 23 divided autistic children into subgroups characterized as follows: socially aloof (not seeking interactions and indifferent to approaches of others), passive (not initiating but responding to approaches of others), active-but-odd (seeking interactions although in an atypical way), and typical. The original descriptions of the styles included behavioral variables such as echolalia, idiosyncratic speech, and repetitive symbolic activities, and were further related to cognitive and emotional factors, 24 and reported to be relatively stable across time, at least in adolescence. 25 Note that while the individual descriptors are intrapersonal (e.g., idiosyncratic speech), the phrasing of the subgroups is primarily interpersonal (e.g., indifference to others’ approaches), highlighting the relational nature of autistic social difficulties, which may arise from individual atypicalities. Given the considerable heterogeneity of autism, we conceptualize social interaction styles as multidimensional profiles with overlaps and differences between both individuals and diagnostic groups.
It takes (at least) two
Because clinical assessments and research are largely based on neurotypical norms 26 and autistic communication and interaction style often does not conform to these norms, 27 it is tempting to conclude that social difficulties experienced by autistic people in the neurotypical world are predominantly due to individual social impairments, following the clinical model. Consequently, numerous studies have compared autistic versus neurotypical groups, or mixed-neurotype versus neurotypical dyads, often attributing any observed difficulties to social impairments in autistic individuals.
However, two overlapping theoretical frameworks offer alternatives to the clinical model’s focus on intrinsic autistic social difficulties. The double-empathy problem (DEP) 28 proposes that communication gaps arise from differing worldviews and information processing between autistic and neurotypical individuals, creating mutual challenges. Similar to cross-cultural communication, difficulties arise not from one party’s deficits but also from reciprocal misunderstandings. 29 As a sociological lens for understanding cross-neurotype interactions, DEP is not positioned to provide precise empirical predictions. This limits the interpretation of available data as either supporting or contradicting its assumptions. Nevertheless, DEP has gained significant academic and public attention. 30 Similarly, the dialectical misattunement hypothesis 31 (DMH) predicts that interactions are more synchronous and enjoyable when the interactants are more likely to predict each other’s actions, for example, due to a similar neurotype, experiences, and expectations. Based on both DEP and DMH, it could be expected that interactions between same-neurotype dyads (i.e., both autistic or both neurotypical) would be on average “better” or more successful than those of mixed-neurotype dyads.
Despite this relational perspective, most literature on autistic social difficulties focuses on social cognition abilities of individuals32,33 rather than relational factors in interactions, 17 generating extensive literature on atypicalities in social cognition, behaviors, and motivation, 34 with social interaction itself remaining a blind spot.35,36 Interestingly, the relationship between interaction outcomes and individual social cognition may be surprisingly weak.34,37 Nevertheless, given the quick emergence of first impressions in interactions,18,34,38,39 cross-neurotype difficulties must be rooted in behavioral differences (or “social presentation styles” 17 ) between neurotypes, with possibly different presentations and outcomes in same- or mixed-neurotype dyads. However, only recently have researchers begun exploring how interactions between two autistic individuals differ from other types of dyads.
Review
Method and scope
To addresses the gap in the literature concerning real-world interactions between autistic and neurotypical individuals, 35 we focus on studies that use interactive paradigms involving participants engaged in real or arranged interactions as opposed to tasks in which a person is evaluated alone. Thus, as inclusion criteria, in this review we include studies that report (1) social behaviors and/or interaction outcomes (2) measured in a social interaction, (3) including autistic dyads (two autistic participants) and neurotypical and/or mixed-neurotype dyads. Our aim was to synthesize the current state of knowledge on social interactions between autistic and neurotypical individuals and to highlight key gaps, inconsistencies, and directions for future research.
No single set of keywords adequately captures studies meeting our inclusion criteria. Instead, we iteratively included articles found in the bibliographies of relevant articles, using the PubMed and Google Scholar databases. We focused on articles published from 2010 onward, reflecting the rise of research with relational views on autism. No restrictions were placed on sample size, age, or intelligence. In total, 25 studies met the criteria and are summarized in Table 1. Supplementary Table S1 in Supplementary Data expands on this with details on dyad types, diagnosis disclosure, task, coder neurotype, age, gender ratio, sample sizes, and type of data collected. This work follows guidelines for narrative reviews.63–66
Studies Investigating Social Interaction Characteristics in Interactive Paradigms Including Autistic Dyads
AUT, autistic dyad; NT, neurotypical (non-autistic) dyad; MIX, mixed-neurotype (one autistic and one neurotypical) dyad.
RD: relational difficulties = whether the main findings of the study support the relational character of autistic social difficulties, that is, whether the measured characteristic differentiates AUT and MIX dyads such that AUT are more synchronized or report higher rapport (or similar interaction outcome) with the interaction partner; or whether an individual characteristic supports more successful interaction of autistic participants with other autistic people. “-” is used where it is not possible to assess (e.g., lack of MIX dyads). The subscript “ind” indicates that the corresponding measurement was calculated per individual (not dyad). For more details on the conceptualization of each measurement and additional study information (task used, neurotype of the coders, age group, gender ratio, sample size), see Supplementary Material.
We organize the social behaviors by whether they are intra- or interpersonal in nature. Intrapersonal social behaviors occur within an individual, despite being developed in a social context (e.g., speech-supporting gestures). Interpersonal social behaviors emerge in interactions between individuals and cannot be observed without another agent (e.g., mutual eye contact). The distinction between them is not rigid. 67
This review is not an exhaustive list of social behaviors measured in an interaction but is rather intended as a starting point to be updated and expanded as the field develops. Where neurotype-matched dyad research is limited, we also discuss results from mixed-neurotype dyads and from varying levels of autistic traits in the general population. However, it is important to note that these last findings should be viewed as potential directions for further research in diagnosed autistic populations, as studies on autistic traits in the general population are not equivalent to research on individuals with a formal diagnosis. 68 This is not an attempt to test DEP, as it does not generate specific predictions 30 (however, Table 1 interprets results regarding relational social difficulties in autism). Instead, this review is an attempt to critically and systematically evaluate the available findings on autistic social interactions in terms of social behaviors and interaction outcomes.
Interpersonal Social Interaction Characteristics
Gaze
The literature on gaze to interlocutor’s face in autism is mixed. Some report reduced attention to eyes and face in autistic children, 69 others find no differences,70,71 and one reported that both neurotypes look more at a conversation partner’s face when discussing an interesting topic rather than a generic one.
Gaze helps coordinate social interaction dynamics, such as turn-taking. Neurotypical individuals typically look away when speaking and return their gaze to signal a turn’s end. 72 Similar gaze functions exist across autistic traits, 73 yet autistic individuals often report difficulties with gaze timing in conversation. 74 Studies have also linked autism to reduced coordination between gaze and other social behaviors, particularly in dynamic face-to-face interactions. 75
Only one study has examined gaze in autistic dyads. Rifai and colleagues 40 found higher mutual gaze in neurotypical than mixed-neurotype dyads, with the latter not differing from autistic dyads. They also reported a positive correlation between mutual gaze and self-reported rapport, but only in neurotypical dyads. Similarly, other studies report less mutual gaze in mixed and high-autistic-trait dyads, 76 and reduced face-directed gaze in individuals with higher autistic traits. 77 While more research is needed, current evidence suggests that autistic individuals establish mutual gaze less frequently, although this does not necessarily impair interaction outcomes such as rapport.
Turn-taking
Turn-taking is a cooperative interaction fundamental to social interaction, 78 aiming to organize an exchange such that only one person speaks at one time. 79 Little is known about turn-taking in autism, particularly in autistic interactions. Recent research by Wehrle and colleagues 80 offers initial insights into turn-taking in autistic dyads, specifically among German speakers.
Wehrle 80 found that neurotypical and autistic dyads showed similar proportion of silence, overlap, and single-speaker speech. However, autistic dyads had less balanced speaking times, with one partner often talking more, especially early in conversations. Neurotypical dyads engaged in longer exchanges with more turn transitions and shorter utterances. At the population level, Peng et al. 76 found that high-autistic-trait dyads had fewer turns than low-trait dyads.
Gaps are silent pauses between speakers, while overlaps occur when one starts speaking before the other finishes. These are quantified using a continuous measure called floor transfer offset (FTO), where positive values indicate gaps and negative overlaps. In neurotypical interactions, gaps are typically brief (∼200 ms) and shorter gaps enhance feelings of connection.81–83
Wehrle, Cangemi, et al. 42 found no overall FTO differences between autistic and neurotypical adults, but autistic dyads had longer gaps early in conversations, suggesting a slower alignment in conversational dynamics. Peng et al. 76 reported fewer overlaps in high-autistic-trait dyads, suggesting reduced predictive abilities in conversation. A few studies have also found longer gaps84–86 and silences before answering a question73,87 in mixed-neurotype versus neurotypical interactions, although these studies involved neurotypical adults conversing with younger autistic partners, often in interview-like settings, with sometimes unclear intelligence range of the samples.
Backchannels
Backchanneling involves verbal (e.g.,” mhm,” “yeah”) and nonverbal (e.g., head nods) cues produced by listeners that signal engagement and encourage speakers to hold the floor. Verbal backchannels account for most within-speaker overlaps, but as an implicit, other-oriented signal, they reinforce rather than interrupt speech. Neurotypical individuals rely on backchannels to maintain smooth interactions, and they often perceive deviations from typical patterns as negative. 88
Few studies have examined backchanneling in autism. Rifai et al. 40 found fewer nonverbal backchannels in autistic dyads and fewer verbal ones in both autistic and mixed dyads. The rate of either type of backchanneling did not correlate with self-rated rapport. Wehrle, Vogeley et al. 41 also reported reduced and less varied verbal backchannels in autistic dyads, especially early in conversations. Autistic speakers showed a stronger preference for specific backchannels and used more rising intonations, while neurotypical speakers varied their intonation based on back channel type. Similarly, a study of mixed dyads found that only neurotypical children used the Japanese backchannel /ne/ for turn initiations, requests, and confirmations, even though autistic children understood its function. 89
Despite limited research, findings consistently suggest that autistic individuals use fewer and less varied backchannels. However, many studies used task-based interactions rather than free conversation, where autistic individuals may prioritize goal completion over small talk, possibly accounting for reduced backchanneling. 80
Smiling and laughter
Smiling and laughter play key roles in conversations, varying in function based on context and reciprocity, 90 for example, shared laughter, unlike solo laughter, fosters social bonding and emotional convergence. 80 In neurotypical dyads, smiling synchrony predicts positive interaction outcomes such as connection and enjoyment.91,92
Only three studies have examined smiles and laughter in autistic dyads. McNaughton et al. 51 reported that mixed dyads showed less synchrony than neurotypical dyads and marginally less than autistic dyads when controlling for overall smiling. Wehrle 80 observed twice as much laughter in neurotypical dyads compared with autistic ones. Bauminger et al. 62 found autistic children made more eye contact with a smile when interacting with neurotypical than autistic peers. Findings from studies without autistic dyads align with these results: autistic children produce less varied laughter than their typically developing peers 93 and are less likely to reciprocate smiles or laughter, despite similar rates of solo laughter (compared with children with Down syndrome 94 ). Similarly, autistic adolescents smile less, receive fewer smiles, and show lower interpersonal smile coordination, with less improvement over time. 95 Contrary to that, a population-based study found no difference in smile coordination across autistic traits. 76 Interestingly, although depression affects smiling and laughter 96 and is common in autism, 97 none of these studies has reported related diagnosis or traits in their samples.
Synchrony and reciprocity
Research consistently shows that neurotypical individuals spontaneously synchronize behaviors from infancy, 98 with interpersonal synchrony fostering enjoyment, prosocial behavior, and positive affect.99,100 Even previously unacquainted dyads watching a movie together synchronize smiles, strengthening social connection.91,92 Synchrony is one of the most studied aspects of social interaction in autism.
Reviews suggest less interpersonal synchrony in autistic and mixed dyads compared with neurotypical ones.101–104 Autistic adolescents show reduced spontaneous and intentional synchrony across tasks involving imitation, interpersonal coordination, facial expression, eye gaze, body language, affect, and motion energy.105–107 However, some evidence indicates successful social motor adaptation in autism, even in individuals with severe learning disabilities. 108
Social motor synchrony refers to the rhythmically matched movements of two or more people. 105 Typically, synchrony is more prominent in spontaneous than in intentional tasks in autism, 102 but it is lower in autistic or mixed dyads than neurotypical ones.
In a series of conversational tasks, Georgescu et al. 48 found that adult neurotypical dyads exhibited more motion energy synchrony than autistic or mixed-neurotype dyads, but Efthimiou and colleagues found no dyad differences in a single 5-minute interaction, although they reported a stronger link between motion synchrony and rapport in neurotypical than other dyads. 50 In another study using a collaborative joint-action task, autistic children showed less benefit from peer assistance and were less likely to synchronize their movements compared with neurotypical children, even though they demonstrated greater stability in solo tasks, suggesting differences in interpersonal coordination rather than motor skills. 53 On the contrary, Glass and Yuill 49 reported that while there were group differences in the overall amount of motion, autistic children showed equal or even greater motor synchrony compared with neurotypical peers.
An alternative and innovative approach was taken by Ward et al., 109 who investigated motor synchrony in autistic children and professional actors during an interactive theatrical play using wearable sensors. They captured motor coupling that might be missed by human observers. While they do not point to a particular pattern or type of social behavior and do not compare synchrony between neurotypes, their approach offers promising possibilities for future naturalistic research.
In the context of communication alignment, the only study including autistic interactions found that autistic dyads were less likely to align nonverbal signals with recent communicative history compared with mixed or neurotypical dyads, indicating reduced nonverbal synchrony. 54 In addition, in mixed-neurotype interactions, autistic children produced fewer contingent utterances and more monologue-like speech, indicating decreased reciprocity. 71 Together, these studies suggest that autistic individuals may be less adaptive to their communication partner in dynamic exchanges.
Finally, oscillatory neural synchrony arises from continuous adaptation of the interacting partners’ behaviors. 110 While it is crucial to explore the neural correlates of interpersonal synchrony in autistic communication, no hyperscanning studies have yet focused on autistic dyads. Existing hyperscanning research suggests decreased neural synchrony in mixed-neurotype dyads. For instance, autistic adults show less neural synchrony with a neurotypical confederate than their neurotypical peers, 111 which is related to atypical eye contact112,113 and greater social difficulties. 114 However, Peng et al. 76 reported enhanced neural synchronization (and self-reported interaction outcomes) in dyads with high autistic traits during a semistructured conversational task, compared with dyads with low autistic traits and mixed dyads.
Intrapersonal Social Interaction Characteristics
In the following, we describe intrapersonal social behaviors as reported in studies including interaction between autistic and/or neurotypical people. However, we encourage the reader to explore existing autism literature investigating similar characteristics in single-person designs, for example, prosody, 115 emotion recognition,33,116 and motor skills. 117
Facial expressiveness
Facial expressiveness—referring to variations in dynamic facial behavior, with more expressive individuals generally being perceived as more agreeable and well-liked in social interactions 118 —may be an important factor in autism. Some studies find greater variance in facial expressions in autistic compared with neurotypical individuals, 119 while others suggest that autistic minds are harder to read,120–122 possibly due to lower expressiveness. 120 However, results are mixed as to whether autistic people are 123 or are not 122 less expressive. Nevertheless, neurotypical raters tend to perceive autistic expressions as less natural and more atypical,18,124,125 which leads them to assign less favorable social ratings. 120
Kinematics
Motor behaviors may differ between autistic and neurotypical individuals in low-level kinematics, potentially influencing higher level social behaviors such as social motor synchrony. Evidence shows that autistic individuals exhibit atypical movement patterns in production and imitation, which may be mediated by motor execution as well as social motivation, action representation, and executive function. 126 For instance, Zhao et al. 127 found group differences in autistic and neurotypical children’s head movements across all three axes, while Edey et al. 121 observed kinematic differences in videos of autistic and neurotypical adults directing geometric shapes to generate animations.
Prosody
Prosody refers to the rhythm, intonation (melody), and flow of speech, including pauses. 128 It plays a key role in expressing emotions and intentions, clarifying grammatical structure and sentence meaning, and managing discourse with others. For example, speakers modulate pitch to signal sentence structure (e.g., raising pitch in questions) and conversational cues (e.g., falling pitch at the end of a turn and rising pitch at the beginning of a turn). A recent review by Grice et al. 128 analyzes both the perception and production of prosody in autism across development. However, to our knowledge, studies involving autistic dyads have so far focused only on production.
Autistic intonation is often described as either “robotic” and “monotonous” (little pitch variation) or “melodic” and “singsongy” (greater variation), 128 with empirical studies supporting the latter and highlighting individual differences.45,80 Furthermore, Wehrle 80 found no prosodic differences in turn transitions between autistic and neurotypical dyads. In another study with a simulated interaction task (“conversing” with an actress in a pre-recorded video), autistic participants showed a higher mean pitch and lower harmony–noise ratio than the neurotypical group. 129 Although social outcomes could not be measured in that study, these basic prosodic differences should be explored in relation to interaction characteristics in the future.
Filled pauses are hesitations in speech, such as “uhm” or “uh,” which serve to maintain the current turn in conversation. Unlike backchannels, filled pauses are produced by the speaker, not the listener, and can occur in a single speaker’s speech. While high rates of filled pauses may be viewed negatively in public speaking, they can facilitate understanding and the flow of everyday conversation. 80
Research on filled pauses in autism is limited, with only one study including autistic dyads. Wehrle, Grice, et al. 46 found no differences in frequency or choice of filled pauses between autistic and neurotypical dyads, although autistic individuals showed more atypical intonation (greater pitch variation) and larger differences between interlocutors in a dyad. This could reflect higher heterogeneity in the autistic population or less adaptation to the partner.
Studies on mixed dyad interactions, including only one on adults, generally show that autistic individuals produce fewer filled pauses than neurotypical peers. Adults use fewer “uhm” and “uh,” 130 while children and adolescents produce fewer “uhm.”131–136 One study found no differences. 137 It should also be noted that most focused on structured interactions (e.g., ADOS interview) or monologic speech, indicating a need for data from natural, unstructured conversations.
Silent pauses, or moments of no speech within a speaker’s turn, are common. The only study of silent pauses from autistic versus neurotypical conversations found that autistic dyads produced more long silent pauses (i.e., longer than 700 ms or 2 seconds, in two analyses) than neurotypical dyads.47,80 This group difference disappeared when analyzing any length of silent pause. Studies not including autistic dyads produced mixed results: autism was linked to fewer silent pauses in children when narrating a book, 138 but similar amount or more silent pauses in adults in a sentence repetition task and in an interview-like conversations, respectively.130,139 Overall, it seems likely that autistic people produce more silent pauses than neurotypicals, but this should interpreted with caution given the small amount of previous work and considerable methodological differences among them (see Wherle 80 for a discussion).
Social Interaction Outcomes
To understand how social behaviors influence interactions, it is necessary to link them to interaction outcomes. However, there is no unified definition of what constitutes a “good” interaction, 34 and the success in conversation depends on individual, often multiple, goals. 140 It is also unclear whether autistic and neurotypical individuals perceive interactions and their outcomes similarly. For instance, while both neurotypes rate interactions with autistic partners as more awkward, autistic raters are more willing to interact again, 17 suggesting differing perceptions of success. Some studies use self-ratings of satisfaction, rapport, or willingness to interact again, while others assess objective outcomes such as information transfer 57 or task performance.43,44 Some studies, however, do not measure success directly and instead conceptualize social behaviors as interaction outcomes. This can lead to the assumption that deviations from neurotypical norms in behavior imply unsuccessful interactions. For example, if increased eye contact enhances rapport in neurotypical dyads but not in autistic dyads, focusing on mutual gaze may mislead us into thinking autistic interactions are less successful, despite similar rapport ratings across groups. 40
Rapport is a commonly used social interaction outcome, but results are inconsistent (see Table 1). Some studies report higher rapport in same-neurotype than mixed dyads,40,57 while others find higher rapport in neurotypical than autistic dyads, 43 or no differences between dyad types55,58 Self- and observer-ratings can also diverge: in one study, mixed dyads were rated lower by participants but higher by observers. 56
Some studies additionally assessed objective outcomes. Crompton and colleagues found that self-rated rapport and task success (information transfer) were higher in same-neurotype dyads than mixed dyads, 57 but in a well-powered replication of this study, they found no differences between dyad types in information transfer, and rapport was higher in neurotypical than autistic and/or mixed dyads. 58 On the contrary, Oates et al., 43 who used a different task and objective measures (drawing accuracy, speed, and communication clarity), found lower self-rated rapport in autistic dyads, but no clear link to task performance. In fact, mixed dyads completed the task faster, and all dyads were similar in clarity, suggesting no consistent relationship between subjective and objective measures.
Other factors may shape these outcomes. In a four-person group study, 59 autistic participants rated rapport highest in all-autistic groups, but ratings declined with more neurotypical peers. Neurotypical participants’ ratings were stable, suggesting they may face fewer relational barriers (see also Geelhand et al. 44 ). Perceived group composition may have influenced responses 39 : neurotypicals might have become or tried to appear more tolerant, while autistic participants might have felt more accepted in all-autistic groups, boosting rapport ratings.
Finally, Morrison et al. 17 assessed interaction quality along broader socio-communicative impressions. Both groups rated autistic partners as more awkward and less warm, although competence ratings (intelligence, quality of conversation) were similar. Also, autistic participants disclosed more and felt closer in same-neurotype dyads, and only neurotypical participants preferred future interactions within their own neurotype. Although socio-communicative mechanisms behind social interaction outcomes are rarely investigated,17,57,59 it should be an important focus in future studies.
Qualitative and observational coding approaches provide a middle ground between behavioral measures and social interaction outcomes. Through a qualitative linguistic analysis, Williams et al. 60 found that autistic conversations showed greater flow, rapport, and intersubjective attunement than those of mix-neurotype dyads, whether familiar or unfamiliar. Crucially, participants were paired multiple times with different partners, revealing that some autistic individuals showed markedly higher communicative competence with autistic than neurotypical partners. This highlights the relational nature of social differences in autism.
In contrast, studies coding autistic children’s and adolescents’ free interactions found greater participation and friendship quality in mixed dyads than autistic-only dyads.61,62 However, the children were not asked to evaluate their own interactions, and the behavioral coders—presumably neurotypical—had significant interpretative flexibility. For example, “expressing enjoyment while playing” can take different forms in autistic and neurotypical children. A neurotypical perspective may therefore favor behaviors more typical of mixed dyads.
Cognitive, Affective, and Other Mechanisms Underlying Autistic Social Difficulties
Beyond social behaviors, cognitive, affective, and contextual factors shape social interactions. For instance, cognitive biases such as rejection sensitivity can lead to negative perceptions of interactions. 141 This is common in social anxiety, 142 which frequently co-occurs with autism. 143 Emotion regulation also plays a role; better regulation enhances well-being and social functioning, 144 while emotion dysregulation, common in autism, 145 may impact interactions. Also, interacting with those sharing one’s neurotype fosters a sense of connectedness to the community. This connectedness has significant benefits for autistic individuals, from increased self-esteem to generally increased well-being. 146 Knowing a partner is also autistic may reduce rejection anxiety and enhance perceived rapport. 59 Yet, despite their importance, few studies control for such cognitive/affective factors in autistic dyads.
One study examined theory-of-mind use in free conversations, finding that while both autistic and neurotypical youth referenced mental states of the partners, autistic individuals violated (neurotypical) conversational norms more often. 55 However, self-rated rapport did not differ across dyad types, nor did it correlate with conversational theory-of-mind skills. Another study tested whether social difficulties stem from deficits in internal action modeling. 52 Autistic dyads showing weaker predictive coordination in a motor task, supporting this hypothesis. Finally, a study on social cognition, skills, and motivation found these factors minimally predictive of actual interaction outcomes in both autistic and neurotypical adults. 34 This challenges the common assumption that deficits in these areas are primary drivers of social difficulties in autism. Instead, other factors (perhaps among the reviewed here social behaviors) may play a more significant role in influencing social interactions.
Discussion
We reviewed studies examining social behaviors and interaction outcomes in autistic, neurotypical, and mixed dyads to characterize social interaction styles in autism. Most research focuses on interpersonal dynamics across verbal and nonverbal domains, highlighting how autistic and neurotypical interactions differ in social coordination and expressive behaviors.
Although research on autistic interactions is still emerging, certain behavioral differences are evident. Autistic dyads generally show less mutual gaze, backchanneling, and reciprocal smiling and laughter than neurotypical ones, aligning with findings that neurotypical negative evaluations of autistic people are more influenced by nonverbal than verbal behaviors.18,147 In contrast, turn-taking appears similar across all dyad types. Other behaviors show mixed results; for instance, while social motor synchrony is lower in mixed dyads than same-neurotype dyads, it remains unclear whether autistic and neurotypical dyads differ in this aspect.
Literature remains diverse in the social behaviors studied (e.g., mutual gaze), interaction types (e.g., free conversation vs. goal-directed tasks), and social outcomes measured (e.g., rapport, willingness to interact again). 30 Few studies account for individual traits such as emotion regulation, cognitive biases, or empathy, which are known to differ between autistic and neurotypical individuals, and which can potentially influence results. We discuss these findings in the context of relational versus individual social difficulties in autism, considering sample variability and study design differences.
Relational versus individual social difficulties in autism
A central question is whether social difficulties in autism stem primarily from individual, autism-related social deficits (as proposed by the clinical model), relational challenges arising from neurotype mismatches (as suggested by DEP/DMH), or a combination of both. As summarized in Table 1, the clearest support for relational difficulties (“better” autistic than cross-neurotype interactions—but see a discussion of what makes an interaction “better” in the Social Interaction Outcomes section), comes from self-reported interaction outcomes. Conversely, studies comparing social behaviors in autistic and cross-neurotype interactions mostly find no differences. The clearest differences between autistic and neurotypical interactions in this review relate to interpersonal behaviors, such as mutual gaze, back channeling, and reciprocal smiling and laughter. Finally, some studies relating social behaviors to interaction outcomes suggest that autistic and neurotypical individuals may rely on different social cues in building rapport and other interaction outcomes.
In the broader autism literature, one line of evidence for relational difficulties comes from studies showing that interactions between autistic individuals are often more successful than mixed-neurotype interactions. For example, in line with numerous first-person accounts,27,148–150 same-neurotype dyads tend to report higher rapport and greater willingness to interact again compared with cross-neurotype dyads.17,40,51,56,57,59,151 Similarly, both autistic and neurotypical students prefer friendships with peers of their own neurotype, and this matching predicts relationship strength better than popularity or neurotype alone.151,152 Similarly, matching level of autistic traits in the general population predicts positive social outcomes and friendship.153–155
Further evidence for relational difficulties comes from studies showing that neurotypical individuals also contribute to communication breakdowns with autistic partners. They tend to blame autistic individuals more than themselves for misunderstandings and rate them as more egocentric, 156 undervalue autistic people’s social competence, 157 overestimate how helpful they themselves are, 158 struggle to interpret autistic facial expressions, 159 behaviors, 122 and mental states, 121 and have problems “reading” autistic peers leading to less favorable social ratings.18,120–122 These biases lead to the rapid formation of negative impressions, reducing willingness to engage with autistic individuals16–18 and increasing their risk of social isolation and loneliness, connected to poor mental health.160–162
However, some evidence suggests that autistic individuals face social challenges even in the absence of neurotype mismatches. For example, autistic individuals may struggle to anticipate their partners’ actions, regardless of neurotype 52 and they tend to evaluate less favorably other autistic people than neurotypicals. 17 Autistic children show less inclination to interact with autistic than neurotypical peers. 62 Finally, difficulties in peer interactions among autistic children and adolescents have been observed even when controlling for their level of social and emotional symptoms. 163
Together, the literature provides evidence for both individual and relational social challenges in autism. Autism may involve inherent social alterations, as suggested by the clinical model. Alternatively, both neurotypes may be equally adept at interacting with those who share their social style, but heterogeneity in autism can still cause mismatches. For example, idiosyncratic social behaviors often observed in autistic individuals—such as differences in facial expressions 159 —can increase the likelihood of mismatched interaction styles, even within the autistic community. From this perspective, social challenges in autism are largely relational but affect both same- and cross-neurotype interactions. Ultimately, it may be simply too early to draw definitive conclusions given the diversity of measurements, tasks, and sample characteristics in current research.
Individual differences in social behaviors
None of the measurements reported here, including those showing robust group effects (e.g., back channeling), clearly differentiates between autistic and neurotypical individuals. For example, Wehrle 80 found that while all autistic dyads in his corpus differed from the average neurotypical dyad, the differences were spread across various parameters. Notably, reducing the social difficulties autistic individuals face to group-level neurotype differences risks “behaviorizing rather than humanizing” them, that is, saying that autistic people behave in a certain way “because of their autism,” which undermines their agency and disregards the individual motivations. 29 Similarly, applying neurotypical standards to autistic needs can be misleading and potentially harmful. For example, some autistic people may prefer less frequent social contact than the average neurotypical person, but this reduced social motivation does not diminish their humanity. This underscores the need for individual- or dyad-level investigations. For example, Wood et al. 164 found that laughter in interactions depends more on an individual’s tendency to laugh than their partner’s behavior, underscoring the importance of individual-level factors. Yet, only one study using a dyadic design for autistic interactions included multiple interactions per participant allowing to model individual tendencies. 60 Such designs would also allow to investigate how individual behaviors of each neurotype change depending on interactions in the same or mixed dyads. Also, mixed-method approaches combining qualitative and quantitative measures, yield valuable insights. 26
Sample characteristics and design aspects
Most reviewed studies focused on interactions between verbally communicative autistic individuals with typical intelligence and no learning disabilities, who are mostly English-speaking. Expanding research to other languages and including individuals who communicate in different ways (e.g., with aided augmentative and alternative communication), and who have varying support needs is crucial for a broader understanding of autism. However, autistic individuals with fewer support needs are more likely to be held to neurotypical standards, leading to negative social judgments. While neurotypicals tend to form more positive judgments when they are informed their partner is autistic, 16 such improvement in ratings occurs regardless of whether the person is actually autistic. 39 This raises the question of whether knowing the diagnosis status promotes genuine tolerance or just inflates reported first impressions, making it a potential confounding factor. Many of the reviewed studies here considered this, with two specifically testing it50,58 (see Supplementary Data).
Another key consideration is autistic masking (or camouflaging), the suppression of natural autistic behaviors to appear more “neurotypical.” While masking influences social interactions, few studies account for it, 148 and none has explored its impact in same versus cross-neurotype dyads. Although the role of camouflaging in social interactions is a complex and context- and individual-dependent phenomenon, we encourage future studies to, as a minimum, statistically control for quantifiable camouflaging behaviors (e.g., using the Camouflaging Autistic Traits Questionnaire 165 ).
DMH suggests that differences in interaction styles between autistic and neurotypical individuals may grow over time, emphasizing the developmental perspective. 27 While the reviewed studies included children, adolescents, and adults (see Supplementary Data), no clear age differences in social difficulties emerged. At the same time, some studies highlight short-term changes, reporting increased behavioral alignment over the course of an interaction, although at a slower rate in autistic dyads,41,42,80 or increased preference for same-neurotype peer interactions over months. 151
Given neurotypical biases against autistic individuals, observer neurotype is crucial in behavioral coding studies (see Supplementary Data for information on coders’ neurotype in the reviewed studies). While most use neurotypical raters,16,120,122 observer characteristics may influence ratings more than the behaviors themselves. 16 Studies with autistic raters explore shared social judgments (higher rapport ratings for autistic dyads, 56 lower naturalness ratings for autistic individuals 125 ) and differences in perceived interaction outcomes (while autistic actors received lower ratings overall, autistic raters were more willing to engage with them in the future 38 ). Notably, the observation that both neurotypes rate autistic individuals less favorably challenges the common assumption that autistic individuals are less sensitive to social cues.
In terms of study design, there are considerable differences in the nature of the interaction task used in the published studies, from free 50 or semistructured conversations, 48 through goal-oriented conversational exchanges, 80 to monologic highly structured verbal tasks57,58 or nonverbal collaborations. 54 This methodological heterogeneity limits the comparability of studies.
Finally, most studies label comparison groups as “neurotypical” or “typically developing” based solely on the absence of an autism diagnosis, often via self-report. However, neurodivergence extends beyond autism, so “non-autistic” should not be equated with “neurotypical.” We encourage future studies to screen for other forms of neurodivergence and caution readers to interpret findings with this limitation in mind.
The importance of the autistic perspective
While a behavior may serve a specific function in neurotypical interactions, its reduced expression in autistic individuals does not necessarily lead to poorer social outcomes. For example, frequent mutual gaze is associated with higher rapport in neurotypical dyads but not in mixed or autistic dyads. 40 Similarly, social motion synchrony carries less social meaning in autistic interactions,50,166 and although both neurotypes rate autistic individuals less favorably, only other autistic individuals are more willing to engage with them in the future.17,38 These examples highlight that the same behavior can have different functions or interpretations across neurotypes, suggesting that autistic social outcomes are less reliant on behaviors that drive neurotypical judgments. This could indicate that neurodivergent intersubjectivity is less dependent on rigid social norms and more on shared understanding with less demand for coordination. 26 Alternatively, autistic outcomes may depend on behaviors not typically measured in neurotypical interactions.
Thus, deeming either autistic or neurotypical behaviors as “correct” would unjustly empower one group over the other based on factors such as majority status or stigma/ableism. Instead, we must investigate social interaction differences and their underlying motivators. For instance, reduced back channeling in autistic individuals may be perceived as lack of interest by neurotypical interactants, but it may instead reflect a strategy to minimize distractions or a distinct communication style with a preference for explicitness. 80 It is essential to identify which behaviors best predict social interaction outcomes for both neurotypes and prioritize these in future research.
Outlook for facilitating cross-neurotype communication
To improve cross-neurotype interactions, we must understand the different social styles, including behaviors and underlying skills (social cognition, affect, and motivation). While interventions often aim to make autistic individuals behave/seem more “neurotypical,” 167 this often does not translate to social success in the real world.10,11 Communication improvements should involve both perspectives, not limiting the burden to the neurodivergent. 29 Some evidence suggests that it may be more effective to target the social skills of the neurotypical, not autistic people, to improve cross-neurotype interactions. 34 For example, neurotypicals who are better able to infer the mental states of autistic people tend to evaluate them more favorably.34,39,120 In that vein, one cognitive behavioral therapy manual 168 aims to support autistic individuals in guiding their communication partners through the following: (1) defining personal goals and success, (2) disclosing communication preferences (e.g., “I may not maintain eye contact, but this doesn’t indicate disinterest”), and (3) providing specific guidance (e.g., “Please verbalize emotions instead of relying on facial cues”).
Conclusions
The traditional, clinical view of autism has focused on deficits, particularly in socio-communicative areas. However, evolving perspectives driven by the neurodiversity movement and greater inclusion of autistic voices are reshaping this understanding. Evidence now emerges suggesting that social difficulties in autism may at least partially arise from mismatches in interaction styles between autistic and neurotypical individuals. This review analyzed studies comparing autistic dyads with neurotypical or mixed dyads, revealing both similarities (e.g., general turn-taking) and differences in social behaviors (e.g., reduced mutual gaze and backchanneling in autistic interactions). To advance the understanding of the different social interaction styles, research needs to consider both same- and mixed-neurotype dyads and incorporate subjective evaluations from both neurotypical and autistic perspectives. We encourage researchers to move beyond neurotypical-centric views by engaging directly with the autism community, ideally through participatory research or, at the very least, by discussing findings with autistic representatives to ensure a more nuanced and inclusive understanding of social interactions.
Footnotes
Authorship Confirmation Statement
M.M.: Conceptualization, methodology, visualization, funding acquisition, writing—original draft, and writing—review and editing. I.D., A.H., and T.W.:—Methodology, writing—review and editing, and supervision. This article has been submitted solely to this journal and is not published, in press, or submitted elsewhere.
Supplemental Material
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
