Synchrony Across Brains

Brief abstract

Second-person neuroscience focuses on studying the behavioral and neuronal mechanisms of real-time social interactions within single, but also across interacting brains. In this review article, we describe the developments which have been undertaken to study socially interactive phenomena and have helped to focus on behavioral and neurobiological processes that extend across interaction partners. More specifically, we focus on the role that synchrony across brains plays in enabling and facilitating social interaction and communication, in shaping social coordination and learning and how reduced synchrony across brains may constitute a core feature of psychopathology.

Introduction

The field of social neuroscience started out with the idea of studying the biological basis of ‘interacting minds’ (Frith & Frith 1999). In early years this was typically done by studying one brain at a time. Moreover, the experimental tasks typically used did not involve study participants in reciprocal social interaction, but rather in social observation, such that a study participant was asked to observe a face stimulus rather than interact with another person. The advent and growing availability of functional magnetic resonance imaging (fMRI) as a powerful, non-invasive technique accelerated the study of the neural correlates of social perception and cognition in subsequent years. Already in 2002, however, it was demonstrated that fMRI could be extended to studying both brains of interaction partners, which was described as ‘hyperscanning’ (Montague et al., 2002). To this end, two MRI machines were linked up technically in such a way that two study participants could simultaneously undergo neuroimaging while interacting by means of a computerized task. In its initial implementation, data analysis of these ‘hyperscans’ was conducted similarly to studies that had been performed in single brains, which raised the question whether hyperscanning could provide specific scientific advantages other than helping to collect neuroimaging data more quickly and when measuring two brains was really necessary to learn something new about social interactions (Konvalinka & Roepstorff 2012). Answers to this question have been given in different ways and using different methodologies, some of which will be presented in greater detail in this article and focus on studying the neurobiology of social cognition and behavior across the brains of interaction partners during live, real-time encounters or in cases where the experimental manipulation recreated an interaction sequence or tested whether participants’ brains respond similarly to social stimuli.

It is important to note that these developments and growing numbers of studies have been scaffolded by conceptual advances that tried to pinpoint why two-brain neuroscience was needed: Based on the idea that being in an ongoing, reciprocal social exchange with another person might be fundamentally different in terms of subjective experience, behavioral coordination and the underlying neurobiological mechanisms as compared to the situation of merely observing another person, the development of a ‘second-person neuroscience’ has been suggested (Schilbach et al. 2013). By drawing upon embodied, enactive accounts of social cognition, this approach has led to a stronger focus on studying socially interactive phenomena while investigating one person’s brain, but also meant obtaining neuroimaging recordings from both interaction partners (Redcay & Schilbach 2019).

In order to study the neurobiology underlying social interactions, it is necessary to use experimental paradigms that involve participants in structured or - ideally - ecologically valid, dynamically unfolding social interactions. The interactions can be real or perceived; however, they will always take place in real time and be reciprocal, such that one partner’s actions directly affect the other and vice versa. These important developments have yielded completely new insights into the workings of the so-called ‘social brain’. For instance, it has been shown that regions of the so-called ‘mentalizing network’ (or: ‘default mode network’, DMN), typically activated by explicitly asking study participants to think about the mental states of another person, also respond similarly during social interaction with or without explicit task demands to engage in mental-state reasoning (e.g. Schilbach et al. 2010; Redcay et al. 2010; Rice & Redcay 2016; Rice, Moraczewski & Redcay 2016; Alkire et al. 2018). In addition, second-person neuroscience studies have questioned the distinction between the mentalizing and action observation networks by showing integration between both networks during real-time social interaction (Schippers et al. 2010; Ciaramidaro et al. 2014; Sperduti et al. 2014). This raises the question why these networks might act in concert during social interaction, but not during observation? Studies investigating the inhibition of spontaneous mimicry give some insight by suggesting that mentalizing regions (particularly the medial prefrontal cortex) may act to control the automatic shared representations between social partners (cf. Wang, Ramsey & Hamilton 2011). Indeed, second-person neuroscience studies point toward greater resonance, or shared representations, during social interaction and have suggested exploration of these phenomena within interacting brains.

In other words, second-person neuroscience findings have challenged the traditional stimulus-response view of how the brain infers and reasons about others’ mental states, including their goals, intentions and beliefs. Furthermore, it has opened up new avenues for research that focus on collecting and analyzing data from the brains of all interaction partners. In fact, simultaneous dual-brain approaches are the only method potentially able to examine the emergent dynamics between two interactors in real time. Such emergent dynamics rely on the contribution of (at least) two autonomous agents and can only be described by using interpersonal measures that capture, for instance, how the gaze behavior of person A changes contingent upon the gaze behavior of person B (Leong & Schilbach 2019). Furthermore it has been shown that people behave differently during real-life social interaction (Becchio et al. 2010) and that interactions with trained confederates during lab-based experiments may not capture natural behavior (Kuhlen & Brennan 2013). Initial simultaneous dual-brain studies as introduced above, however, were limited in that they often used highly constrained tasks originating from game theory (such as economic bargaining games), which did not allow for freely forming, face-to-face interactions but rather relied on a sequential exchange of symbols. In recent years, paradigms involving more ecologically valid social situations including gaze-based and verbal communication tasks (Hirsch et al. 2017; Kinreich et al. 2017) have been developed and have been put to more frequent use. With simultaneous dual-brain approaches, researchers can, thus, identify inter-individual synchrony at the behavioral and neural level when two (or more) participants are engaged in a reciprocal and freely forming interaction. In other words, research has shifted towards the investigation of how real-time social interactions lead to synchrony across bodies and brains and it is to the discussion of this topic that we now turn.

Making minds more similar: Measures of interpersonal behavioral and neural synchrony

Behavioral synchrony.

Interpersonal coordination, i.e. coordinating one’s behavior with that of conspecifics, is recognized as an important human ability across different fields of research. It has been suggested that the act of keeping together in time with others may foster our social bonds (Wheatley et al., 2012; Vicaria & Dickens 2016). Furthermore, synchronized and coordinated behavior across individuals may help groups of individuals to act as single or social units, which allows them to achieve goals jointly that would otherwise not be attainable (e.g. Kourtis et al. 2019; for review see Sebanz & Knoblick 2021).

Behavioral synchrony can be created in multiple ways including synchronizing the same basic actions (e.g., walking together or imitating a partner’s actions (e.g., behavioral mimicry), or coordinating different actions with another person. In the category of synchronizing basic actions is the phenomenon of behavioral mimicry, which refers to the unconscious or automatic imitation of speech and movements, gestures, facial expressions and eye gaze (Chartrand & van Baaren 2009). Here, social psychology has demonstrated that producing the same behavior as someone else increases sympathy and rapport with others, unless one realizes that such mimicry is done intentionally and not spontaneously and can help to induce more complex forms of interpersonal coordination (e.g. Lakin 2013; Duffy & Chartrand 2015). Apart from mimicry in terms of observable behaviors, there is also evidence for mimicry on other levels, such as heart rates and pupils (Palumbo et al. 2016; Wohltjen & Wheatley 2021; Dumas et al., 2014). Furthermore, such instances of behavioral mimicry appear to be closely connected to what is called emotional contagion, i.e. the human ability to be affected and share affective states with others (Hatfield, Cacioppo & Rapson 1994). According to the perception-action model of empathy (Preston & de Waal 2002), emotional contagion and our ability to automatically track and integrate the bodily and affective states of another person can be described as a basic form of empathy, which may allow us to intuitively grasp what goes on in another person. Indeed, a large body of literature in social neuroscience demonstrates that observing others’ actions and emotions activates brain regions that are involved in generating these same behaviors or emotions oneself, which has been described as a ‘simulation’ or ‘mirror’ mechanism (cf. Rizzolatti & Sinigaglia 2010 & 2016).

Apart from studies that focus on the temporal link and coupling of simple and similar movements or actions, other studies have also studied task-directed complementary or joint action tasks, which address the notion of behavioral or interactional synchrony. Interactional synchrony refers to situations, in which people coordinate their movements, which may or may not be similar, to coincide with those of others. Here, one needs to produce actions, but also anticipate those of others in order to coordinate with them to produce ‘joint’ actions (Knoblich, Butterfill & Sebanz 2011). In a study by Richardson et al. (2015), dyads of participants performed a targeting task, in which they both moved computer stimuli without colliding with one another. Results demonstrated that participants were able to establish an asymmetric pattern of synchronous movement, which was essential to task success. In other words, patterns of complementary, interpersonal action synchronization can sustain more complex, joint actions. Interestingly, recent evidence in adults suggests that behavioral coordination across people also affects their cognitive systems, in particular those that are involved in reasoning about others’ mental states. A study by Baimel et al. (2018), for instance, demonstrated that physically moving together with others increased mental state attribution, and feelings of social connection specifically to those with whom participants had behaviorally synchronized. In other words, aspects of interpersonal coordination appear to go deeper than the skin and influence both emotional and cognitive processes of those involved by making them – in some cases – more similar and more susceptible to reading each other’s minds.

This brings us to the question whether we can also observe synchrony across persons at the neural level, and what such measures could help to explain. Below we discuss two forms of neural synchrony: neural similarity (NS) and interactive neural synchrony (INS) (see Definitions). Neural similarity is an “offline” measure of how similar two brains are while interactive neural synchrony is an “online” measure that captures alignment between brains in real-time.

Neural similarity.

Seminal work by Hasson and colleagues has repeatedly demonstrated that in early sensory areas shared neural patterns across individuals, so-called intersubject correlations (ISC) - a type of neural similarity - are coupled to the low-level properties of the stimulus, while in high-order brain areas, shared neural patterns are coupled to high-level aspects of the stimulus, such as meaning (e.g. Silbert et al. 2014; Chang, Nastase & Hasson 2022). Nummenmaa et al. (2012) used fMRI to investigate brain activity while participants were individually watching movies depicting unpleasant, neutral and pleasant emotions. Results demonstrate that during movie viewing, participants’ brain activity was, indeed, synchronized both in lower- and higher-order sensory areas, but that valence ratings obtained after the movie watching also showed increased ISC, which suggests that emotions play a particularly important role for binding people together, because they strongly contribute to similarity in neural responses across individuals, i.e. to stronger synchrony across brains, which might facilitate social interaction and communication. In another set of ground-breaking studies, Parkinson et al. (2018) investigated whether familiarity, social network proximity and friendship with other persons is related to interpersonal similarity of neural responses obtained during movie watching. Results, indeed, demonstrate that neural responses are exceptionally similar among friends and that the similarity in neural responses, in fact, decreases with increasing distance of the participants in real-world social networks. An additional study demonstrated that the extent to which neural responses are synchronized across persons when viewing naturalistic stimuli is closely related to their personality profile and may further reflect related similarities in the interpretation of the seen stimuli (Matz et al. 2022). An open question is whether neural similarity leads to more social interactions (and ultimately friendships) or whether greater social interactions leads to greater neural similarity. Recent work suggests the latter as NS during movie viewing was greater following conversation between strangers (Sievers et al., 2024). Additionally, conversation increased neural synchrony during movie co-viewing following conversation between partners that was unrelated to the movie (De Felice et al., 2024). As these examples demonstrate, completely new insights into the working of the ‘social brain’ and its functional relevance for real-life social interactions and social relationships have been made possible by focusing on measures of synchrony across brains without actually synchronously measuring two interacting brains (Figure 1).

Figure 1. Interactive vs non-interactive neural synchrony.

Neural similarity refers to how similarly a pair or group of individuals process the same audiovisual stimuli. Because individuals are not in a real-time interaction this reflects similarities in existing brain function and organization rather than a real-time alignment between partners. Interactive neural synchrony (INS) refers to synchrony calculated during a real-time interaction using a variety of methodological approaches. INS can reflect existing neural similarity as well as conceptual alignment and behavioral coordination between partners that emerges over the course of the interaction. Unlike neural similarity, INS is also influenced by the social partner’s behavior in real-time. Social partners who begin their interaction with higher neural similarity will also show higher INS during interaction. Further, repeated interactions with high INS with the same social partner have the potential to create more neural similarity between partners (illustrated by arrows between terms in the figures).

Interactive neural synchrony.

Interactive neural synchrony (INS) or hyperscanning studies as described above, however, are able to synchronously measure two interactive brains by examining coherence in neural activity and behavior in two (or more) persons in the context of a live, reciprocal social interaction. Crucially, this approach helps to investigate how behavioral and neural processes in one person affect those, which are present and/or developing in the interaction partner. INS studies do not only increase the ecological validity of studies in social neuroscience, but also make it possible to investigate the neural mechanisms of how human beings communicate and affect one another during reciprocal social interactions by simultaneously focusing on the social exchange at the behavioral level and linking the behavior to synchrony across brains (Figure 1). Grounding INS in behavioral metrics of social exchange are important to disentangle co-created INS from the similar processing of the shared environment.

Studies measuring INS during real-time interactions demonstrate links between verbal or non-verbal behavioral features of a conversation and neural synchrony, For example, a study by Kinreich et al. (2017), for instance, used EEG hyperscanning to investigate brain-to-brain synchrony in 104 adults during a male-female naturalistic social interaction, comparing romantic couples and strangers. INS was found for couples in temporo-parietal regions, but not strangers and was linked to measures of behavioral synchrony, in particular the exchange of social gaze. In other words, using a micro-level analysis of social behavior revealed a tight link between synchrony across brains and bodies. Neural synchrony in this study was not related to speech duration or conversation content, which suggests that some aspects of INS may be driven by the non-verbal rather than verbal aspects of the social interaction.

INS also appears to be greater during periods of information sharing between partners. In a study using a complex, multimodal approach that included eye and face-tracking as well as functional near-infrared spectroscopy (fNIRS) during conversations in which participants did or did not disclose biographical information to each other, Canigueral et al. (2021) demonstrated a modulation both of social behavioral signals and brain activity. Specifically, participants gazed more at each other’s face and produced more facial displays during disclosure. At the neural level, greater brain activity was found in the temporo-parietal junction and INS was observed during the sharing of information. In other words, the ability to communicate information to another person, indeed, modulates both non-verbal behavior and brain activity patterns and, again, leads to synchrony across brains. Importantly, the analysis model used by Canigueral et al. accounted for task- and stimulus-driven effects, which suggests that INS was not merely driven by aspects of the tasks or stimulus input.

A ground-breaking study further demonstrates the importance of the temporo-parietal junction for INS. Bilek et al. (2015) used hyperscanning fMRI and an ingenious setup that allowed for immersive audiovisual interaction of the two participants as they were both lying inside (separate) MRI scanners. To characterize information flow between the two interacting brains, the authors adopted a data-driven approach to extract components from all participants’ data and were able to identify temporo-parietal junction as a brain region specifically coupled across brains during gaze-based social interaction.

Causal approaches to neural synchrony.

Dynamic causal modeling (DCM), relies on effective connectivity, or the influence one neural system exerts over another, and has recently been applied to hyperscanning fMRI data (Bilek et al. 2022). In this setup, correlated neural responses become data features that have to be explained by models with and without between-brain connections. This is an important addition to measures of functional connectivity across brains, because such correlative measures of synchronization between brains can be partly explained by exposure to the same sensory information. Measures of functional connectivity per se do not evaluate whether a connection between different brains needs to be assumed and how it is measurably instantiated. Bilek at al. (2022), therefore, plausibly suggest that we have to test empirically whether INS provides a better explanation for the neural data than single brain approaches. In their study, they use hyperscanning DCM, because DCM can be used to distinguish and quantify potential causes of correlation. In the context of two-brain hyperscanning data this means to evaluate both shared sensory input and effective connectivity between brains. Conceptually, the approach taken by Bilek et al. is related to the notion of generalized synchrony, i.e. the characteristic behavior of loosely coupled dynamical systems (Hunt et al. 1997), whereby knowing the state of one system would allow to predict the state of the other (Jiruska et al. 2013). Importantly, according to Bilek et al. (2022) generalized synchrony can only occur if there is formal or structural similarity between the coupled systems. In other words, two brains can only become coupled via generalized synchrony when they share the same sort of dynamical structure. Previous work by (Friston & Frith 2015) has addressed this question and used the active inference framework to generate simulations based on this premise. On this view, communication can be seen as a process between individuals that use the same model to process and attend to sensory input that is interchangeably produced by the interaction partners. Attending to sensations then allows for a shared narrative to predict sensations generated by another or to articulate the narrative oneself. According to Friston & Frith (2015), this produces a reciprocal exchange of sensory signals that induce a generalised synchrony between brain states in both agents. Following this logic and by using hyperscanning DCM, Bilek et al. (2022) demonstrate between-brain effective connectivity that was specific to social exchange in a two-person joint attention task and directed from the sender’s to the receiver’s right temporo-parietal junction. In other words, a causal connection was present between the two brains and necessary to explain the data, while accounting for the shared perceptual input of the two interaction partners. These findings highlight the importance and novel methodological opportunities of investigating INS beyond measures of functional connectivity and demonstrate how brain systems are dynamically coupled during reciprocal social interaction, whereby a sender’s brain - in a control theoretic sense - has a causal impact on the receiver’s brain. This causal connection may reflect processes of conceptual alignment to one’s partner, representation of affective states, or increased prediction of a social partner’s actions (see ‘Mechanism’ section below). Future exploration of the causality of cross-brain phenomena may profit from the use of multi-brain stimulation (cf. Novembre & Iannetti 2022).

Synchrony across brains enables and facilitates social interaction & communication

As discussed above, synchrony across brains comes in different forms and flavors. In this section, we focus on work that demonstrates how communication relies upon interpersonally shared similarities in brain activation patterns (i.e., neural similarity) and upon integration with neural networks that evaluate incoming information against prior and conceptual knowledge to align between partners and generate a neural signature of shared experiences (i.e., interactive neural synchrony). We also discuss possible cognitive explanations for the occurrence of INS.

Synchrony across two brains can occur when two persons – who have sufficiently similar brains – are being exposed to the same signal or stimuli. This synchrony emerges early and is foundational to the development of communication. As a matter of fact, human beings have the spontaneous tendency to experience (and know) the world together. Importantly, this is what the word consciousness stemming from the Latin term ‘conscientia’ meaning “knowledge shared with others” used to mean (Charlton & Short, 1879). In other words, falling into synchrony with other brains appears to be an important aspect of the sharing of experiences. This sharing of experiences and understanding is fundamental to the emergence of any communication system as a signal’s meaning must be shared to be able to communicate (Wittgenstein 1973). Such a shared understanding of signals and objects in the world is developed early during human development by means of various practices that can be subsumed under the term ‘shared intentionality’ (cf. Rakoczy & Tomasello 2007). The first instances of shared intentionality are already apparent during the first months of life when infants demonstrate responsiveness to the reciprocity of face-to-face interactions with the caregivers. Later the interactively constituted phenomenon of joint attention emerges, whereby interaction partners actively coordinate their attention towards aspects of the environment, share experiences and create a ‘common ground’ that is known to be highly important for the development language (e.g. Mundy et al. 2007). In other words, forms of (increasingly sophisticated) communication emerge through embodied social interactions and the immediate context they create (Galantucci 2005) and typical adults continue to exert this tendency to align their behaviors and perspectives with others. This alignment or shared intentionality that emerges from repeated interactions shapes similarity between brains, leading to greater inter-brain alignment in future interactions between social partners (De Felice et al., 2024) (Figure 2).

Figure 2. Correlates and consequences of neural synchrony.

Shared audiovisual stimuli, behavioral coordination, social and affective cues, personality characteristics, and cognitive alignment all affect the degree of neural synchrony between partners. Contextual factors such as affect, relationship status (e.g., parent vs. stranger), proximity between partners, and goals (e.g., cooperative vs. competitive) also influence the degree of synchrony. Synchrony also leads to many positive outcomes including increased social learning, improved communication and cooperative performance, greater prosocial behavior, shared enjoyment of an interaction and formation of relationship bonds as well as the development of cognitive and affective abilities and mental health symptoms. However, most studies provide only correlational evidence of these associations so directionality of these effects can not be determined. Further, these interactions are likely reciprocal, for example, dual-brain transcranial magnetic stimulation studies have shown that neural synchrony can induce coordinated body movements, suggesting bidirectional relationships between these proposed mechanisms and outcomes with neural synchrony.

Conversations are characterized by various forms of synchronization: for instance, breathing patterns during conversation are correlated. Also, conversants tend to coordinate postural sway and eye gaze, even when they cannot see each other (Shockley et al. 2003; Richardson et al. 2007). During conversations, people are also likely to synchronize their word use (Garrod & Pickering 2004; Ireland et al. 2010). In other words, speech signals seem to be particularly efficient at creating interpersonal alignment at various levels, i.e. the bodies of the conversation partners get coordinated in ways that they weren’t before the interaction started. And linguistic coordination is also an important cognitive tool that helps human beings to cooperate and reach better performance levels than an individual could (Bahrami et al. 2010).

Using neural similarity approaches, Hasson and colleagues have demonstrated that during successful communication, the speaker’s and the listener’s brains exhibited joint, temporally-coupled, response patterns (Stephens et al. 2010). Speaker-listener coupling in early auditory regions was shown to reflect shared processing of low-level acoustic properties of the stimulus. By contrast, in higher-order brain regions the responses in the listeners’ brains lagged behind the speakers’ brain. Most interestingly, speaker-listener coupling in language and higher-order social cognitive regions reflected communication and shared understanding of narratives. In addition, it was shown that NS in higher-order brain regions was not observed when communication is disrupted (Silbert et al. 2014) and that the quality of the communication correlated with the degree of interpersonal similarity in brain responses (Dikker et al. 2014). In this context of language-based communication the development of concepts that integrate and represent our past experience is also highly important as a mechanism, by which the brain makes meaning of sensations and which can be used to communicate experiences, thereby potentially leading to shared or synchronized neural activity arising from the tendency of our social brains to align thoughts and create meaning (Yeshurun, Nguyen & Hasson 2021). Indeed, the degree of NS between two social partners has been shown to predict their communication success in a communication game. In that study, neural similarity interacted with empathic abilities such that for those who were more neurally similar, empathic abilities did not impact communication; however for those who were less neurally similar empathic abilities were important in predicting communication success (Dziura et al., 2023).

Building on studies that use single-brain neuroimaging to investigate neural similarity, emerging work also uses hyperscanning and dual-brain neuroimaging to investigate how partners build common ground or mutual understanding, during interaction. Stolk et al. (2013) asked participants to jointly reproduce a spatial configuration of two tokens on a digital board and compared a so-called communicative and an instrumental condition. In the communicative condition, the goal of the communicator was to make sure that both his token and that of the receiver were arranged according to a configuration visually presented only to the communicator. This required that the communicator used the movements of his token to signal to the receiver how she should configure her own tokens. This simple task was effective in reliably inducing dyad-specific communicative behavior. In other words, the same movements were used by different dyads to signal different meanings. Interestingly, the communicative condition elicited more mutual adjustments by the interaction partners. As a key finding, Stolk et al. demonstrated that the communicative condition elicited comparable neural responses in medial prefrontal cortex and anterior temporal lobe in both communicator and receiver, which again highlights the importance of synchrony across brains. Stolk et al. (2014) followed up on their findings by using the same task for two-brain neuroimaging. Here, they found that cross-brain correlations in superior temporal gyrus are stronger during communicative episodes, which demonstrates that INS between communicators is relevant for the development of shared meaning and concepts. Similarly, a study by Liu et al. (2023) used a ‘coordinating symbolic communication paradigm’, in which two communicators are required to create an interpersonal communication system. In dyads that were able to establish communication, significantly increased levels of INS were found in right superior temporal gyrus. Furthermore, positive correlations between INS and measures of shared intentionality and communicative accuracy were found. The authors also used transcranial alternating current stimulation (tACS) to further explore the potentially causal role of INS enhancement for communicative success. In-phase stimulation, in fact, led to an enhancement of INS in the right superior temporal gyrus and resulted in higher communicative accuracy as compared to sham or anti-phase stimulation. Thus, INS appears greatest during periods of communication and causal approaches suggest this neural synchrony may be a mechanism for communicative success.

Reciprocity of social interaction affects synchrony across brains

Hyperscanning studies also demonstrate how the reciprocity of social interaction contributes to synchrony across brains (e.g., Dumas et al., 2010). An fNIRS study by Fishburn et al. (2018) found greater INS in participant pairs when they completed a puzzle together in contrast to a condition where an identical puzzle was completed individually. In addition, it was shown that the time course of neural responses of one person predicted that of their partner, but not that of another person completing the puzzle individually. A recent study by Koul et al. (2023) demonstrated that INS emerges spontaneously and can be predicted by the natural occurrence of dyadic behavior that is typically shown when human beings are in the presence of one another, such as reciprocated eye-contact, body movement and smiling. Importantly, Koul and colleagues ran control analyses to ensure that INS was not simply a by-product of individual EEG variation, but that it rather reflected dyad-specific neural dynamics. They did so by comparing models, in which an individual behavior of one participant would be sufficient to induce INS or alternatively where such behavior needs to be reciprocated and occurs in both persons simultaneously. Results, indeed, demonstrate that reciprocated social behaviors predicted INS better than unreciprocated behaviors. It is important to note that the observed phenomena in Koul et al.’s study were found in the absence of a structured social interaction task, which is consistent with the idea that human beings may have a natural tendency to socially connect with others (cf. Coan & Sbarra 2015).

In addition to the affiliative benefits of synchrony, the communicative benefits of INS might also be tractable at a computational level. According to the Bayesian brain hypothesis and predictive coding accounts, brains are probabilistic prediction machines that build up mental models of the external world to predict and explain incoming sensory information. Importantly, such mental models need to be continually updated in order to reduce the so-called prediction error, i.e. the difference between the model’s prediction and the observed evidence (Friston 2005). A predictive social brain should, therefore, attempt to predict another person’s social behavior and observe what the person actually does (Lehmann et al. 2023). Such predictions and expectations of social behavior can sometimes be so strong that they can lead to false positive social perception (Friedrich et al. 2022). In the context of social interactions, however, behavioral synchrony might be helpful for predictive processing, because it could help to predict the interaction more easily and might allow for the relevant mental models to become more similar over time, thereby reducing prediction error and contributing to social understanding (Mayo & Shamay-Tsoory 2024).

Group Dynamics.

Importantly, while most studies described above focused on dyadic synchrony, INS across members of a group is important for communication, learning, and cooperation and may feature different dynamics, for example, due to imbalance in leader vs. follower roles and in vs out-group factors. A study by Jiang et al. (2015) has investigated whether the occurrence of INS across individuals is linked to leader emergence, i.e. when and how initially leaderless small groups decide to select one person as the leader. Results demonstrate that INS for leader-follower participant pairs was higher than for follower-follower pairs. Also, INS for leader-follower participant pairs was higher during leader-initiated communication than during follower-initiated communications. INS was related to leader’s communications skills, but not communication frequency, which the authors interpreted as evidence for the importance of timing and qualitative aspects during social interaction that successful leaders seem to possess and which make them effective in having influence over others. In another important study conducted by Yang et al. (2020), measures of INS were investigated in a large fNIRS study, for which the authors organized 546 individuals into 81 three vs three-person intergroup competitions. They used in-group bonding manipulations and demonstrated an enhancement of INS, which led to participants to give more money to in-group members and to be more willing to give money to outcompete rivals. These results highlight the importance of INS, but also show that synchrony across brains does not always contribute to prosocial behavior, but can accentuate in-group vs out-group behavior. Although in-group bonding can increase INS, other work with teams has shown dissociable contributions between feelings of in-group identification and INS in predicting collective team performance. Specifically, Reinero et al., (2021) had participants wear EEG caps and solve problems either working with their team or individually. A strength of this study is that tasks were all done through a computer interface that was matched between team and individual conditions so any stimulus entrainment should be comparable across groups. Teams outperformed individuals and the collective performance of the team was predicted by whole-brain INS.

Synchrony across brains shape neuro-cognitive and social development & learning

Biobehavioral synchrony shapes socio-emotional development.

Biobehavioral synchrony is a defining feature of early social interactions between infants and caregivers. Already by 2 months of age, infants are sensitive to non-contingent responding from their caregiver (Murray & Trevarthen, 1985). These face-to-face interactions provide a foundation for the sharing of emotions, exchange of communicative signals, understanding of others and self, and development of self-regulation abilities. The majority of work on biobehavioral synchrony has been between parents and children (most often mothers and infants). Biobehavioral synchrony among parents and children is not simple mimicry, but involves a coordinated attunement between parent and child to their social signals, affective state, and communicative bids (Feldman, 2012). Typically this synchrony is coded in mother infant dyads by identifying periods of coordinated positive engagement in which the mother coordinates gaze and social touch during periods of infant positive affect, vocalization, and gaze (Atzil et al., 2011; Leclere et al., 2014). Maternal sensitivity, or parental responsiveness, promotes coordinated, synchronous interactions between child and caregiver and has a significant effect on children’s development of social-interactive and cognitive abilities (Landry et al., 1988; Legerstee et al. 2007) and their emotion discrimination and regulation abilities (Yaniv et al., 2021; Feldman, 2012; Bell, 2020). Synchrony between parent and child can be measured in terms of this behavioral coordination as well as physiological arousal (e.g., respiratory sinus arrhythmia – RSA) (Bell, 2020). RSA synchrony between mother and child is linked to their affect during the interaction (Capraz et al., 2023; Han & Tronick, 2009), is affected by risk status such as a history of maltreatment (Miller et al., 2023). These effects of biobehavioral synchrony are long-lasting: maternal-infant behavioral synchrony predicts neural discrimination of emotions in adulthood (Yaniv et al., 2021). Outside of the parent-child context, synchrony between young children promotes prosocial behavior (Kirschner & Tomasello, 2010) and infants are more likely to help an adult if they previously engaged in synchronous movement with that adult (Cirelli et al., 2014). Thus, synchronous coordination of behavior and affect between child and parent or peer has powerful and sustained positive effects on social, emotional, and cognitive development.

With advances in non-invasive neuroimaging technologies, including fNIRS and EEG, synchronous neural activity (or INS) can be measured between parents and children (including infants) during real-time social interactions (Turk et al., 2022; Alonso et al., 2023; Wass et al., 2020; Nguyen et al., 2020). INS is higher when behavioral synchrony is high, including when adult and child are engaged in direct gaze (Leong et al., 2017; Piazza et al., 2020) and during turn-taking in natural conversation (Nguyen et al., 2021). Affect also modulates INS such that periods of high positive (but not negative) affect synchrony relate to high INS within medial and lateral frontal and temporoparietal brain regions (Santamaria 2020). Further, maternal sensitivity, or the coordinated attunement of the mother’s behavior to the child’s, predicts mother-infant neural synchrony while maternal intrusiveness, in which the mother engages the child during non-receptive periods, predicted lower synchrony (Endevelt-Shapira & Feldman, 2023). From these data, INS could be interpreted as simply a complementary approach to identify behavioral or affective synchrony, but one that is more objective and does not require detailed frame-by-frame coding. And, indeed, shared audiovisual environments and coordinated behaviors will produce similarities in neural activity that do not necessarily indicate a mutual understanding or real-time alignment between two people. However, beyond shared environments, neural synchrony can also reflect conceptual alignment that can not be identified based on behaviors alone. This conceptual alignment includes shared perspectives, goals, or affective states, and transfer of information (Hasson & Frith 2016; Wheatley et al., 2012) that occurs during these rich learning contexts for infants and children.

Methodological approaches can help tease apart synchrony that reflects dynamics between interaction partners beyond shared environmental features. For example, use of granger causality and graph theory can identify directional influences between parent and child inter-brain network organization (Box 1), and thus goes beyond effects simply due to a shared environment. For example, when mothers and infants were engaged in a social referencing task during EEG data recording, greater integration was found between nodes of the parent and child inter-brain network during positive compared to negative maternal affect (Santamaria et al., 2020). In other words, fluctuations in neural activity between regions of the child and parent were more similar when the mother was displaying positive affect. Using directed connectivity analyses, Santamaria et al., (2020) found mothers had greater effects on inter-brain density (i.e., roughly the degree of connections between brains) during positive affect whereas infants had greater effects on inter-brain density during negative affect. It is important to clarify that these neural “influences” during dyadic interaction occur via observable behavior (Semin & Cacioppo 2008), which then impact conceptual models, and thus incorporating behavioral coding into models of inter-brain directed connectivity in future studies will strengthen our understanding of the mechanistic role synchrony plays in emotion development. Lagged analysis approaches are another means to identify neural synchrony that goes beyond shared environment effects. Piazza et al., (2020) used lagged inter-subject correlation approaches with fNIRS and detailed behavioral coding during a face-to-face interaction in which an adult experimenter read and sang to an infant and revealed that behavioral synchrony (e.g., periods of mutual gaze) is preceded by activation within prefrontal cortex in the adult experimenter and child. Thus, here INS may reflect an anticipation of joint behavior, rather than a result of shared behaviors. In another example, a weighted phase lag index that avoids time 0 (i.e., simultaneous neural responses) to examine cross-brain synchrony identifies EEG phase coherence between brains that is not time-locked to sensory events but rather reflects the biobehavioral attunement between social partners (e.g., Endevelt-Shapira & Feldman 2023).

Synchrony methods box.

Approaches to Measure Neural Synchrony.

For a more extensive review and discussion of neural synchrony methods we point readers to Hakim et al., 2023. These categories were taken from their systematic review. Here we highlight common analysis approaches that are discussed in this review and used to investigate synchrony across brains.

Correlation –

Correlation approaches measure the temporal relation between time series, also described as inter-subject functional connectivity. This the most common synchrony approach used with fMRI data, though other modalities use it as well. Correlations can be non-interactive (neural similarity or inter-subject correlation) or interactive (INS; see Figure 1). They can encompass the whole interaction period or within time windows throughout the interaction to assess changes in synchrony over time. Correlations can be synchronous or lagged depending on the question of interest.

Regression –

Regression approaches relying on the General Linear Model (GLM) are more common with fMRI and fNIRS. Here one person’s brain activity is used to predict the other person’s in a cross-brain GLM. One can also look at time lags to identify lead-lag relationships between partners. A benefit with this approach is that behavior can also be included as a separate regressor in the prediction model.

Coherence –

Coherence is a measure of the correlation in the frequency or time-frequency domains between participants. This method is most common in fNIRS studies. One common approach is wavelet transform coherence (WTC) where time series data are transformed to the time-frequency domain so that correlations in frequency bands over time can be calculated.

Phase Synchrony –

This method is used only within EEG studies due to their high temporal resolution and examines the extent to which two signals are in phase with each other. One instantiation of this approach is the weighted phase lag index (wPLGI). This approach allows researchers to identify synchrony that is independent of shared environment effects by examining only non-zero phase lags.

Causality –

These approaches determine the causal influence of one brain on the other using approaches such as Granger Causality, Dynamic Causal Modeling, or partial directed coherence. Causal approaches are important in verifying that synchrony is an emergent property and due to the mutual influence between brains rather than alignment to shared external features of the environment.

Perhaps even more interesting than what drives INS is work demonstrating what INS predicts during development. As with behavioral evidence, synchronous neural activity during parent-child social interactions is related to child’s emotion regulation abilities (Reindl et al., 2018). Problem-solving or cooperation tasks provide a useful context to assess variation in synchrony and relation to emotion regulation. Overall, INS is higher during cooperative compared to competitive contexts (Reindl et al., 2018; Miller et al., 2019) and neural synchrony predicts problem-solving performance beyond behavioral synchrony alone (Nguyen et al., 2020). This improved prediction power from INS may be due to its ability to capture alignment in shared goals between partners. Further, neural synchrony is greatest during cooperation with a parent, compared to with a stranger, suggesting a pre-existing understanding of, or similarity to, one’s partner promotes synchrony beyond behavioral coordination alone (Reindl et al., 2018, 2022). Indeed, in sequential dual-brain studies, the similarity between parent and adolescent functional brain network organization predicts their similarity in emotional fluctuations throughout the day and adolescent’s emotional competence (Lee et al., 2017). Problem-solving tasks, such as the tangrams puzzle, can also be manipulated to induce higher or lower levels of frustration, and thus, provide an opportunity to test the hypothesis that child emotion regulation is shaped during synchronous parent-child interactions (Feldman et al., 2012). After preschoolers and parents completed a frustrating puzzle task they were given a recovery play period. Mother-child neural synchrony in the lateral prefrontal cortex during the recovery period predicted child (but not mother) irritable temperament, with lower INS predicting higher irritability. Higher child irritability was also related to lower behavioral synchrony between mother and child, demonstrating that child characteristics may also affect opportunities for synchronous contexts, leading to fewer opportunities to co-regulate and develop self-regulation as a consequence (Quinones-Camacho et al., 2020). Because this study was correlational and only one time point, it is not possible to disentangle directional effects of child irritability and interpersonal behavioral and INS between mother and child.

In the only study to date to examine longitudinal effects of neural synchrony on child socio-emotional or mental health outcomes, Quinones-Camacho et al., (2022) used the same frustration and recovery task design and demonstrated that greater parent-child PFC neural (but not behavioral) synchrony at 4–5 years of age predicted a more rapid decrease in internalizing behaviors over the subsequent year and a half. This first longitudinal evidence of relations between parent-child neural synchrony and behavioral outcomes is important in highlighting a potential causal role of parent-child neural synchrony in developmental outcomes. Future longitudinal studies can incorporate analytic techniques examining lagged coherence or directionality of influence across multiple cross-brain regions as well as behavioral and personality measures to test whether factors reflecting alignment between brains during interaction uniquely predict child outcomes beyond shared environments or child characteristics.

Neural synchrony promotes social learning throughout the lifespan.

One of the most powerful learning mechanisms, particularly in early life, is social learning (Herrmann et al., 2007). When caregivers and infants coordinate their attention together on objects of shared interest (i.e., engage in joint attention), infants learn about the object (e.g., the name of the object, its function, emotions toward the object, etc) (Mundy & Newell, 2007). Engaging in joint attention is itself a synchronous activity and involves ostensive and communicative cues of mutual gaze and pointing, shared affect, coordination of attention and mental states, and transfer of information between social partners. This social learning mechanism continues to shape our attention, knowledge, desires, and preferences throughout our lives (Mundy et al., 2018; Redcay & Saxe, 2013).

Interactive synchrony may facilitate learning, and social signals, such as mutual gaze, may be a mechanism driving this INS (Leong et al., 2017; Wass et al., 2020; Leong et al., 2021). Specifically, the Learning through Interpersonal Neural Coupling (LINC) hypothesis (Leong et al., 2021) suggests that ostensive cues like mutual gaze may reset the learner’s (or receiver’s) neuronal oscillatory rhythms to match those of the sender, which allows for optimal information transfer. This “resetting” would be reflected in greater interbrain phase synchrony (Leong et al., 2021; Leong et al., 2019). INS can, therefore, provide unique explanatory power beyond what could be learned simply from understanding single brain learning mechanisms (e.g., Leong et al., 2019; Pan et al., 2022; Dikker et al., 2021). For example, during a social referencing task, the likelihood of learning (i.e., the emotion associated with the object) per trial was related to greater INS between parent and child, but not the learning valence (i.e., the propensity of an infant to select positively or negatively labeled objects). The learning valence, on the other hand, was associated with infant intra-brain connectivity but not inter-brain connectivity (i.e., INS) (Santamaria et al., 2020). These findings highlight the importance of incorporating dual-brain, second-person neuroscience approaches to fully understand social processes, including social learning. How well we learn from others depends on us and them, and only dual-brain perspectives can identify and characterize that mutual influence.

While learning occurs naturally during social interactions, formal instruction as in a classroom, also relies on social learning mechanisms such as coordination of attention and representation of mental states for successful information transfer. These coordination processes may be reflected in INS between teacher and learner. Much of the work on social learning and INS has been conducted with adults or examining teachers and students and has shown that greater interactive neural synchrony between teacher and student relates to greater student engagement (Bevilacqua et al., 2018; Dikker et al., 2017; Davidesco et al., 2023) and better learning outcomes (Dikker; Davidesco et al., 2019; Pan et al., 2021; Pan et al., 2022; Davidesco 2023; Zhang et al., 2022). As with the study in infants, INS predicts learning even when intra-brain metrics do not (Davidesco et al., 2023; Davidesco et al., 2019).

Several mechanisms have been proposed for why INS relates to learning. Synchrony may reflect behavioral alignment and this alignment itself (rather than interbrain synchrony per se) may facilitate learning (Pan et al., 2022). Beyond behavioral alignment, INS may reflect two individuals in a shared attentional state. This shared attentional state may amplify processing of audiovisual input and thus facilitate learning and memory for those shared objects (Dikker et al., 2017; Shteynberg et al., 2015). An alternative explanation is that INS represents each person’s own neural activity and prediction of one’s social partner. During teaching, however, the teacher’s representation of and prediction of the student’s mind may be more critical for effective information transfer. Examining lagged relationships can identify these leader-follower patterns. For example, using fNIRS Zheng et al., (2018) predicted that the best teaching outcomes would be reflected in a lagged coherence between teacher and student’s brains because the teacher would represent the learner’s mind prior to effective transmission of information (Prediction-transmission hypothesis). Specifically, coherence between the teacher’s temporoparietal junction with that of the students anterior temporal cortex ten seconds later predicted better teaching outcomes (Zheng et al., 2018). Similarly, Pan et al., (2018) demonstrated that during an interactive song learning task that INS predicts song learning and that the instructor’s brain activity is best predicted by the learner. However, while INS may reflect these alignment and mutual prediction processes, it remains an open question as to whether interbrain synchrony plays a causal role in social learning. Novel multibrain approaches in animals and humans are beginning to shed light on the causal role of synchrony (e.g., Liu et al., 2023). Pan et al., (2021) used multi person transcranial alternating-current stimulation to show that INS is causally related to social learning. They synchronously stimulated the inferior frontal gyrus, an area important for song learning, between the teacher and student during the active song learning task. This synchronous stimulation led to spontaneous synchrony of body movements and improved learning outcomes. Further spontaneous body synchrony was a partial mediator of the relation between INS and learning outcomes. While this study contained only 15 students and thus caution is warranted in the interpretation, it provides a promising approach to identify causal mechanisms.

Reduced synchrony across brains as a core feature of psychopathology or ‘disorders of social interaction’

Psychiatric disorders are ubiquitously characterized by social difficulties. Furthermore, social interactions can either constitute a protective factor that contributes to quality of life and mental health or in the case of social stress and exclusion act as a risk factor that increases the probability of developing a mental health problem. Autism spectrum disorder can be considered as a paradigmatic case of a ‘disorder of social interaction’ (Schilbach 2016), because it is defined by difficulties in social interaction and communication. In spite of this most research has focused on single brains or single individual approaches to understand social challenges in autism. In contrast to this, we have suggested that it is important to address the importance of dyadic context in driving behavioral and neural responses to social stimuli and synchrony across brains in autism and other psychiatric conditions. Also, experimental tasks that focus on social interaction and are high in ecological validity are likely to be more sensitive in their objective assessment of those social impairments that are most therapeutically relevant and, when combined with computational approaches, may be used to develop more sensitive neural signatures of atypical social interaction (e.g. Lahnakoski et al. 2022). For example, social impairments have been shown to be less pronounced (or even completely absent) when two people with autism interact with each other than in a situation in which one person with autism and one person without autism interact. These clinical observations might be related to evidence indicating that the empathy shown by autistic individuals is greatest when it is directed towards others with autism (Komeda et al. 2015), potentially due to greater mutual understanding. Similarly, individuals without autism find it easier to infer the mental states of individuals without autism than of individuals with autism (Edey et al. 2016).

At a more abstract level, these findings may be taken to suggest that social impairments in autism, but also other psychiatric conditions, could be more closely related to (dis-)similarities between interaction partners than they are to the characteristics of each individual, which we have termed the social interaction mismatch hypothesis (Bolis et al. 2017; Redcay & Schilbach 2019). Evidence of greater social difficulties in dyads with more dissimilar partners may be because an interaction partner’s behavior can be more easily and accurately predicted when he or she is similar to oneself as previously discussed (cf. Friston & Frith 2015; Dziura et al., 2023). Behavioral research has, for instance, shown that interpersonal difference values of autistic traits are more closely linked to friendship quality than autistic traits per se (Bolis et al. 2020). In other words, similarity across interaction partners appears to be relevant for interaction success, which is consistent with a recent meta-analysis that demonstrates similarities across different variables for partners (Horwitz et al. 2023). Behavioral research has also shown reduced behavioral synchrony in dyads with an autistic individual (Glass & Yuill, 2023; review: McNaughton & Redcay, 2020). Consistent with the mismatch hypothesis, evidence suggests higher behavioral synchrony among autistic pairs than mixed neurotypes (i.e., autistic (AUT)-neurotypical (NT)) (McNaughton et al., in press; but see Georgescu et al., 2020).

Here, hyperscanning has the potential to provide unique new insights into ‘disorders of social interaction’ and might help to address aspects of heterogeneity in autism. In recent years, a number of hyperscanning studies have, in fact, studied INS in autism and the majority have demonstrated decreased INS for dyads consisting of AUT and NT individuals. Tanabe et al. (2012) used hyperscanning fMRI to investigate the neural correlates of live gaze-based social interactions between persons with and without autism. Results demonstrated that INS in right inferior frontal gyrus was greater in dyads of NT individuals than AUT-NT dyads. Quinones-Camacho et al. (2021) used fNIRS hyperscanning to investigate neural synchronization during conversations between a neurotypical experimenter and adults with or without autism. fNIRS measures demonstrated that neurotypical individuals showed more neural synchrony with the experimenter than autistic individuals in the temporoparietal junction. Less neural synchrony in the TPJ was associated with higher social impairments. Similarly, Key et al. (2022) have shown that lower levels of INS were associated with increased behavioral symptoms of social difficulties in autistic adolescents. Hirsch et al. (2022) have used fNIRS hyperscanning to investigate INS during in-person eye-to-eye contact and demonstrated reduced cross-brain coherence in autism. With regard to the studies that have demonstrated reduced INS in autism, the social interaction mismatch hypothesis might serve as a possible explanation for this finding, based on the assumption that interpersonal dissimilarity might disrupt processes of synchronization. In fact, recent work compared neural synchrony and behavioral patterns between dyads that were either matched or mismatched in autistic-like traits (i.e., high/high, low/low, or low/high). While dyads in which both individuals were high in autistic traits showed different communicative behaviors than the other two groups their neural synchrony was greater than between the other dyad types (Peng et al., 2024). Future research may help to further address this issue by systematically manipulating interpersonal differences across dyads in order to assess their impact on social interactions and their relationship to brain structure and function of both interaction partners. In addition, future hyperscanning studies should include AUT-AUT dyads to investigate whether similar or higher levels of INS would be observed compared to AUT-NT dyads, as suggested by the social interaction mismatch hypothesis (Bolis, Dumas & Schilbach 2022).

As discussed earlier, psychiatric disorders other than autism are also characterized by social interaction difficulties that could also be related to disturbances of behavioral and/or neural synchrony. Persons with schizophrenia (SCZ), for instance, are known to exhibit a variety of abnormalities in social perception, facial emotion recognition, mentalization and interpersonal coordination (Green et al. 2019; see Pan et al. 2023 for a recent review article). Aberrant social processing is known to negatively affect interpersonal interactions in schizophrenia, leading to poor social integration and quality of life (cf. Couture et al. 2006). A recent review of studies that investigate behavioral synchrony in SCZ has demonstrated synchronization impairments across different modalities, which are also found in relatives of persons with SCZ (Dean, Scott & Park 2021). Kupper et al. (2015) have demonstrated that the severity of so-called negative symptoms of SCZ, i.e. avolition, anhedonia, social withdrawal and affective flattening, are linked to less interpersonal synchrony. Negative symptoms are a core aspect of SCZ, do not respond well to antipsychotic medication (Correll & Schooler 2020) and account for a large part of the long-term disability and poor outcome of patients.

Other studies demonstrate that impaired behavioral synchrony between persons with and without SCZ can be improved by means of pro-social priming (Raffard et al. 2015), therefore, may point towards an important new avenues for research. With regard to the neural correlates of social deficits in SCZ, single-brain neuroimaging studies have demonstrated connectivity differences in the DMN, but also in the action observation network that is likely to contribute to interpersonal coordination (Saris et al. 2022). Furthermore, studies have implicated temporo-parietal cortex to be involved in deficits in controlling representations that relate to self and other that are commonly observed in SCZ (cf. Eddy 2016). Whether aberrant processing in temporo-parietal junction is related to INS differences in SCZ is not known. A recent study by Wei et al. (2023) has used fNIRS hyperscanning to investigate a neurotypical (NT) cohort and a clinical high risk (CHR) group of psychosis. Here, it was found that during a cooperation task the CHR-neurotypical dyads showed reduced INS compared to NT-NT dyads in right inferior frontal gyrus. Interestingly, reduced levels of INS in the CHR-NT group were linked to symptom scores of suspiciousness and persecutory ideas characteristic of the CHR status. Future hyperscanning research will help to understand the relationship of previously demonstrated activation differences in schizophrenia and their possible contribution to INS.

Depression is one of the most prevalent mental health conditions and is known to strongly affect social interaction behavior by leading to social withdrawal. Also, cases of chronic or persistent depression have been explicitly linked to social interactional difficulties. In fact, the Cognitive Behavioral Analysis System of Psychotherapy (CBASP) has been tailored specifically to meet the demands of this patient group, who are sometimes described as ‘disconnected from the social environment’ putatively due to difficulties in formative relationships, which affect their expectations of interaction partners. These difficulties can prevent chronically depressed persons from having the kinds of positive social experiences that would help to alleviate depressive symptoms and strengthen feelings of self-efficacy and self-worth. Indeed, depression at the neural level has been linked to brain networks associated with social cognition and action observation, but also recall of relationship episodes (Schilbach et al. 2014, 2015; Wade-Bohleber et al. 2020). Studies of behavioral synchrony in depression have, for instance, been conducted in the field of parent-infant interactions, where it is well established that parental depression negatively affects dyadic synchrony important for infant development (e.g. Golds, Gillespie-Smith, Nimbley & MacBeth 2022; Leclère, Viaux, Avril et al. 2014).

As we have seen, emerging evidence demonstrates that psychiatric conditions negatively affect or are associated with lower levels of interpersonal synchronization. Consequently, restoring synchrony across patient and therapist appears to be an important goal for psychotherapy, where patients and therapists are known to spontaneously synchronize their behavior, aspects of their voice and even physiological processes such as heart rate and where behavioral synchrony is linked to therapeutic success (see Atzil-Slonim et al. 2023 for a recent review). Koole & Tschacher have already in 2016 argued for an Interpersonal Synchrony (In-Sync) model of psychotherapy according to which synchrony plays a crucial role in shaping the so-called ‘therapeutic alliance’. Meta-analytic assessments have demonstrated that this alliance accounts for a robust portion of outcomes in individual therapy (Flückiger et al. 2018). The model by Koole and Tschacher (2016) suggests that patient-therapist synchrony may foster the therapeutic alliance and could promote adaptive emotion regulation abilities and outcome-related variables in the patient. According to the model, the alliance is grounded in the coupling of the patient’s and the therapist’s brain and behavioral synchrony helps to establish this INS. By drawing upon developmental psychology, Koole and Tschacher (2016) describe how interpersonal synchrony can be considered as a form of external emotion regulation that continues to be effective across the lifespan and constitutes an important part of psychotherapy. Work by Xie and colleagues (2016) in single brains has demonstrated that socially induced cognitive emotion regulation, i.e. a psychotherapist helping to down-regulate participants’ emotions, relies upon differential activations in key nodes of the DMN. In other words, it is conceivable that the DMN might constitute a network candidate for INS in relation to psychotherapeutic interventions. INS in the DMN might facilitate those complex social cognitive processes and shared mental representations that play an important role during psychotherapy and help to formulate goals and intentions needed for long-term changes. An important hyperscanning study by Bilek and colleagues (2017) has demonstrated that abnormalities of the DMN, in particular lower neural coupling of the temporo-parietal junction in patient-control dyads in a study investigating borderline personality disorder (BPD), can no longer be found when patients remit during psychotherapeutic treatment and no longer meet the clinical criteria of BPD. This demonstrates that hyperscanning may help to generate state-associated biomarkers for mental ill health, which track neural synchronization differences during treatment.

Mechanism of interbrain neural synchrony

As we have reviewed above, behavioral, cognitive, and personality factors drive interpersonal neural synchrony between social partners (Figure 2). Both interactive and non-interactive synchrony are affected by how similar two individuals are in terms of their personality traits and shared perspectives or conceptual alignment (Matz 2022; Yeshrun et al., 2017; Lahnakoski et al., 2014). For example, personality profiles (Matz et al., 2022), irritability (Quinones-Camacho et al., 2020) and intolerance of uncertainty (van Baar et al., 2020) predict neural synchrony (either NS or INS) between individuals. Characteristics of the social interaction also drive interactive INS. These include perceptual-motor features such as shared audiovisual input as well as the coordination of body movements in joint action. Ostensive or social signaling behaviors are particularly powerful drivers of synchrony, such as eye contact (Wass & Leong, 2020; Leong et al., 2017; Kinreich et al., 2017; Hirsch et al., 2017) or touch (Nguyen et al., 2021) and may serve to reset oscillatory rhythms of one’s social partner (Leong et al., 2017) (Figure 2). Interestingly, while eye contact is related to greater INS it serves to decrease pupillary synchrony (an index of shared attentional processing) during conversation (Wohltjen & Wheatley 2021) suggesting unique and potentially complementary mechanisms depending on the type of synchrony observed.

At the cognitive level, multiple non-mutually exclusive interpretations have been proposed to relate INS to cognitive processing between individuals (Wheatley et al., 2023). One group of explanations is grounded in conceptual alignment (Stolk et al., 2014, 2016; Hasson & Frith 2016). Conceptual alignment can reflect shared knowledge, shared goals, or shared affective state. During an interaction individuals come to a mutual understanding, or common ground, through repeated probing and updating of a shared conceptual space. This shared conceptual space is reflected in similar temporal and spatial patterns of brain activity that are on different temporal scales than sensorimotor events (Stolk et al., 2013, 2014, 2016). Individuals also can come into alignment affectively through building up a “shared space” of affect (Anders et al., 2011) over repeated communication in which each partner’s affective state is represented in the other’s brain. Conceptual alignment claims are particularly compelling when synchrony is examined over the course of interaction, for example, over multiple blocks of a communicative game (Stolk et al., 2014) or using dynamic INS (Li et al., 2021; Likens & Wilkshire 2021). Otherwise, alignment may reflect simply pre-existing neural similarity between partners. During any interaction a combination of existing neural similarities in perspectives and processing style combined with the ability to mutually align between partners will contribute to levels of INS (Figure 2).

Mutual prediction frameworks, discussed above, offer a related interpretation of INS that is grounded in behavioral action and prediction. During a social interaction both partners will represent both the actions of themselves and their social partners. The INS signal reflects the summed activity of the partner’s own behaviors as well as predictions of the other’s behaviors (Hamilton, 2021). Combining mutual prediction with active inference frameworks (e.g., Lehmann et al., 2023; Friston & Frith, 2015), Mayo & Shamay-Tsoory (2024) propose that social partners work to minimize the prediction error of themselves and their partner. Through this active inference process inferential models become more similar over time. And this similarity in inferential processes may be reflected in INS (cf. Friston & Frith 2015).

Behavioral Neuroscience approaches to identify mechanisms

While work in humans has begun to test causal hypotheses on the role of synchrony, this work is hampered by limits in the temporal and spatial resolution of methods available for recording during dyadic interaction and the degree of experimental control available to test hypotheses. Behavioral neuroscience methods allow for cellular level resolution, even while rodents are engaged in a social interaction. Behaviorally, rats and mice engage in interpersonal behavioral synchrony in ways that parallel human interactions. For example, social synchrony is seen in parent-infant or parent-pup behaviors in infancy during feeding and grooming as well as between peers as pups reach adolescence (Ham et al., 2022, review). Recently, research in bats, monkeys, and mice has demonstrated that INS is reflected at the neuronal level, provides better predictive power than behavior alone, and predicts future social interactions (Zhang & Yartsev, 2019; Tseng et al., 2018; Kingsbury et al., 2019). Further, recent methodological advances allow for collection of neural responses across the entire cortical mantle at cellular resolution from freely interacting mice using optical imaging methods (Scaglione et al., preprint 2024). While preliminary, this approach opens up avenues for more direct comparison of human and rodent synchrony, with the greater ability to probe mechanistic roles of synchrony in the rodent studies.

In an elegant series of studies with mice, Kingsbury et al., (2019) provide compelling evidence for the mutual prediction theory at the neuronal level – specifically INS is the product of neural activity reflecting both behaviors of self and prediction of other in both partners. Using calcium imaging to record from hundreds of dorsal medial prefrontal cortex (dmPFC) neurons across individuals simultaneously, they found correlations between dmPFC neurons when mice were engaged in social interaction. The correlation strength did not differ during periods of high compared to low periods of concurrent behavior suggesting shared behaviors alone were not driving synchrony. As in human studies, synchrony was greater when in social interaction compared to interactions with a barrier between mice, which suggests synchrony is not purely due to a shared environment. Importantly, using a cross-brain general linear model (GLM) approach they found that one animal’s neural activity could be predicted based on the behavior of both animals but that including the other animal’s (i.e., the social partner’s) neural activity in the model significantly improved prediction performance. At the level of single neurons, they found cells within the dmPFC that coded for specific behaviors of the mouse during the interaction while other neurons coded for behaviors of the interacting partner. These cells are spatially intermixed within the population leading to synchrony at a population level between brains. And in fact the cells coding the partner’s behavior had the greatest effect on interbrain synchrony. They found similar improvement in prediction when looking at the level of single cell recording when using “behavior” cells (i.e., cells within the dmPFC ensemble that code for specific behaviors of the interaction) rather than neutral cells. Overall, these findings demonstrate a neuronal mechanism for interbrain synchrony. That is, both partner’s represent their own and the partner’s behavior. This common behavioral repertoire leads to similar patterns of activity between brains and the degree of INS predicted future interactions between dyads (Kingsbury et al., 2019).

Conclusion & Outlook

As we have reviewed in this article, emerging evidence points towards the importance of synchrony across brains in order to enable, facilitate and realize social interaction and communication. We have seen how synchrony and shared experiences may emerge automatically during social interactions and how it affects subjective experience and the views we may hold. But we also see how we can work on synchronizing our brains by reaching consensus during conversations (Sievers et al., 2024) and using other forms of explicit communication to share thoughts and our mental models of the world (Frith & Frith 2024). Synchrony also waxes and wanes and is even disrupted. These disruptions are not necessarily bad, but can also be helpful by allowing for complementary and/or independent modes of thinking (Wohltjen & Wheatley 2021; Mayo & Gordon, 2020). Rather than always swinging in synchrony like pendulums, we can also resist synchrony or intentionally break it, when we try not to be influenced by the opinions and behaviors of others or social convention, which is also an important ability in many instances of human relations. Understanding how these moment-to-moment transitions in and out synchrony are beneficial depending on the specific context and goals of interaction will be an important area for future research (e.g., Mayo & Gordon, 2020).

Whether or not brains synchronize likely relies on sufficiently large and robust similarities in brain structure and function. Research has provided striking evidence for neural homophily, i.e. a tight link between familiarity and friendship between persons and similarity in brain activity when individuals are exposed to the same stimuli (Parkinson et al. 2018; Matz et al. 2022). Future research will help to further investigate how measures of brain structure and function, as well as cognitive and personality characteristics, across interacting dyads can help to predict who synchronizes with whom, how strongly, and how this relates to social interaction success. Such empirical work is needed to test hypotheses such as the ‘dialectical misattunement theory’ (Bolis & Schilbach, 2017) and the ‘double empathy problem’ (Milton, 2012), which are consistent with the idea that communication outcomes are due to the extent to which partners align in features such as their brain organization, lived experiences, understanding of each other, and communication styles. This work could help to scientifically substantiate the notion of neurodiversity, which - originating in the autism rights movement - is tied to the idea that all brains are to a degree unique and that differences across individuals may explain disabilities rather than deficits ascribed to individuals. It has also been recognized that the study of divergence in neurodevelopment should move away from conventional categorical differences and attempt to include and model the developmental dynamics that capture the emergence of differences (Astle, Bassett, Viding 2024).

Another important area for future work is understanding when synchrony is a mechanism or an epiphenomenon. Progress is being made in developing theoretical models of synchrony as a mechanism of social affiliation, communication, and learning (e.g., Leong et al., 2021; Hamilton, 2021; Friston & Frith, 2015; Mayo & Shamay-Tsoory, 2024) as well as approaches (in humans and animals) to directly test the causal role of synchrony in these processes (Liu et al., 2023; Kingsbury et al., 2019; Pan et al., 2021). Several models suggest that synchrony reflects greater predictive processing of one’s partner (e.g., Hamilton, 2021; Kingsbury et al., 2019; Mayo & Shamay-Tsoory, 2024; Friston & Frith, 2015). Studies that use computational approaches, which have been successfully employed to investigate and mathematically describe the cognitive and neural processes that underlie social perception and cognition in individuals, could be extended to study and mechanistically explain the emergence of synchrony during social interaction (Pott & Schilbach 2022; Bolis, Dumas & Schilbach 2022; Dumas et al., 2014). An important future direction will be continuing to formalize and test computational models that assess how predictive models of self and other are updated in real-time during social interaction and how (or whether) this updating reflects changes in behavioral and neural synchrony between partners in real-time, as well as longer-term changes in neural similarity between partners. Relatedly, longitudinal studies of synchrony are critical to test whether neural synchrony between peers or parent and child is predictive of developmental outcomes or social connection, respectively. However, currently longitudinal studies tracking the effects of synchrony are very limited (see Quinones-Camacho et al., 2022).

Finally, most INS studies are face-to-face but as communication increasingly occurs outside of face-to-face contexts it will be important to understand how synchrony may differ in these virtual contexts. Virtual interactions lose key aspects of co-presence, including body cues, eye contact, and even smell – features that are critical to social inference and social bonding (Endevent-Shapira et al., 2021). Further, work reviewed above suggests eye contact, touch, and interpersonal body coordination may drive synchrony. Indeed, preliminary work suggests neural synchrony is decreased during virtual, texting interactions compared to face-to-face, though information transfer is similar across contexts (Schwartz et al., 2024).

Taken together, social brains have the amazing ability to spontaneously and effortlessly synchronize. This ability appears to be fundamental for the human ability to become conscious, share experiences and communicate, which, in turn, modulates the degree of synchrony across brains. Being able to share and reflect upon our experiences of the world allows us to jointly develop mental models of the world, which we can use to transmit information, educate each other and engage in other culture-building activities that have transformed the world we live in. But synchrony across brains (or between groups of brains) may not take place or can even go awry and lead to misunderstandings and failures of communication that can have bitter consequences. Here, differences at the interpersonal level may be relevant and should be systematically addressed in future research including how synchronization differs depending on the neurotype match or mismatch between social partners. By doing so, future synchrony research will elucidate the factors that influence whose brains synchronize well with whom, the underlying mechanisms shared across individuals and the compensatory strategies and techniques, which might help to get communication back in sync, even when spontaneous alignment does initially not occur. In light of the many misunderstandings and conflicts that continue to characterize human existence, the potential relevance of this work is enormous and could help to alleviate human suffering by pointing towards mechanisms and techniques that support communication and reconciliation.

Definitions

Hyperscanning

The process of collecting neuroimaging data from at least two individuals while they are engaging in mutual interaction.

Simultaneous dual-brain approaches

This term is synonymous with hyperscanning and refers to approaches in which neuroimaging data is acquired from two individuals concurrently while in an interaction.

Social brain

A group of brain networks and regions that are associated with social processing. The default mode network is a core network within the social brain, but the social brain also includes networks associated with affect and reward processing, social salience, social perception, and action understanding.

Default mode network

This network includes bilateral anterior and posterior midline regions, temporoparietal junction, and anterior temporal lobes at its core. The default mode naming refers to the fact that these regions have higher metabolic rates at rest (i.e., no explicit task), reflecting their role in internally-directed thought. This network is also reliably engaged during tasks involving social processing.

Mentalizing network

This network overlaps with the default mode network but refers to regions that respond more when making judgements about another person’s mental state (or mental state reasoning).

Action observation network

This social brain network is engaged both when performing actions and observing others perform actions, suggesting an important role in self-other representation as well as imitation

Interpersonal behavioral synchrony

Interpersonal behavioral synchrony refers to the coordination of behavior between interacting individuals. This can be conscious or unconscious and involve mirroring (or imitation) or coordination of complementary actions.

Non-interactive synchrony (or Neural Similarity)

Neural similarity (NS) reflects how similarly individuals’ brains respond to the same stimulus and can be measured through sequential data acquisition.

Interactive neural synchrony (INS)

INS measures real-time neural synchrony and is thought to reflect a process of coming into alignment with an interactive social partner (though it can also be influenced by neural similarity between partners)

Intersubject correlations

This is a method in which brain response (or timeseries) between individuals are correlated. This is the dominant method used to measure neural similarity.

Respiratory Sinus Arrhythmia (RSA)

RSA is a physiological measure of heart rate variability linked with respiration and may reflect arousal and engagement.

Granger causality

Granger causality is a statistical approach to determine the influence of one time series over another and is used to look at causal influences between brains during interaction.

Graph theory

Graph theory is an approach used to characterize brain networks based on the strength and number of connections between nodes of the network. Some INS studies use graph theory to look at between brain network organization.