Abstract
Interpersonal synergy offers a conceptual and methodological framework for analyzing coordinated interaction in dyads and groups. The framework aims at an inquiry of how individuals contribute to a collective behavior and how, in doing so, they coordinate and co-adapt their actions, hereby establishing specific patterns of interplay. The present review retraces the origins of the notion of interpersonal synergy, assesses its scope, delineates synergies from other forms of interaction, and discusses possible mechanisms (such as mutual affordance responsiveness or shared intentions) that mediate the creation of such synergies. It then presents a contrastive “bestiary” of synergy types, which highlights the many shapes and forms in which the phenomenon occurs (e.g., collaborative vs. competitive, spontaneous vs. planned). The final two sections survey analytical approaches, from quantitative metrics which formalize “dynamic fingerprints” of collective dynamics to qualitative and mixed methods studies, which include 1st person perspectives on synergy formation.
Introduction
When tango dancers execute a complicated off-axis technique together, perfectly balancing out each other’s instability, when soccer forwards jointly thwart the defense through well-coordinated passes, when workers carry a heavy trunk together in a crowded stairwell, or when a group of academics discusses a topic in a mutually enlightening way, we can describe the underlying patterns of behavior as interpersonal synergies, which display a precise and task-specific coordination between individual actions.
The study of interpersonal synergy is an approach to behavioral analysis within the wider interdisciplinary enterprise of interaction research. 1 Briefly put, an interpersonal synergy is said to occur when people come to interdependently regulate their behavior and dynamically adapt it to some collective task in coordinated ways. Interpersonal synergies have been defined as “higher-order control systems formed by coupling movement system degrees of freedom of two (or more) actors” (Riley et al., 2011, p. 1). Similarly, Araújo and Davids (2016, p. 1) speak of a “collective property of a task-specific organization of individuals, such that the degrees of freedom of each [...] are coupled, enabling [them] to co-regulate each other.”
The aim of this review is to introduce the approach to a non-specialized audience. I will attempt to strike a good balance between conceptual reflections and an overview of existing approaches. The section “Origins and Scope” retraces the origins of the synergy concept and its application to interaction, before evaluating its scope and general merits. The section “What Defines Synergies?” turns to definitional dimensions of synergy, delineates synergies from non-synergies, and addresses cognitive and sensorimotor mechanisms that make synergistic interactions possible. The section “The Synergy 'Zoo'” discusses different manifestations of interpersonal synergy and cautions against over-generalizations from one topic of study to another. The last two sections (“Measuring Synergy” and “Micro-Practices and Meanings of Synergy Formation”) survey methods and perspectives that have been applied to interpersonal synergies, from macro-level inquiries of collective dynamics down to micro-level inquiries, typically with a more qualitative focus.
Origins and Scope
To get started, I will retrace the motivating assumptions of interpersonal synergy research, before discussing the concept’s possible application fields and scope.
A Mechanistic Science of Coordination
The word “synergy” is Greek for working together. Although it has become a well-worn concept in business and even everyday contexts, it is also a fruitful technical concept (e.g., Kelso, 2009; Latash, 2008). For an example from physics, a laser exemplifies a synergy of photons that have been stimulated to self-organize into a coherent wave (Haken, 1977). In biomechanics, synergy can be exemplified through coordinating limb segments in tasks such as hammering (Bernstein, 1967), manual grasping (Latash, 2008), or in coordinating speech articulators (Saltzman & Kelso, 1987). We can speak of synergy when there are “functional groupings of structural elements (e.g., nerves, muscles, joints) that are temporarily constrained to act as a single coherent unit” (Oullier & Kelso, 2009, p. 1537). This means that system components are interdependently configured and adjusted to produce an ensemble function. Note that this is a relatively abstract and scale-free definition, which can apply to the organizing principles of a broad range of multi-component systems from neurons and muscles, via whole-body behaviors, up to dyads, groups, and even whole societies. Across its various manifestations, the synergy perspective seeks to identify lawful patterns of interacting system components and how they are globally organized.
Interpersonal synergy, more specifically, provides a dynamic perspective on how people act as a combined unit to manage complex tasks rather than merely acting as separate individuals. We can speak of interpersonal synergy when two or more individuals are engaged in a collective action and meaningfully coordinate their contributions through an ongoing coupling process. In doing so, individuals must stay aware of the state of the collective and renounce many degrees of freedom they would normally use when acting alone.
Scholars of interpersonal synergy typically define organized behavior over one or several relational parameters expressing how the collective dynamics is coordinated—for instance, whether players on a soccer field move at stable angles or distances or if social dancers time their steps together. In principle, possible expressions of organized collective dynamics can relate to action synchronization, other periodicities of action, behavioral matching, or complementary behaviors that contribute to some complex collective function. Quantitative studies of interpersonal synergy (e.g., Fusaroli et al., 2014; Schmidt & Fitzpatrick, 2016) proceed by analyzing how a collective behavioral parameter behaves over time, often after performing data reductions through appropriate mathematical techniques. The behavioral time-series is checked for its process signatures, also referred to as dynamic “fingerprints,” which provide clues about the presence of interpersonal synergy (or not) and about its specific nature (see section 5). Prior to conducting such a study, researchers must make an informed decision about a useful collective parameter with realistic chances of “summing up” how a dyad or group coordinate their dynamics. This often requires “educated guesses” or the wisdom of experts and coaches who may, for example, have informally observed that the spatio-temporal coordination of two soccer forwards approaching the goal affects if they end up scoring.
Historically, inquiries into interpersonal synergy have roots in the coordination dynamic approach to motor control (see Schmidt & Fitzpatrick, 2016). Vial and Cornejo (2022, p. 2) sum up the origins well: Developments in the complexity research program, such as Haken (1977) contributions to synergetics and Bernstein’s (1967) to motor coordination were crucial for the early development of a dynamical systems account specifically designed for interpersonal coordination [...]. Concepts such as constraints, phase transitions and order parameters have allowed to consider interpersonal coordination through the lens of non-linear mathematical models—for example, the HKB model proposed by Haken et al. (1985). More recently, several authors have taken a step forward stating a model that uses complexity theories not only to facilitate non-linear methodological tools, but as a conceptual framework to describe the very nature of interpersonal coordination, which has taken shape in the interpersonal synergies theory [...].
Scholars of interpersonal synergy frequently aspire to provide a “mechanistic science of coordination” (Kelso, 2021, p. 305). They stress that interaction follows the principles of non-linear dynamic systems and aim at the formalization of measurable emergent properties of social systems, including functionality and robustness, as well as dynamic changes (Richardson et al., 2014).
Import and Scope
The general import of the interpersonal synergy perspective is that it views interaction behavior as more than the sum of its individual contributions; it recognizes that social interactions obey their own dynamics and can be conceived as emergent wholes. Hence, the framework does not reduce social cognition to “individuals in their context” as methodological individualist approaches would, but looks at interaction in terms of its emergent dynamics.
Interpersonal synergy is a concept of considerable breadth for social cognition (Marsh et al., 2006; Marsh, Johnston, et al., 2009; Marsh, Richardson, et al. 2009). Many expert skills such as dance, martial arts, rowing, or field sports have been researched, as have conversations and joint problem solving (discussed in detail later). Numerous other types of human behavior have not yet been researched but furnish intuitive candidates, such as play, making music together, stage interaction, sex, childrearing, or interactions with animals, such as horseback riding and driving husky sleighs. Corporate teamwork, rituals, collective worship, as well as the proverbial military esprit de corps equally seem to give expression to a synergistic ethos. Possibly, synergy is even a deep principle underlying organizational and societal dynamics more generally.
Although movement studies currently dominate empirical research, interpersonal synergy has been postulated to underlie various forms of extended cognition (Anderson et al., 2012), sensorimotor empathy (Chemero, 2016), human-tool systems such as the blind person’s cane (Dotov et al., 2010), conversational activity (Cummins, 2013; Fusaroli et al., 2014; Marsh, 2015), and linguistic sense-making (Di Paolo et al., 2018). For example, in brainstorming or joint problem solving (Wallot et al., 2016) different kinds of behavioral alignment and coordinated cognitive activity can be interpreted as synergy. Even the alignment of mental representations through conversation may be thus understood, especially when it stands in constant interplay with lower-level coordination of speech or para-verbal embodiment (Garrod & Pickering, 2009). Another line of research has shown that coupling behavior gives rise to neurological synchronization, which apparently enhances collaborative activity (Cacioppo et al., 2014).
The synergy framework is also affine to research of joint improvisation which presupposes constant mutual adaptation (e.g., Sawyer, 2003). Even the concept of distributed cognition (Hutchins, 1995) may—at least in part—be interpreted as a synergy with specialized, but coordinated contributions that incorporate tools, work ecologies, and social arrangements. That said, it remains open to debate how to assess conceptual or otherwise “abstract” synergies which intuitively seem to fall within the scope of the concept, but are not as easy to measure as physical action is. For example, it seems natural to think of two academic co-authors, who pool their writing skills and specializations, as profiting from “knowledge synergies.”
Cautionary Points
What theoretical and methodological scope interpersonal synergy should be accorded is a somewhat tricky matter. As Vial and Cornejo (2022) point out, we do not yet fully understand if social interaction ipso facto follows the principles of non-linear dynamic systems. They hit the nail on the head when criticizing that “we simply do not have enough evidence to claim that all naturally occurring interaction follows synergic principles” (p. 5). This untested assumption is further compounded by a “tendency to use synergy as a synonym for coordination” (p. 5), which conflates explanans and explanandum in inadmissible ways. The second problem is a theoretical neglect of subjective meaning. Vial and Cornejo warn that “social systems cannot obey the physical dimensions of natural systems” (p. 5) where personal affectivity, aspects of alterity and autonomy, and personal histories matter a great deal. We may add socio-cultural contexts and skills to the list of factors.
Furthermore, a certain tension between descriptive and causal meanings of the term synergy meets the eye. The definition of interpersonal synergy as “higher order control system” by Riley and colleagues (2011) implies a (self-organizing) generative mechanism that is held to cause behavioral coordination. 2 The underlying semantic ambiguity is well-worth acknowledging, because even if we know that something is a synergy from a descriptive standpoint the question which mechanisms underlie its creation may admit of different answers (see section “The Synergy 'Zoo'”).
Another potential problem is that the literature often equates synergy with a set of methods derived from complexity theory, creating the impression that the synergy concept literally “lives” in particular formalizations. This, however, seems to conflate a conceptual perspective with a particular choice of method. It is possible to add depth and nuance to quantitative findings through methods focusing on the subjective meanings and practices of synergy, as the section “Micro-Practices and Meanings of Synergy Formation” will show. (It is also not necessarily true that, as has sometimes been argued, lower-level aspects of complex systems are impossible to access).
Similar to Vial and Cornejo, I take a slightly cautious view of the scope of a “mechanics of interaction”, due to possible methodological limitations. Formalizing time-series data is easiest to do with behavioral measures such as movement, vocalization timing, or neural rhythms. An exclusive reliance on measurable data, however, may have limited construct validity for certain topics. That is, the dimensions that are easiest to measure are not always what is most representative of the phenomenon. In some cases behavioral measurements may offer but a weak proxy for it (e.g., if we wish to study “conceptual synergies” in joint problem solving) or represent only a small subset of what is meaningful to the participants of the interaction. A second limitation is that behavioral researchers most often base the analysis on, if we will, coarsely grained collective parameters. Whatever collective parameter is chosen, any such choice omits or even conceals other dimensions of interaction.
What Defines Synergies?
We now turn to three fundamental conceptual questions: What are the theoretical criteria for speaking of interpersonal synergy? How can interpersonal synergy be held apart different from other forms of social coordination? And what are possible sensorimotor and cognitive mechanisms that allow individuals to coordinate synergistically?
Which Dimensions Is a Synergy Composed of?
An interpersonal synergy is a collective dynamic that is organized at a macro-scopic scale in a relevant sense. The current behavior of a dyad or group can therefore be expressed with reference to the values of a particular collective or performance parameter, such as rugby attackers forming diamond shapes for the duration of an attack. When analyzing interaction data mathematically, this macro-scopic organization is a testable feature known as dimensional compression. We speak of a synergy if, and only if, a low-dimensional collective system is created (Oullier & Kelso, 2009; Riley et al., 2011). This presupposes that individuals are tightly enough coupled that “solo” degrees of freedom are renounced for the sake of the task. We may also think of this as data points of multiple individuals behaving with similar coordination that we would expect to see if the same “point cloud” belonged to a single person. Consequently, the collective organization can be parsimoniously described, as synchronized or periodic movement rhythms, shared velocities, specific spatial geometries, and so on. The low-dimensional organization that is expressed by the performance parameter can be interpreted as reflecting a functional dimension (Fusaroli et al., 2012), such that synergy subserves some task or meaningful ensemble behavior.
Riley et al. (2011) propose as second criterion reciprocal compensation between individual actors. This can mean that when the action of one individual changes (mistiming, too much of this, too little of that), then others adapt to “protect” the synergy. The ongoing dynamic adaptation is often thought of as stabilizing the collective variable, such as a specific attack or defense geometry in field sports. This dynamic stability is interpreted as evidence that components of a network are mutually responsive and organized as a synergy. Despite reciprocal compensation being having achieved a canonical status in the literature, I believe that interdependent adaptation is a preferable expression, as it is more inclusive. 3 The problem with reciprocal compensation is its being predicated on the assumption that a synergy is collaborative. Non-collaborative synergies (see section “The Synergy 'Zoo'”) display very different kinds of interdependent adaptation, such as exploiting errors, vying for dominance, or perturbing a dynamically stable interaction pattern. Furthermore, there are improvisational synergies in which actors maintain the interaction in other ways than just compensating for errors. They may, for example, opportunistically co-opt a short deviation for another contextually meaningful purpose (see section “Micro-Practces and Meanings of Synergy Formation”).
In addition to dimensional compression and interdependent adaptation, recent research points to two further analytic levels of synergy research:
Sharing patterns (cf. Latash, 2008) designate specific contributions of a person to a task, for example, in ball sports players in a defense or offense role. Complementary contributions may be distributed in role-specific ways to begin with, as in social dances. In other cases, complementary roles may spontaneously crystallize, for example, when people solve problems together. More generally, sharing patterns are a notion that helps us contrast modes of interplay and ways that persons are relationally organized. They qualitatively characterize the contribution of each component to the whole and; they define the specific co-dependency relation. Sharing patterns are a mid-level concept that allows us to understand the manner of interplay between individuals within a specific coordinated behavior.
Lastly, we can ask which components are part of a synergy. This may concern which players in a larger team contribute to a specific task such as an attack, but it may also concern lower-level elements of individual action systems such as the body parts a person uses for a collective task. Relevant questions attaching to the component-level concern which components are functionally most central for a synergy, which are more variable, and if a particular component contributes to many other possible component assemblies, a feature known as “degeneracy” (e.g., Hristovski et al., 2012). As will be argued later, a component-level analysis can also contribute to a more detailed understanding of skills. Such an analysis may, for example, point to partially isolated elements that are exempt from the synergy, resource competition between elements, newly added components that change the coordination pattern, components that get “switched on” to master particular contingencies, and components that need to be subtly active in the background for new synergies to emerge.
What Qualifies as a Synergy?
Not all interaction phenomena qualify as synergies. We need conceptual criteria to delineate synergy from non-synergy. A first point to stress here, however, is that synergy is not an all-or-nothing concept. Some aspects of a given interaction may be more synergistic than others. For example, Fusaroli et al. (2012) demonstrate that conversations include forms of verbal alignment that are independent of the discussion topic (i.e., the collective function in the foreground of the conversation) and do not enhance the degree of coordination, although verbal alignment is a linguistically relevant level of description.
Second, we might ask if externally coordinated events should be interpreted as synergies. Think of a park full of people running for cover under one tree as a thunderstorm sets in (Searle, 1990). On the face of it, this behavior lets us think of a synergy. Examples such as these, however, lack the required collective functionality and are not mutually coordinated, hence: non-synergy. Sports researchers, meanwhile, point to externally mediated forms of coordination that do have a joint functionality, for example, when two players of a soccer team do not coordinate directly with each other but respond to the actions of a particular opponent in coordinated ways (Bourbousson & Fortes-Bourbousson, 2016; Feigean et al., 2018). Cases such as these may legitimately be called a synergy.
Third, what does it mean to speak of collective dynamics and collective functionality? Importantly, we must not mistake the presence of a functionality for a joint purpose. When people collaborate, this functionality implies a joint goal. Yet, collective dynamics and collective functionality can also occur in non-collaborative contexts. In competitive sports or martial arts individuals demonstrably behave in closely coordinated ways. The difference is that one person tricks, lures, or coerces another person into a collective dynamic in ways that run counter to the interests of the losing party. People can even be engaged in synergistic behavior despite their own interests, as paradoxical interaction dynamics discussed in the next paragraph illustrate.
Fourth, deliberateness—qua conscious intentional attitude—is not required for interpersonal synergy. Unintentional synchrony is highly frequent (Davis, 2016). The concept of synergy should not be linked to any particular form of intentionality, but defined with respect to collective functions. For example, falling into lockstep is nothing people usually do on purpose. But it feels nice and has a subtle social purpose by smoothing interaction. One can potentially even treat as synergy unwanted entrainment or more complex violations of what people consciously “want.” Think of mirroring movements when two persons try to walk past each other in a narrow corridor (De Jaegher & Di Paolo, 2007), the inability to discontinue an unproductive conversation when people keep triggering each other (Granic et al., 2003), or intractable political conflicts (Vallacher et al., 2010). Also, human tendencies to synergize may be exploited for manipulative purposes in certain contexts. A synergistic coupling can notably arise as an unwanted byproduct of a basic entrainment tendency of humans. All these examples conform to the criterion of dimensional compression of behavior and qualify as synergies, despite being both dysfunctional and non-deliberate.
Lastly, should we reserve the term synergy for fully successful ensemble actions or fully completed tasks? Soccer forwards engaging in a double pass over several seconds, even if they fail to score in the end, clearly exhibit a strong synergy in the process—sports commentators would praise their admirable coordination. The criteria of collective low-dimensionality and interdependent adaptation apply here, despite the incomplete success. More generally, discussions of synergy should move the coordination process into focus, rather than a final achievement. It can reasonably be assumed that most “peak interactions” occur in a substrate of already coupled behaviors that may not be as conspicuous, but still qualify as synergy. A related question is how to best characterize synergies that are insufficiently coordinated right from the start. My tentative answer is that modes of partial or insufficiently constrained collective dynamics can be fruitfully discussed under a synergy heading, because their dynamic regularities are relevant topics of inquiry.
Which Mechanisms Mediate Synergies?
The most frequent mechanism that is held responsible for the creation of synergies is sensorimotor coupling in a situation of embodied co-presence. The claim that organisms dynamically couple with their environment is a key focus in ecological psychology (Marsh et al., 2006; Schmidt et al., 2011; Schmidt & Fitzpatrick, 2016) and ecological dynamics (Araújo & Davids, 2004, 2016; Araújo et al., 2006; Davids et al., 2014). This dynamic coupling not only concerns the external surroundings but also other people one interacts with. To participate in an interpersonal synergy, actors must couple their behavior by dint of some (usually, but not always bidirectional) connection through vision, touch, or other information. We need to think of such synergies as real-time embodied interactions that work precisely because individuals continuously adapt or modulate their own actions in relation to those of others. This does not exclude that some interpersonal synergies are simultaneously sensitive to prior influences, such as tactical instructions from a coach or known cultural rules.
Typically, it is an incipient interactive engagement itself that makes useful perceptual information available that mediates the synergizing process (Marsh et al., 2006; Marsh, Richardson, et al., 2009). To monitor the collective dynamics, actors attune their perception to relational aspects (cf. Araújo & Davids, 2004; Fajen et al., 2009; Passos et al., 2012; Silva et al., 2013) such as spatial geometries, angles, or distances, speed differentials to others, relative force vectors in martial arts, or harmonic lines when making music together. A well attuned perceptual apparatus may exploit informational variables that specify (mostly local) action rules for actors, which in turn support the synergistic behavior. If we situate the affordance notion coined by Gibson (1979) in an interaction setting, we can therefore interpret interpersonal synergies as being mediated by a mutual affordance responsiveness. Interaction-related affordances are gleaned from where others are and what they do in relation to oneself, information that is used to guide behavior. In addition, several experiments (reviewed in Marsh & Meagher, 2016) indicate that in order to exploit interaction-related affordances, individuals should also be attuned to relations others themselves perceive in a joint ecology, that is, affordances of others, as well as be attuned to the specific capabilities of others.
The discussed sensorimotor coupling mechanisms have the most evident capacity to ensure interdependent adaptation, without which there can be no synergy. However, other frameworks point to potential complementary mechanisms. Joint action scholars, for instance, appeal to representations of shared action goals, of action sequencing, or of the expected co-actions of others (Knoblich et al., 2011; Sebanz et al., 2006; Vesper et al., 2010; Wenke et al., 2011). Especially in so-called planned coordination (see section “The Synergy ‘Zoo'”), pre-process representations may in part explain the general kinds of actions that inform synergies, whereas in emergent forms of coordination they have very limited power because they impede flexible enough adaptation. One possible hybrid explanation is that of minimal joint intentions (Saint-Germier et al., 2021) that only define in advance the kind of activity one jointly engages in, but leave further details to sensorimotor coupling. Other potential cognitive mechanisms cited in the interaction literature may play a role but only in very limited ways. This notably includes kinesthetic empathy through neural mirroring (Gallese et al., 1996) and common coding between observed and executed action (Hommel et al., 2001; Prinz, 1997). These mechanisms can intrinsically apply only to symmetric behaviors, if they explain anything at all. Mechanisms of action prediction of others through anticipatory resonance (Aglioti et al., 2008) are of potentially broader causal relevance, but restricted to situations where the underlying skill set is shared and where anticipation as such is not harmful (as would be the case in some high-velocity action domains).
When searching for a sufficiently inclusive model of synergy-mediating mechanisms it is instructive to look to full-scale improvisation contexts, which best illustrate the breadth of skills and levels of organization to reckon with. When improvising together, interpersonal synergies are successfully coordinated in a field of myriad options (Kimmel, 2021; Kimmel & Hristova, 2021). Improvisers in dance, music, or theater coordinate their actions spontaneously, producing a flow of continuously changing synergistic states. As they explore new possibilities together, they may frequently also converge into novel types of synergy that have not been tried before yet. When improvising a dance together, for example, this process is mediated by action invitations and mutual enablement, but also through mutual constraints and challenges, as the dancers say “yes, and” or “no, but” to each other. All this requires an action system with a capacity for finely-adjusted interpersonal co-assembly. Improvisers must have the capacity to flexibly complement what the other person is concurrently doing without any reliance on fixed forms, but by mixing and matching action components as contextually required.
As champions of the ecological dynamics approach have streesed, such spontaneous co-assemblies happen within a set of multiple constraints (Araújo & Davids, 2004) that emerge from intentions, tasks, and ecological factors. A good illustration are rugby attackers who maintain a diamond-shaped formation and adapt their speeds to each other, but when they approach the other team’s defense they adopt evasive behaviors with slightly changed mean distances (Passos et al., 2011). Researchers therefore need to get a sense of different types of constraints that license collective synergy formation at multiple levels and timescales (Paxton et al., 2016). This “force-field” of convergent constraints includes the already discussed affordances that result from what others are doing. Such affordances may change rapidly but also may include informational invariants that attach to the broader situation. Second, task dynamic constraints (Kimmel, 2021) may temporarily narrow down the field of possible actions. A good example is a dance couple’s need to regain safe grounding after an acrobatic lift or jump before attempting a new action, or the need to build an attack as a team before shooting at the goal in soccer. This means that some interpersonal synergies are subject to short-lived path-dependencies, which momentarily rule out a wide range of otherwise afforded options until points of reconsolidation are reached. Third, intentions—qua internal system constraints—shape synergy formation, including long-term preferences that direct the collective dynamic toward particular “regions of interest.” What a group is interested in may be shaped by a shared practice history and trust (Golvet et al., 2024; Wilson & MacDonald, 2017). However, it can also be the interaction itself that gives rise to interests, themes, or local goals that temporarily emerge as shared intentions (Goupil et al., 2020, 2021). Similarly, shared concerns (e.g., Bourbousson et al., 2011) can make individuals orient toward a specific subset of behaviors in advance. Fourth and finally, general domain constraints delimit the interaction space more globally. This includes stylistic conventions (Sawyer, 1996), communication rules, roles (e.g., leading and following), and etiquette. This level of constraint implies that the range of admissible means for creating synergies is always to an extent part of shared orientations of practice, both with regard to avoiding injury or social impropriety, but also with regard to aesthetic or related group conventions.
Interim Conclusions
Taking the different mechanisms and constraints into consideration, we may conclude that synergistic behavior must always be negotiated in real time, but may be simultaneously constrained by shared basic orientations, prior expectations about the interaction “theme,” constraints of joint purpose, or the logic of specific tasks. Any “one-size-fits-all” view of the specific mechanism mix is unlikely to be found, a point that the section “The Synergy ‘Zoo'” will further reinforce.
In an interdisciplinary context, readers may also wonder what to make of the complexity-informed emphasis on the self-organizing and non-linear nature of collective behavior. Several cautionary points are in order: First, exploring self-organizing dynamics should not impede cross-talk with approaches such as joint action that use different terminology. Second, the rhetoric of self-organization needs to be handled circumspectly. It should avoid any implication that some numinous causal mechanism necessarily overrides what individuals want. The term self-organization is a simple shorthand for system regulation through local component interactions, hence the absence of central governance that “scripts” how individuals interact. As has been established, all interpersonal synergies must self-organize locally to be able to dynamically adapt, yet this may not always be the sole constraint or mechanism at play. Third, complexity-informed approaches frequently claim that mediating mechanisms operate through causalities that are “spread out” over a system encompassing the brain, the body, tools, and spaces. How (and just how much) this is true remains an empirical question for case-based demonstration. By a similar token, whether non-linear effects (e.g., self-stabilization, resistance to perturbation, sudden state shifts) or even interaction paradoxes occur in a self-organizing collective cannot just be assumed ex hypothesi.
The Synergy “Zoo”
Interpersonal synergies come in many shapes and forms, a fact which cautions against “one-size fits all” claims and against an over-reliance on a single type of research design. This section distinguishes multiple axes along which interpersonal synergies can be contrasted.
Dimensions and Deliberateness of Coordination
A first point to consider concerns what is coordinated to create a synergy. Some kinds of social synergies require matching the behavior, others display behavioral complementariness. In some contexts it best fits the phenomenon to focus on synchronization either of similar actions (such as kicking) or synchronization regardless of specific actions. Furthermore, some temporal regularities may only occur at a higher level of analysis.
Another difference to consider, as has been previously hinted at, is the role that conscious intention plays. While much synergistic behavior is deliberate, it can also be induced through sub-personal mechanisms such as motor entrainment or involuntary mimicry. Such synergies can emerge despite what people consciously intend, yet be very useful. Especially in-phase entrainment, but also behavioral matching (e.g., similar gestures), may act as low-level “smoothers” of collaborative interaction (Vesper et al., 2010). Some interactions may begin with low-level synergizing and gradually lead to more complex forms of synergy (Tollefsen et al., 2013). Low-level entrainment is not always a boon, though. In some dance contexts the natural seeming in-phase coordination that comes easiest needs to be “detuned” through training (Kimmel & Preuschl, 2015). Similarly, in some martial arts, opponents can exploit excessive entrainment to their advantage (Kimmel & Rogler, 2019; Kimmel & Schneider, 2026). The tension between “automatic” tendencies and advanced skills in an artistic context is well characterized by Miura et al. (2015, p.19) in the following way: motor learning of rhythmic artistic performance may be interpreted as a process of acquiring freedom from the intrinsic constraints that are associated with pre-existing, self-organizing tendencies
Emergent vs. Planned Interaction
Equally worth of consideration is how much synergies are cognitively specified prior to their creation. There is a spectrum ranging from planned synergies, often based on familiar interaction routines, to fully emergent improvisational synergies created on the spur of the moment. This roughly follows the contrast between emergent and planned coordination, as joint action theorists define it (Knoblich et al., 2011). 4
At one end of the spectrum sit genuinely emergent synergies. Both which specific kinds of actions are selected and how they are coordinated is decided spontaneously; the complementariness of actions is negotiated in real time through the interaction itself. In the extreme case, for example, in improvisational partner dances of in free jazz, there is no set task to begin with, but a wide space of possible options. This implies that both the task itself and the chosen means to complete it are negotiated in real time, as may be the roles (e.g., who currently leads and who follows).
Further along the spectrum, we find synergy building processes that are moderately emergent. These can be roughly pre-defined through a set of agreed task constraints or thematic preferences. This can prefigure sketchy ideas about “ballpark” types of action, which still allows for a sizable range of possibilities of real-time assembly. An example would be when improvisers decide to practice a circumscribed set of interaction techniques together (Kimmel et al., 2018). As joint action theorists stress, interaction can be constrained by a general understanding of the task and the role distribution, albeit without defining specific actions in advance (Vesper et al., 2010).
At the far end of the spectrum sit planned synergies, subject to actual pre-process coordination (Eccles & Tenenbaum, 2004). Stronger forms of pre-coordination move into focus explanations via shared action representations (Sebanz et al., 2006), which may variously specify goals, roles, or task characteristics such as the kinds of actions each person is in charge of, or the “who does what when” when co-assembling a causally complex action. Even in some improvisation contexts such “ready-mades” are in evidence (Sawyer, 1996, p. 199) that pre-specify many parameters of the interpersonal synergy. An example would be acrobats that my team has researched (unpublished), who practice different “tricks” with standard terms and a standard sequence of steps in their training sessions. Such coordination in advance is absolutely necessary in this high-risk context, as it allows the practitioners to concentrate on the precarious execution of each practiced “trick.” However, even here interaction details are subject to dynamic negotiation at lower levels, which determine the precise timing, action intensity, small corrections, and so on. More generally, it is very rare that interaction can be planned fully and still remain adaptive; plans are incapable of organizing behavioral control all the way down (cf. Suchman, 1986). This is why planned coordination, provided that it only concerns global interaction characteristics and is mutually well-adapted in the details through real-time mechanisms, can qualify as interpersonal synergy.
Degrees of Symmetry and Reciprocity
The next question concerns whether contributions to a synergy are similar or functionally complementary. It would seem as though the greater proportion of social synergies is of the complementary type. Examples include organized linguistic turn-taking in conversations (Fusaroli & Tylén, 2016), joint model car building (Wallot et al., 2016), or leader-follower dances where, for example, one person walks forward, the other backward (Kimmel, 2019). Such contexts invite a careful consideration of the mechanisms that license well-timed complementariness.
A related question arises as to reciprocity and power distribution in a synergy, the question “who influences whom how much.” Certain synergies are said to be isotropic in terms of displaying a bidirectional mutual influence between participants. For example, conversation partners may often mutually synchronize their rhythms. Yet, many other synergies are anisotropic (de Poel, 2016). The study by de Poel proposes to define this criterion from a coupled oscillator perspective: “Interaction between the components can be stronger in one direction than in the other, which implies an asymmetry in the strength of the coupling” (p. 2). Anisotropy is associated with leader-follower systems, where leaders influence the summary dynamic more, such as social dancing. 5 The extreme case would be unidirectional coupling, such as when persons in the last row of a lecture needs to coordinate their head movements with the persons in front (De Jaegher et al., 2016). Not infrequently, role-specific differences regarding access to information shape how non-isotropic coupling plays out, as in rowing crews where “the bow rower can see the movements of the stroke rower but not vice versa” (de Poel, 2016, p. 3).
As to the role distribution between leaders and followers, its likely raison d’etre is to ensure simpler governance of the interpersonal system, so as to curtail potentially disruptive non-linearities in the interaction. To take the case of tango dancing once again, technically demanding and rapidly improvised interaction to external music and on a crowded dance floor is only possible when a leader decides about the kinds of movements to be done and if the follower uses only an allotted space for their own creativity (Kimmel & van Alphen, 2022).
Collaborative vs. Adversarial
Another important comparative dimension concerns whether an interpersonal synergy is collaborative or competitive/antagonistic. Krabben et al. (2019) characterize the latter type in the following words: Two combatants in a fight self-organize into one interpersonal synergy, where the perceptions and actions of both athletes are coupled. To be successful in combat, performers need to manipulate and take advantage of the (in)stability of the system.
To formally define antagonism, de Poel (2016) suggests an oscillator model of repulsive and inhibitory coupling; collaboration, in contrast, requires persons to attract toward each other’s behavior. Kimmel and Rogler (2019) take a slightly different tack and define collaborative domains as featuring shared synergy aims, while characterizing competitive domains as being defined by opposing synergy aims. This means that the synergy of the winning party succeeds, while the loser’s desired synergy never materializes. For example, in grappling arts, wins are created despite the opponent’s attempt to resist or counteract them (Kimmel & Rogler, 2018, 2019; Krabben et al., 2019). Antagonism obviously includes the use of counterforce, but it may be less obvious how this involves a synergy. An argument can be made that, in soft martial arts, the interaction starts as “tuning in” to the opponent, and only then ends in a “hostile take-over” of the their physical degrees of freedom. Hence, a prior synergistic interaction is needed. The path toward this physical “take-over” initially involves subtle informational strategies of such as hidden preparations, feints, or lures. In domains with less or less constant physical contact such as field sports, competitive success is based on rapid reaction speed as well as feints.
In relation to the criterion of interdependent adaptation (see section “What Defines Synergies”), collaborative and antagonistic synergies differ sharply. In antagonistic contexts, errors of the opponent are encouraged and amplified, while reciprocal compensation is absent. In collaborative contexts, people often compensate for the errors of others. In light of this very fundamental difference, collaborative and antagonistic types may display different self-organizing dynamics (de Poel, 2016). It has been suggested that, when people vie with each other, the interaction can hang in the balance for longer, result in gridlock, or create paradoxes (Kimmel & Rogler, 2019).
Informational vs. Mechanical
The next distinction concerns synergies with and without direct contact. Different possible mechanisms used to negotiate the interaction are at issue here. Virtually all interpersonal synergies, that is, both those with and without direct contact, are informationally coordinated through visual or acoustic communication. In contexts where agents touch physically, synergies may additionally be mediated through mechanical coupling. To be clear, this means that physical forces passing between bodies contribute to coordinating the synergy. Even without such “push-pull” causalities, touch can serve as a subtle means of communication, as worksharing strategies based on haptic coordination illustrate (Reed et al., 2006; Sawers et al., 2017; van der Wel et al., 2011). For instance, Slomka and and colleagues (2015) conducted an expertiment with force plates in which connecting haptically through one arm was crucial.
One important type of mechanical effect is known as passive dynamics (Pfeifer & Bongard, 2007), that is, movement coordination that exploits the self-organizing mechanical dynamics of human bodies and their morphology. For example, when two persons jointly carry a rigid foam block the way that forces pass through their (indirectly connected) bodies synchronizes their gaits in a complex pattern; what emerges looks similar to a quadruped animal (Harrison & Richardson, 2009). This type of mechanism is somewhat analogous to clocks that come to be synchronized via the wall (Kugler & Turvey, 1987) or clever mechanical arrangements like Watt’s Governor for steam engines. Passive mechanical effects can stabilize joint carrying, as a simulation study by Lanini and colleagues demonstrates (2017). Being connected through a boat similarly facilitates joint rowing and stabilizes the coordination (Cuijpers et al., 2019). In domains where bodies interact at close quarters such as partner dance, individuals may literally extend their body structure or dynamics into another body (cf. Froese & Fuchs, 2012). A complex “collective physics” can result here. For example, acrobats create physical architectures in virtue of precise weight sharing, counterbalance, skeletal alignment, connective muscle arcs, levers, and stabilizations based on the interplay of rigid skeletal struts with the elasticity of muscles and fascia (Kimmel & Schneider, 2022). Similarly, in the earlier discussed soft martial arts context, experts establish a structural connection into the opponent’s body, which helps them to “hi-jack” their degrees of freedom. One important aspect of mechanical couplings and collective physical structures is that they can simplify tasks and reduce the cognitive processing load needed to regulate the interaction.
Further Distinctions
A scarcely noted distinction is that some synergies are ephemeral and “one-shot,” while others obey a more cumulative causal logic. In much social dance or musical co-improvisation, each moment of synergizing stands for itself, even if larger creative and thematic lines may be aimed at. Other synergies inherently build up toward an overall goal. For example, in a psychotherapy (Tschacher & Dauwalder, 2003) or bodywork treatment (Kimmel & Irran, 2021), the therapist aims at an overall beneficial effect on the client through interventions spanning 1 hour or even multiple sessions. Dozens of specific interventions may be combined in an effort to build toward this effect, first preparing the ground and then creating a sequence of interventions that meaningfully add up. Of course, there also exist a broad range of intermediate cases that are non-exactly “one-shot,” but display a moderate degree of cumulativity. In field sports, for example, each attack can be interpreted as a cumulative attempt at building synergy in the team, but after every attack (or every new kick-off the very latest) things are “set back to zero.”
Further possible distinctions between interpersonal synergies are thinkable. For example, creativity researchers may want to distinguish effect related synergies from motor synergies. Consider an orchestra who jointly generate a complex musical texture in virtue of their excellent lower-level motor coordination. The effects of their coordinated motor actions come together as sound effect in ways that make it intuitive to speak of a synergy. The focus of this synergy is a summary esthetic effect that the audience can perceive. Similarly, how a painter assembles brush strokes into a visual effect or how a cook mixes or processes ingredients for a new flavor may be interpreted as creating effect related synergies through appropriate motor actions. The future will tell if researchers find merit in such ideas.
Measuring Synergy
The most frequently discussed perspectives on interpersonal synergy draw on quantitative data and their mathematical formalizations. In this section, I will survey these approaches and explain how synergy relates to similar concepts from dynamical systems theory, a branch of complexity science.
Dynamic “Fingerprints” of Interaction
The most basic questions are why something qualifies as interpersonal synergy at all, and to what degree. Riley et al. (2011) propose that the criteria of dimensional compression and reciprocal compensation define synergies, as discussed in the section “What Defines Synergy?”.
Different formalizations have been proposed. Principal Component Analysis (Ramenzoni, 2008) tests whether only few dimensions can account for the variance in the data-set. The Uncontrolled Manifold (Black et al., 2007) determines the proportion of variance that is compensated for at the collective scale. 6 The notion of the Uncontrolled Manifold refers to the subspace within the set of possible actions where the shared goal remains unchanged. If variance is concentrated here this indicates that interaction participants are actively co-adapting to stabilize the required goal parameter, so for example when one person pushes slightly harder, the other pushes slightly less. The ratio between compensated and uncompensated variances can be used to characterize the presence or absence of (collaborative) synergies as well as their strength. Note that both discussed methods are linear. Variance is calculated along one particular axis, which has certain limits when analyzing complex real-world data. Other non-linear operationalizations of synergy have emerged, such as auto-encoder neural networks (De Feudis et al., 2021).
Let us illustrate how an Uncontrolled Manifold analysis proceeds: Riley and colleagues (2011) applied this method to a cooperative manual precision task, where it was expected that participants would compensate for deviations from the ideal dynamic when needed. The authors quantified to what degree variations in the position and velocity of each hand preserved the task’s goal value φ, the overall performance parameter, concluding that the ratio of compensated to uncompensated variance was >1, indicating the presence of synergies for intrapersonal and interpersonal coordination. Dimensional compression occurred because a lower dimensional variable, φ, was selectively stabilized via reciprocal compensation among the components. (p. 5)
Thus, the presence of interdependent adaptations pointed to a synergy here.
To provide another example, Silva et al. (2016) attempted to capture synergy strength in soccer teams by focusing on the speed of mutual adaptation when a player adjusted to relevant teammates. The authors used co-positioning in pairs to calculate an index of synergistic coupling strength between the players and found that the latter improved through learning over a few weeks. A study of the ball carrier in rugby and the configuration with a support player in two-against-one situations (Passos et al., 2018) found that synergy strength fluctuates over time with respect to different attack phases.
In a study of conversation, Fusaroli and Tylén (2016) explicitly compared the predictive power of synergy metrics to the interactive alignment approach. The latter highlights imitative reuse of each other’s words, speech/pause rhythms, and prosodic patterns between interlocutors, whereas synergy uses jointly constituted patters such as questions and answers to measure the strength of coordination. Thus, the synergy approach “points to structural organization at the level of the interaction—such as complementary patterns straddling speech turns and interlocutors” (p. 145). The authors argue that synergy provides the better statistical predictor of how well the dialog works, although they also acknowledge that “synergies and alignment capture different aspects of the dynamics of dialog” without any “single mechanism that makes conversations easy and effective” (p. 166). This means that, to understand conversations, emergent dynamics operating at the level of the interaction itself need to be taken into account. Reinforcing a similar point, researchers have explored more abstract statistical forms of behavioral alignment known as complexity matching that span multiple timescales of conversation and go far beyond simple mimicking or synchrony (Abney et al., 2014).
The Coupled Oscillator Perspective
A frequently researched topic concerns temporal properties of interpersonal synergy. A trailblazing study by Newtson et al. (1987) identifies periodicities in four interaction tasks of several minutes: unloading a truck, setting up a tent, explaining how to change a tire, and playing one-on-one basketball. The four tasks were manually coded in detail, allowing the authors to model the time-series in terms of excitatory and inhibitory and processes between the two persons. Although the authors examined what is at face value a complex and non-rhythmic interaction form, they argue that thinking of this phenomenon in terms of coupled oscillators is productive due to the repetitive periodicities (e.g., spike structures) that emerged in the coordination, which are similar to oscillator waveforms.
A wide range of lab experiments have investigated interpersonal limb synchronizations, for example, when swinging pendulums or legs together (Schmidt et al., 1990, 1998). Spontaneous phase transitions have been repeatedly documented. As movement frequency increases, the system shifts from an unstable anti-phase (i.e., alternate) to a more stable in-phase (i.e., simultaneous) movement patterns. Similar studies address rhythmic activities such as joint walking, rocking chairs together, finger tapping, carrying objects together, or joint rope turning (Fine & Amazeen, 2014; Harrison & Richardson, 2009; Huys et al., 2018; Schmidt et al., 2011; Sylos-Labini et al., 2018; van der Wel et al., 2011), or investigate coordinated force tasks (Slomka et al., 2015; Solnik et al., 2015). Other research investigates tasks such as hand-clapping games, joke telling, or aikido exercises (Black et al., 2007; Richardson et al., 2007; Schmidt et al., 1990, 2011). More complex rhythmic patterns are in evidence as well, such the earlier discussed spontaneous gait patterns that resemble those of quadrupeds (Harrison & Richardson, 2009). When people entrain their movements to auditory beats, we see the same intrinsic preferences for in-phase and anti-phase periodicities, although it is possible to produce others with sufficient practice (Miura et al., 2015).
Interestingly, the documented phase transitions are comparable to intra-personal limb coordination. Apparently, similar principles of self-organization govern intra- and interpersonal behavior (cf. Schmidt et al., 1990), even if interpersonal systems involve no mechanical or neural couplings between limbs. This notwithstanding, interpersonal and intra-personal synergies do not behave identically. The coupling strength between limbs of a single person is stronger than interpersonally (Schmidt et al., 1998, Richardson et al., 2008) and when each person needs to coordinate multiple limbs simple patterns may get destabilized (Kodama et al., 2015).
An influential mathematical formalization describing how changes in movement tempo produce specific interpersonal periodicities is known as the Haken-Kelso-Bunz model (Haken et al., 1985). The model couples the equations of two individual oscillators and describes how sudden transitions in periodicity are brought about by the logic of their entrainment, in response both to their intrinsic rhythmic tendencies (which can be similar or different) and to external constraints. The model has grown more inclusive since its inception to increase its applicability to real-world complexities, including multi-agent scenarios (Tognoli et al., 2020). Here, influences from other persons figure as one variable among several that determine a person’s rhythmic preference.
Synchrony, however, is not always an ideal index of optimal interpersonal synergies. Even though synchrony is generally taken as sign of rapport and its disruption can be indicative of pathology (e.g., Fitzpatrick et al., 2013) or conflict (e.g., Paxton & Dale, 2013), a study by Wallot et al. (2016) indicates that synchrony is not necessarily predictive of collaborative success in complex tasks. The authors conclude that “optimally coordinated joint action might actually lie on a continuum between synchrony and action diversification depending on specific constraints” (p 21). Oscillators that become more “detuned” with respect to one another can make complex behaviors possible in the first place (Richardson et al., 2016). Other findings suggest that de-synchronization can be integral to collaborative tasks such as creativity (Laroche et al., 2024).
In parallel, it bears mention that it is often practical to combine the coupled oscillator perspective with formalizations of non-rhythmic properties. Nalepka et al. (2017), for example, claim that coupled oscillator modeling can incorporate different behavioral modes and interaction strategies when doing a simulation of herding a flock of sheep, but in fact add other descriptors to create a well-rounded analysis. Yet another suggestion is that low-level dispositions toward rhythmic in-phase entrainment often simply trigger higher-level mechanism that may require different forms of analysis (Tollefsen et al., 2013).
Applying Dynamic Systems Theory to Complex Forms of Social Interaction
Numerous applications of dynamic systems theory to social domains share conceptual roots with synergy theory. They specifically share in common that interactions are interpreted as complex dynamic systems that (a) converge on particular patterns in a state space and (b) do so in lawful ways and sensitive to external constraints.
Of special note is the continued importance of Hermann Haken’s synergetics framework (1977), which originates in theories of far-from-equilibrium complex systems. Haken’s analytic strategy proved to be very influential for synergy theory: He proposes to define an order parameter that describes the system’s collective state and to identify an external control parameter that, when it changes, causes the order parameter to transition to a new state in a self-organized fashion. Movement researchers were the first to apply this to human behavior (see Haken-Kelso-Bunz model).
Synergetics concepts have since been variously applied in psychology and psychotherapy research (Tschacher & Dauwalder, 2003; Tschacher et al., 1992). In a psychotherapy, a powerful control parameter for success is the rapport between therapist and client. Hence, a good therapist can harness the interaction quality to the benefit of the “target system,” the client’s well-being. The task of the therapist is to help a client to move away from “a dysfunctional or psychopathological region of the client’s state space [...] establishing a novel attractor in a different, healthier region of state space” (Tschacher & Haken, 2020, p. 1076). The therapist’s job therefore is to destabilize undesirable dynamics in a client and facilitate transitions toward new and healthier system states or “re-awaken” hidden system states so the client can access them again.
The synergetics model has been empirically applied to psychotherapy sessions by drawing on non-verbal interaction data or measurements such as heart-rate and breathing. Other work uses subjective process ratings by clients and treats therapists implicitly (Schiepek et al., 2014) to study specific hypotheses such as the occurrence of sudden pattern transitions, the occurrence of critical instabilities before transitions, or the need for stable boundary conditions as prerequisites for change to occur. Haken and Schiepek’s (2010) framework proposes a set of complexity-based principles for psychotherapeutic work. It is rooted in a systemic way of thinking about how therapeutic changes happens and defines the therapist as someone who accompanies or enables self-organized and sustainable system reorganization. This is, for example, facilitated by exploiting the intrinsic dynamics of the client’s system, by creating constraints conducive to change, as well as by recognizing key tipping points in the dynamics and utilizing these at the right moment. Quite possibly, many of these rules, although they remain quite generally defined, are helpful precepts for skillful interpersonal synergizing in other professional domains as well.
Similar work has been done in the field of dynamic social psychology (Vallacher et al., 2002; Vallacher & Nowak, 2007). There it has been argued that “complex interpersonal phenomena can be understood in terms of simple models involving principles and mechanisms common to a wide variety of dynamic systems” (Vallacher & Nowak, 1997, p. 73). Social interaction is modeled as the coupling of two non-linear dynamic systems, albeit in a different form than coupled oscillators. For example, Vallacher et al. (2005) mathematically simulated the emergence, maintenance, and disruption of coordination in close personal relationships, using the degree of motor synchrony as indicator. It was found that how much dyads synchronize depends both on their intrinsic internal state and on the intensity of mutual influence. High behavioral synchronization only occurs when internal states of the persons are inherently similar or when they exert a strong mutual influence. Weakly similar internal states may risk a coordination breakdown; and under weaker coupling conditions more diverse interaction patterns arise, including complex rhythms as well as independent behavior. At the same time, the authors found indications that even partners that are only weakly coupled can mutually stabilize their respective dynamics.
Attractor Dynamics
Applications of dynamic systems theory frequently use mathematical tools that characterize interpersonal behaviors with respect to their state space. The interpersonal system is thought to be structured by attractors carving said state space into predictable regions of behavior. The central idea is that social processes from a wide range of starting points may converge on, or repeatedly transition between, a small set of dynamic attractors (cf. Richardson et al., 2014). While an interaction unfolds the transitions that occur between attractors describe the evolution of the collective behavior. This can signify changes between types of synergistic coordination, between modes of stronger or weaker coupling, or moving from non-synergy to synergy and vice-versa. More generally, which attractors are found in an interaction system describes how it is configured as a whole.
A typical phenomenon observed in such dynamics is symmetry breaking (Richardson et al., 2016). It means that the interaction converges on one attractor after “hanging in the balance” between several possible ones. Symmetry breaking is, for example, documented for competitive behaviors. This happens as the interaction transitions from a stalemate to a win the moment one opponent prevails. For example, in kendo fencing opponents move back and forth with equidistance for some time, a symmetry that is broken when one opponent gets close enough to strike (Yamamoto et al., 2016). Similar, in basketball, symmetry holds as long as a defender mirrors the attacker’s right and left movements to block his or her passage; symmetry breaking here stands for getting past (Bourbousson et al., 2010a). Symmetry breaking can thus be applied in all cases when a dynamic equilibrium between two attractors suddenly transitions to one of them.
Attractor dynamics can also contribute to the analysis of how undesirable synergies stabilize through long-term histories of interaction, as is the case in intractable social or political conflicts (Coleman et al., 2007; Vallacher et al., 2010) and problematic family dynamics (Granic et al., 2003). Some readers may be surprised that dysfunctional dynamics can be interpreted as self-organizing synergies. However, it follows from the notion of self-organization that system dynamics can take on a harmful “life of their own,” from self-reinforcing psychological and health problems, via uncomfortable social roles, to political conflicts. These paradoxical system dynamics can be elucidated in terms of Haken’s (1977) notion of circular causality, where downward causation from the collective dynamic constrains components despite their intrinsic tendency. The collective dynamic, once in place, can “enslave” components and “bring them into line,” for good or ill. To help a system that has become stuck in a dysfunctional dynamic, the detailed analysis of its self-reinforcing loops can provide a way out (e.g., Schiepek, 1986).
Synergies in Team Sports
In complex naturalistic settings such as team sports, it is not always easy to cut through the many different realizations of synergistic behavior. However, research on ecological dynamics in team sports has been on the forefront of developing sophisticated techniques that elucidate how strong, of what kind, and responsive to which context factors synergies are, questions that all have relevant implications for training.
Studies of sports such as soccer, futsal, or rugby have examined the nature of adaptive collective behaviors, for example, by asking which distances between teammates are most adaptive. Functional intra-team coordination has been found to follow spatial formation rules, which are adapted to the game situation, for example, a study of rugby found that “players moved closer together to face the first defensive line and diverged to face the second defensive line” (Passos et al., 2011, p. 163). Studies have also shed light on synergistic coupling patterns with opponent players. For example, in 1-vs-1 contests in many team sports (including basketball, rugby union, and association football), the relative velocities of a ball carrier and the closest intercepting defender and the distance between them have been extensively studied. Defenders typically synchronize their speed with attackers, with breakdowns in this coordination leading to successful breakthroughs, a pattern that is however sensitive to contextual factors such as proximity to the goal or location of the ball (Headrick et al., 2012; Vilar et al., 2012, 2014). How a player couples with an opponent depends on factors such as interpersonal distance and speed difference (Passos et al., 2008). It also depends on the opponent’s specific action. Defenders in futsal may be attracted to an in-phase mode of coordination in general but use anti-phase coupling when the attacker receives the ball (Vilar et al., 2012). Similar measures that predict synergy formation in a dyads have been applied in many studies, including kendo, boxing, and duels between two basketball players (Bourbousson et al., 2010b; Correia et al., 2014; Esteves et al., 2011; Hristovski et al., 2006; Travassos et al., 2012; Yamamoto et al., 2016).
Furthermore, research on ecological dynamics has proposed a number of aggregate measures to understand synergistic patterns of a whole team (which can be averaged and used to compare different scenarios of a match). These include metrics for team center, team synchrony, and team dispersion, as well as team communication channels (Araújo & Davids, 2016). For example, Duarte and colleagues (2013) use a technique known as Cluster Phase Analysis to assess the overall synchrony of a soccer team as well as synchrony with the opposing team. Their study found that, on a soccer field, interpersonal coordination in the longitudinal dimension tends to be stronger than laterally, that local variability of individuals may stabilize the macro-scopic order of a team, and that “a team performing in a synergistic manner may attract the other team to behave in a synchronized way too” (p. 563).
Other studies describe types of team behaviors (Araújo & Davids, 2015), coupling modes (Bourbousson et al., 2010a), or degrees of behavioral fusion (Araújo & Davids, 2016), where some moments may display somewhat looser assemblages than others. Duarte et al. (2012) discuss this as “super-organismic” properties of a team. Furthermore, sports researchers have tried to distinguish forms of division of labor in a team. This notably includes preferentially occupied spaces, area and range covered by a player on a field, or the spatial variability of a player (Araújo et al., 2015; Fonseca et al., 2012). Heat map visualizations have additionally been employed to identify spots where a player lingers most throughout a match, providing clues to the type of synergistic engagement likely to take place there.
Group Synergies Viewed as Networks
Yet another methodological avenue for studying synergies in large groups are (small worlds) network models (Grund, 2012; Passos et al., 2011), which identify statistical regularities of pattern formation dynamics. Applied to sports teams, network models capture the coordination among players in terms of how they build temporary interactive links during the game. The network represents possible communication channels among players; nodes represent the players themselves; and lines between them are weighted according to the number of passes or positional changes completed between them.
By ascertaining the preferred interaction channels, researchers can determine which players enter into closer local synergies or are most involved in an attack and related tactical behaviors. For example, Duarte and colleagues (2012) compared link strengths on a pair-wise basis in water polo players to investigate dominant interaction probabilities, passing accuracy between players, as well as position switching. As Araùjo and colleagues (2015) propose, the key contributors to a team’s performance can be identified through their stronger and more numerous network connections. In addition, it is possible to compare different match networks to describe a team’s tactical and strategic character based on the preferential attachments between players.
Furthermore, network analysis can identify which role-specific actions positively contribute to the formation of subsequent team synergies, say which technique a goalkeeper uses. Network characteristics can also be used to predict team success or different game styles (Pina et al., 2017). Metrics that have been proposed include clustering, centralization, and interaction density. Centralization here refers to the degree of task distribution in a team and interaction density between player dyads is indexed, for example, by passing rates. To provide an example, in a study of 23 English premier league teams Grund (2012) found that high interaction density and low centralization are predictors of good team performance. 7
Finally, team synergies can be explored through the use of multi-level formalizations such as hyper-networks that represent “micro (interactions between players), meso (dynamics of a given critical event, e.g., an attack interaction), and macro (interactions between sets of players) levels” (Ramos et al., 2017, p. 1). This multi-level nesting of synergies is an important theoretical aspect I shall return to shortly.
Micro-Practices and Meanings of Synergy Formation
The research discussed so far emphasizes collective-level quantitative modeling. In this section, I will survey complementary literature which investigates specifics of synergy formation from a qualitative or mixed perspective.
Unique Synergies and Their Meaning
An increasing number of authors argue that the synergy concept can be implemented through 1st person (i.e., phenomenological) descriptions. The sports scientists Seifert and colleagues (2016, p. 109) defend this possibility as follows: although constraint manipulation can be well controlled in lab contexts, it can be more challenging to determine how a set of constraints [...] interact in the ecological context of performance [...] when a phenomenon is fundamentally embedded in a complex interacting set of multiple constraints [...] phenomenological data can provide fruitful information about how participants experienced these constraints as a whole, which is meaningful to their own point of view.
The informative nature of phenomenological data can be nicely exemplified by situations in which two rowers notice a small synchronization gap between them, but typically do so with reference to different perceptual cues accessible from their respective positions. The phenomenological data also helped Seifert and colleagues interpret their quantitative data on variability, pattern switching, and local destabilizations, which could be made sense of from the 1st person viewpoint.
A number of studies apply micro-phenomenological methods, which employ in-depth interviews about specific moments of synergizing. Two noteworthy 1st person studies on soccer teams investigated the dynamics of the lived experience of the game (Feigean et al., 2018; Gesbert et al., 2017). Post-game interviews were employed, supported by video-stimulated recall, which focused on ball carriers, player positions, salient errors, and so on. Both studies reconstruct which meaningful units of interaction participants differentiate and what their experiential characteristics are. On this basis, Gesbert et al. (2017) describe team coordination over time “as the succession of collective regulation modes” (p. 10), which changes their characteristics when the micro-context changes, such as when ball possession is regained. The interviews also probed into the informational resources used by players to coordinate with their teammates and to contribute to the team. This could either be based on local, global, or mixed regulation modes. For example, local information involves nearby team members or some distant 1-vs-1 interplay between a teammate and an opponent, whereas global information might concern larger team patterns such as a line of midfielders. The authors discuss how frequent each collective regulation mode is in four different game scenarios “(a) reorganization in play formation, (b) adaptation to actions of putting pressure on the ball carrier, (c) availability to get the ball out of the recovery zone, and (d) shoot for the goal” (ibid, p. 7). The study by Feigean et al. (2018) is similar and focuses in somewhat greater detail on the content of informational resources that support the player’s adjustments in different game scenarios. The study distinguishes different ways of perceptual focalizing on, respectively, a single player, a global spatio-temporal array, a player in relation to such an array, the area of current play, surrounding behavior captured by a comprehensive awareness. Additionally, anticipations based on previously built observations are described.
Other micro-phenomenological studies conducted by my own team employ a combination of video-stimulated interviewing, thinking aloud, and informal experimentation in a workshop setting. Among these, two studies of contact improvisation (Kimmel, 2021; Kimmel & Hristova, 2021) trace how dance partners negotiate complex synergies under conditions of direct physical contact. The studies follow the dancers’ interaction as it gives rise to continuously shifting weight sharing configurations, at a timescale of a few seconds. Both studies highlight how ongoing affordance responsiveness between the participants mediates their interaction. The self-organizing interaction dynamics itself brings about new, often unexpected affordances for collective action. A similar micro-scale approach has been applied to the martial arts aikido (Kimmel & Rogler, 2018, 2019) and the taichi practice of “push-hands” (Kimmel & Schneider, 2026). Again, these studies confirm the self-organizing causality of interaction but this time in synergies that are adversarial and that require luring the opponent into a, for them, detrimental collective state through clever combinations of force and subterfuge. In addition, the study by Kimmel and Rogler (2019) explores which synergy creating strategies are preferred by aikido experts depending on the initial relative timing and spatial geometry as the opponents approach each other.
Overall, micro-phenomenological methods allow exploring the details of selected situations and of the synergy building practices employed in them. They allow us to take a closer look at the micro-scale “give-and-take” between partners, in other words, the reciprocal causality that brings an interpersonal synergy to life in its unique context. This “inside” perspective also renders accessible the skills and strategies employed to create technically sophisticated synergies. I will refer back to some of the studies below, discussing further details of these aspects.
Micro-phenomenological studies are nicely complemented by micro-genetic studies done from a 3rd person perspective. These are based on the labor-intensive manual coding of observational data, but without conducting interviews. Torrents and colleagues (2010) annotated contact improvisers in a fine-grained manner and then subjected the annotation to statistical measures. They distinguish several dozens of basic synergy components across different dance techniques and contrast their relative frequencies; another study contrasts different instructional constraints (Torrents Martín et al., 2015).
Connecting Experiential with Objective Synergy Properties
As the abovementioned study of rowers by Seifert et al. (2016) demonstrates, 1st and 3rd person methods may display complementary strengths. A similar argument for complementariness emerges in a study by Kimmel and Preuschl (2015), who studied tango argentino. It was found that kinematic measures best revealed the details of relative timing and micro-variations between dance styles that were otherwise difficult to verbalize. Meanwhile, interviewing the dancers was more helpful for learning about technical background details as well as the relevant perceptual cues that allowed the partners to coordinate.
Other studies addressing improvisational interaction employ mixed method designs, which provide clues about counterparts between subjective meanings and objective metrics. Himberg et al. (2018) conducted a study with groups of dancers to explore if their experience of “togetherness” has specific kinematic counterparts, such as the smoothness of global group coordination. Somewhat similarly, Noy et al. (2015) combine ratings with kinematic measures and heart-rate measures that index the experience of “togetherness.”
Studies have also investigated how individual intentions affect objective measures of group synergizing. In a study of musical improvisation, Goupil et al. (2020) had improvisers annotate whether they wanted to continue or change the trend produced by the larger group and analyzed how this related to metrics for group alignment. Similarly, Golvet et al. (2024) had improvisers rate their convergent, divergent, or loosened synergy intentions (playing music “with”/“against”/“without”). Goupil et al. (2021) linked the similarity of what music improvisers wanted, their shared intentions (or not), to objective coordination measures and checked if the quality of joint improvisation as rated by external listeners confirmed this. In my estimation, methods of this type are vital for progress in synergy research. They connect overt behavior with unobservable cognitive factors such as intentions and help evaluate what the synergy concept can achieve in relation to psychological constructs such as “shared intentions.”
Understanding Components and Their Relations
Despite the ostensible relevance to both skill researchers and coaches, there is a scarcity of micro-level investigations of sophisticated interaction techniques, which synergistically integrate dozens of components across the skin boundary. First inroads come from three micro-phenomenological studies, two of them previously mentioned. Kimmel (2021) takes stock of action components that contact improvisation dancers use to create technically complex synergies. For example, a multi-stage acrobatic lift of several seconds is analyzed in terms of actions contributed by each dancer. The study traces how intra-body actions, such as increased core tension or skeletal alignment, feed into the interpersonal physics that aligns skeletal structures and creates a fulcrum for the lift between the two bodies. The individual actions remain perfectly complementary based on the dancers’ knowledge of interaction principles, although the situation evolves spontaneously. Similarly, the taichi study by Kimmel and Schneider (2026) explores the technical means that experts combine for highly effective pushing of an opponent without much active force. The combinations of action parameters employed can be trained through basic practice routines, but remain highly responsive to factors such as relative timing and opponent countermeasures. In fully improvisational practice contexts, parametric combinations can additionally generate different hybrid techniques and complex sequential combinations of technical elements. How technical elements are combined into “architecturally” complex interpersonal synergies in partner acrobatics is discussed by a similar study (Kimmel & Schneider, 2022).
Furthermore, all three studies indicate that the analysis of component interactions would be incomplete without considering how domain-specific habits contribute core elements to situated synergies. Good basic postural alignment, muscle pre-activations, and other readiness patterns furnish “kernels” that contribute to a wide range of possible interpersonal synergies, ensuring that activations do not need to be built from scratch.
More generally, certain component actions may be so tightly coupled with others that they form “coalitions” (Turvey et al., 1978), while other components remain weakly coupled. A good example is the elbows and chest center of two tango dancers that (almost) always move together to create a stable frame, while the legs act more independently (Kimmel & Preuschl, 2015). Furthermore, different body parts can display qualitatively different coupling patterns, for example, in joint tango walking both torsos are tightly coupled in virtue of an in-phase synchronization, but the legwork is slightly out of phase as one person walks forward and the other backward. Also, partly autonomous behaviors may be partitioned off of from the interpersonal synergy. The supported legs and torso of a dancer need to be closely coordinated with the partner’s, but the unsupported leg has a “creative life of its own” (Kimmel & van Alphen, 2022). Persons may synergize selectively, such as when acrobats, at the beginning stages of a handstand on another person, synchronize their torso with the supporting partner receiving part of their weight, but dissociate their legs from this synergy, as these are still responsible for stability on the ground, as unpublished data from my team suggests.
One level up from components, the concept of sharing patterns allows us to specify the general relationship of components of a synergy. Sharing patterns can be used to contrast types of synergy with respect to the global “division of labor” between individuals. This can unfold as parallel vs. alternating actions (Bell & Kozlowski, 2002), sequential vs. parallel initiation, symmetric vs. complementary movements, or mere element aggregation vs. deep fusion (Araújo & Davids, 2016).
We may also see different kinds of co-dependency relations between the components of a synergy. Individuals may facilitate each other, and strengthen the coordination, but there can also be scenarios of necessary compromises: A rather typical case is the need to manage the relative competition between aims, for example, if a boxer wants to swing at the opponent with maximum strength but can only achieve this by sacrificing his balance. However, besides necessary compromises, we may also see clever all-in-one solutions known as “simplexity” (Berthoz, 2012), such as when good inner alignment in social dances simultaneously supports individual balance, motility, and being precisely felt by the partner (Kimmel, 2016).
Level Interdependencies
Understanding the precise causal links between individual and interpersonal synergies offers a rich topic for more detailed inquiry. We may typically expect that lower-level patterns change interdependently with the interaction level (Paxton et al., 2016). This suggests a nested view in which lower-level coordinative structures are embedded in higher-level ones (cf. Tollefsen & Dale, 2012). Such a view can be best expressed in terms of circular causality. Richardson et al. (2014, p. 256) describe this well: elements or agents at the lower level of the system modulate the macroscopic order of the higher level and at the same time are structured by the macroscopic order of the system.
These hierarchical interdependencies prompt asking what trade-offs between different levels of regulation may be found. When exploring this micro-macro relationship, it can pay off to study how collective states and individual synergies feeding into them co-evolve. A previously discussed micro-phenomenological study (Kimmel & Schneider, 2026) sheds light on how bodily self-cultivation specifically contributes to winning a taichi bout. Taichi experts do this by perfecting their inner alignment, relaxation, specific forms of using the muscles, and movement habits. This creates a configuration that allows the opponent’s incoming force to be channeled back into a rebound or related collective effects of a passive dynamic sort (see section “The Synergy ‘Zoo'”).
Similar concerns can, of course, also be addressed quantitatively, for example to understand which level is causally stronger in the circularly causal relationship between levels. Montull and colleagues (2021) studied slackliners who balanced while holding hands. They found that interpersonal synergies had a dominant role and that individual synergies (in relation to their balancing task) were nested within this co-adaptive whole. Yet another question of interest concerns whether the general nature of synergizing across nested levels is similar or different. For example, Bourbousson et al. (2010a; 2010b) argue that in basketball similar interaction principles govern dyads and the higher level of teams considered as wholes.
In team sports, a characteristic and fascinating complexity is that multiple nested synergies between different players are happening at the same time. Multiple small-group or dyadic synergies add up to the team’s overall performance. In such contexts, interpersonal couplings at different levels may intersect, for example, a player can contribute simultaneously to a dyadic and to a group synergy, but may also compete or require prioritizing. Santos and Passos (2021, p. 3) hypothesize that “interpersonal synergies involving more players result in the weakening of lower level synergies. For instance, a synergy formed between two center-backs ‘disappears’ due to a new intra-sectorial synergy, formed by the whole defensive set.” Who a player synergizes with can quickly change. For instance, in one moment a defender needs to stabilize the interpersonal distance to a teammate and then switches to a more inclusive group behavior.
The Micro-Evolution of Synergy
Turning our gaze to how modes of coordination evolve over time, further topics for inquiry meet the eye. How interacting persons move in and out of closely coupled coordination is one such topic. In jazz ensembles, there is a waxing and waning of the tightness of coupling between relaxed “almost solo” moments and integrated group playing (Berliner, 1994). It has also been shown that free improvisers “play against” their partners at times, but also to “play without” them, seeking creative autonomy for moments (Golvet et al., 2024). Collective improvisers move dynamically through a huge space of co-functional patterns. Some synergies are convergent and support others; many other synergies are divergent and challenge others, while trying to maintain the overall coordination. There may be strategic reasons for tightening and loosening synergies. For example, Goupil et al. (2020, p. 12) suggest that in joint music improvisation tightening occurs “at strategic points during the performance, precisely when the group is on the verge of imploding under the weight of its internal contradictions.”
As a synergistic coupling evolves, many small-scale adaptations to perturbations or contingencies occur. This can happen, as Riley and colleagues (2011) predict, through task-preserving behavior that saves a synergy that appears to be slipping. In many improvisational tasks, however, actions that serendipitously co-opt chance deviations or errors may be preferred by experts. The perturbation may be picked up on or even amplified. An example is a tango leader who perceives a slightly exaggerated weight shift in the follower and, instead of enforcing his original intention, serendipitously locks onto to this to switch from a linear into a circular dance technique in the fraction of a second (Kimmel, 2019).
Larger-scale changes in interpersonal synergy are frequent as well. Some of these include abrupt phase transitions, such as switching from defense to offense tasks in field sports. Other changes are continuous. In free improvisation contexts such as dance or jazz, synergies may gradually develop or “morph” to maintain continuity (Kimmel et al., 2018). Such continuous transitions have been termed synergy flows (Kimmel, 2021), characterized by a fluid evolution of the component mix of the synergy. Often continuity within change is seen here, as some parameters shift more rapidly while others ensure sufficient continuity. For example, in contact improvisation dancing variables, such as the number of limbs that touch the ground (e.g., a quadruped stance) or whether the torso itself is on the ground, evolves more slowly than nested but more rapidly evolving small-scale actions (Torrents et al., 2016).
Of course, the flow of synergies is often constrained by a wider task logic, as the cumulative synergies discussed in section “The Synergy 'Zoo'” show. Our own unpublished research on partner acrobatics indicates that complex lifts move through a complex sequence until one person can, for instance, stand on the others head with a single hand for several seconds. Such sequences include highly dynamic phases, but are interspersed by points of greater stability where practitioners can correct or pause more easily, before complexifying the synergy further. To take another frequent observation we made in the context of soft martial arts, we see here that the opponent’s situation is gradually deteriorated by synergizing through subtle informational mechanisms, such as by creating a lure, before adding decisive mechanical manipulations.
I would like to close by pointing out how research on improvised dance, music, or theater can benefit from the synergy framework. Synergizing behavior can propel interaction toward creativity, provided that it includes well-constrained exploration and responds to serendipity. This would explain how the dynamics of an encounter acquire creative power, a causal mechanism Sawyer (2003, p. 2) refers to as “collaborative emergence.” Even the physical interplay of individual actions that combine non-linearly (Kimmel & Hristova, 2021) can instigate various forms of self-organizing dynamics (Torrents et al., 2016) and move the interaction toward surprising new attractors (Miura et al., 2015). The creative potential inherent in collaborative emergence may be understood as a general feature of complex multi-level coupling behavior (Walton et al., 2018). From a more qualitative perspective, it can also be seen as resulting from specific creativity skills whereby improvisers court emergence, for example, by deliberately gravitating toward states of indeterminacy (Kimmel et al., 2018), states in which small actions result in large qualitative changes (Hristovski et al., 2011; Orth et al., 2017).
Conclusion
The interpersonal synergy framework offers an original perspective on the principles of collective action dynamics that transcend skull and skin. This provides an antidote to methodological individualism in other branches of interaction or social cognition research. The framework comes with a set of guiding notions, such as performance parameters, sharing patterns, dynamically assembled component arrays, and interdependent adaptation.
At its origin, the framework is linked to methods capturing “dynamic fingerprints” of interaction, a tradition that has spawned innovative techniques for quantifying collective dynamics. This may hold promise for developing intelligent systems for sports performance analysis, psychotherapy research, or embodied conversation markers, to name but a few applications. In addition, a new wave of qualitatively oriented studies has embraced a more “bottom-up” perspective, throwing into relief communication, perceptual resources, skills, intentions, and strategies. The advent of mixed methods studies which associate behavioral with experiential characteristics seems equally auspicious for moving out of the narrow confines of a purely behavioral research paradigm. How synergy analysis relates to subjective meanings, skills, and strategies awaits much further analysis.
Undisputedly, interpersonal synergy is widespread in the human lifeworld; yet only future research will tell what portion of interaction behaviors “in the wild” are indeed synergistic and if so, by which standards. It also remains to be seen how broadly concepts originating in movement research can be readily applied to other kinds of interaction. For example, what could it mean to talk about degrees of freedoms in a collective “thought-space” being collectively organized to create cognitive synergies such as academic co-authorship?
A cautious assessment of the research field would suggest the following four tentative points: First, the synergy construct clearly overlaps with others that think about interaction systemically (e.g., distributed cognition) and may be implicit in many studies that do not actually speak of synergy. Second, formalizing interaction through a “mechanistic” approach remains a project of uncertain scope, and one with possible methodological limits. Third, future research will need to explore which types of constraints synergies respond to, the hierarchical nesting of synergies, the process of interactive synergy assembly, and the different possible profiles of cognitive mechanisms that underpin synergies. Finally, in view of the diversity of synergies their specific boundary conditions need to be taken into account to avoid premature generalizations, which the “bestiary” provided here has hopefully contributed to.
In closing, I would like to express my conviction that scholars of interpersonal synergy stand to benefit from methodological pluralism and engaging in cross-talk with other branches of interaction scholarship. Given that the framework is now coming of age, the time is also ripe for greater philosophical reflection that scrutinizes its assumptions carefully and compares it—and where possible connects it—with other frameworks.
Footnotes
Acknowledgments
I wish to thank Joshua Bergamin and Julian Zubek for their helpful comments on an earlier draft.
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author was funded by Austrian Science Fund grant P-33289. Award Recipient: Michael Kimmel.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.