Nonverbal synchrony and affect in dyadic interactions

ORIGINAL RESEARCH ARTICLE published: 24 November 2014 doi: 10.3389/fpsyg.2014.01323 Nonverbal synchrony and affect in dyadic interactions Wolfgang Tschacher*, Georg M. Rees and Fabian Ramseyer Abteilung für Psychotherapie, Universitätsklinik für Psychiatrie und Psychotherapie, Universität Bern, Bern, Switzerland Edited by: In an experiment on dyadic social interaction, we invited participants to verbal interactions Michael W. Kraus, University of in cooperative, competitive, and ‘fun task’ conditions. We focused on the link between Illinois at Urbana-Champaign, USA interactants’ affectivity and their nonverbal synchrony, and explored which further variables Reviewed by: John F. Rauthmann, contributed to affectivity: interactants’ personality traits, sex, and the prescribed interaction Humboldt-Universität zu Berlin, tasks. Nonverbal synchrony was quantified by the coordination of interactants’ body Germany movement, using an automated video-analysis algorithm (motion energy analysis). Traits Katja Schlegel, University of Geneva, were assessed with standard questionnaires of personality, attachment, interactional style, Switzerland psychopathology, and interpersonal reactivity. We included 168 previously unacquainted *Correspondence: Wolfgang Tschacher, Abteilung für individuals who were randomly allocated to same-sex dyads (84 females, 84 males, Psychotherapie, Universitätsklinik für mean age 27.8 years). Dyads discussed four topics of general interest drawn from an Psychiatrie und Psychotherapie, urn of eight topics, and finally engaged in a fun interaction. Each interaction lasted Universität Bern, Laupenstrasse 49, 5 min. In between interactions, participants repeatedly assessed their affect. Using Bern 3010, Switzerland e-mail: [email protected] hierarchical linear modeling, we found moderate to strong effect sizes for synchrony to occur, especially in competitive and fun task conditions. Positive affect was associated positively with synchrony, negative affect was associated negatively. As for causal direction, data supported the interpretation that synchrony entailed affect rather than vice versa. The link between nonverbal synchrony and affect was strongest in female dyads. The findings extend previous reports of synchrony and mimicry associated with emotion in relationships and suggest a possible mechanism of the synchrony-affect correlation. Keywords: nonverbal synchrony, mimicry, imitation, embodiment, coordinated body-movement, motion energy analysis (MEA), body movement INTRODUCTION describes and models the laws that underlie such pattern formation When people are affectively moved, they tend to move accord- and self-organization in systems across disciplines (Haken, 1977). ingly – the close and bidirectional link between emotion and In the present empirical study, we focused on the combination bodily movement (Blake and Shiffrar, 2007; Hatfield et al., 2009) of both, embodiment and synchronization, in investigating the constitutes the core premise of the present empirical project. aspect of coordinated body-movement of individuals interacting This assumed link is consistent with the concept embodiment, in dyads. which has recently received support from research in psychol- This coordinated body-movement will be termed nonverbal ogy and cognitive science (Gallese, 2005; Tschacher and Bergomi, synchrony. We were interested whether nonverbal synchrony is a 2011). Embodiment denotes the theoretical perspective that men- manifestation of the affective states of the individuals in interac- tal processes must not be viewed isolated from bodily processes; tion and how synchrony and affect relate to each other, in different the essence of cognition is recognized in sensorimotor couplings types of verbal interaction. rather than in abstract information processing. A similar focus, In the context of psychotherapy dyads, synchrony was previ- a ‘corporeal turn,’ is currently observed in the humanities and ously found associated with positive affectivity reflected by better in philosophy (Fuchs and Jaegher, 2009; Alloa et al., 2012; Fuchs rapport and a positive quality of the therapeutic relationship and Koch, 2014). Thus, psychology is becoming increasingly sensi- (Ramseyer and Tschacher, 2011, 2014). Rapport was conceptu- tized to investigate the close association between mental and bodily alized according to Tickle-Degnen and Rosenthal (1990), who parameters such as body motion, gesture, or facial expression. identified three aspects: Mutual attentiveness, positivity, and coor- In a dynamical systems view, synchronization is a pervasive con- dination. Our present operationalization of nonverbal synchrony cept relevant to a large number of physical (Nicolis and Prigogine, covers their aspect of coordination, and the affective component 1977), biological (Rodriguez et al., 1999; Iacoboni, 2009), and inherent to positivity was measured by self-report questionnaire social (Salvatore and Tschacher, 2012; Schmidt et al., 2012; Char- data. trand and Lakin, 2013) systems. Synchronization means that pre- In the domain of social psychology, synchrony has been viously independent variables of a system can become entrained, studied as behavioral imitation (chameleon effect: Chartrand i.e., increasingly correlated, thereby reducing the degrees of free- and Bargh, 1999), and numerous investigations were concerned dom of the system. Synchronization events are typically found in with mutual adaptation during social exchange (interpersonal complex systems undergoing transitions from disordered states to adaptation: Burgoon et al., 1995), and movement entrainment states of higher order and coherence. Non-linear systems science (Richardson et al., 2008). Again, there is a link to emotion www.frontiersin.org November 2014 | Volume 5 | Article 1323 | 1 Tschacher et al. Synchrony and affect regulation (emotional contagion: Hatfield et al., 1994). The major- from previous research on nonverbal synchrony between patient ity of empirical studies concerned with nonverbal synchrony and therapist: Using the above mentioned conceptualization of have been conducted in affiliative contexts, i.e., within interac- synchrony, we found evidence for the association between syn- tional affordances that encouraged rapport between interactants chrony and positive aspects of the current state of a relationship (e.g., Nelson et al., 2014). To our knowledge, direct comparisons (measured at the level of single sessions of psychotherapy, i.e., of cooperative versus competitive interaction settings and their micro-outcome) as well as at the level of relationship develop- associations to nonverbal synchrony have almost never been sys- ment and maintenance (measured at termination of all therapy tematically investigated (Bernieri et al., 1996). A rare exception is sessions, i.e., macro-outcome). We therefore sought to extend Paxton and Dale (2013a), who explicitly addressed the impact of these findings from the psychotherapy setting (Ramseyer and conflict on nonverbal synchrony and found that it was disrupted in Tschacher, 2011) to an experimental context of dyadic verbal comparison to cooperative interactions. This scarcity of research interactions. We created an interactional setting that provided on nonverbal behavior in the two opposing settings of cooperation analogous instructions for both the cooperative as well as the com- and competition is rather surprising because the relevance of these petitive conditions. This was achieved by instructing participants aspects for negotiation or debate has long been discussed in social to engage in verbal discussions with the aim to either convince psychology (e.g., Thompson, 1990; Graziano et al., 1996; De Dreu the other fellow participant (=competition) or with the aim to et al., 2000; Seiter et al., 2009; Dunbar and Abra, 2010). Much defend a shared argumentational position against a third party research on interpersonal conflict was conducted in the clinical (=cooperation). ‘Micro-outcome’ in the present, more general field of marital interaction, specifically in marital conflict resolu- context was assessed by repeated ratings of participants’ affectivity tion and its association with marital satisfaction and divorce (e.g., directly after interactions. Gottman and Notarius, 2000). In our study, we sought to directly The hypotheses of the present study were fourfold. As our compare the effects of cooperative and competitive settings on specific methodology was not before applied outside of clinical set- nonverbal synchrony in two kinds of verbal debates. tings, we hypothesized that nonverbal synchrony was significantly Chartrand and Lakin (2013) report that, in various settings, present also in dyads of unacquainted individuals who engage in people synchronize and ‘mimic’ more whenever they perceive prescribed conversations (hypothesis 1, one-sided). We expected or wish a positive relationship. For example, the frequency of that synchrony would be associated positively to positive affect mimicry behaviors is predicted by attachment traits of adults and negatively to negative affect resulting from these conversations (Hall et al., 2012). Vice versa, synchrony entailed liking, coop- (hypothesis 2, two-sided). We wished to assess temporal sequences, erative behavior, and further prosocial effects. Thus, synchrony a potential indicator of the direction of causality between syn- was found to be both a consequence and antecedent of prosocial chrony and affect: Is nonverbal synchrony better explained by behavior and positive emotions. affect ratings prior to, or subsequent to, the respective interac- Several explanations were proposed for this linkage between tions where synchrony occurs (hypothesis 3, two-sided)? Finally, nonverbal synchrony and affect. Synchrony between interactants in an exploratory approach, we wished to model the dependence may support (or result from) empathic understanding (Bavelas of positive and negative affect on synchrony together with inter- et al., 1987; de Waal, 2007). Synchrony may also have a com- action type, sex, age, and personality traits of the interactants; municative function, creating a shared perspective of a situation additionally, possible differential effects of synchrony in com- (Scheflen, 1964; Wallbott, 1996). A number of studies inves- bination with sex or interaction type were tested (hypothesis 4, tigating the chameleon effect have shown that imitation has exploratory). beneficial effects on relationship quality and rapport (e.g., Stel and Vonk, 2010). Studies have demonstrated that synchronized MATERIALS AND METHODS motor activity increases both cooperation and affiliation (Hove SETTING and Risen, 2009; Wiltermuth and Heath, 2009). Even simple body- The project consisted of staged dyadic interactions between pre- movements, such as walking, are more synchronized in dyads with viously unacquainted persons of the same sex. Each dyad had positive relationships (Miles et al., 2010). Most of the studies ref- five interactions; four concerned social or political topics of com- erenced above explored the effects of synchrony that occurred mon interest, which were randomly drawn from an urn (without outside of participants’ conscious awareness. Similar findings for replacement); the fifth interaction was a fun task. For the initial the case of deliberate (instructed) synchronous motor activity have four interactions, eight different topics were prepared, such as ‘do been reported, demonstrating increases in, e.g., self-esteem and tuition fees at university make sense?,’ ‘media influences on child affiliation to interaction partners (Lumsden et al., 2014). development,’ ‘voluntary army or conscript army?’ and similar. Many areas of human social interaction generate synchronized These topics provided the basis for verbal debates between mem- behavior (Vallacher and Nowak, 2009): the list includes religious bers of each dyad. Participants of a dyad were provided with one settings (chorusing, dancing), sporting events (‘Mexican waves’ of two different and opposing written lists of specific arguments in the stadium), and the military (marching). These behavioral fitting these topics, which they could read in a 2-min preparation examples describe ritual processes where belongingness is nego- period prior to the interaction. The dyads were given instructions tiated and positive affect is desired. Research on interactional for the respective interaction; two instructions encouraged cooper- synchrony has likewise been conducted in diverse contexts (Davis, ation and two instructions encouraged competition. Cooperation 1982; Bernieri and Rosenthal, 1991; Burgoon et al., 1995). We instruction 1 was to develop a shared position with the strongest derived the interactional setting for the present experimental study arguments from the lists; the other cooperation instruction (2) Frontiers in Psychology | Personality and Social Psychology November 2014 | Volume 5 | Article 1323 | 2 Tschacher et al. Synchrony and affect additionally invoked an imagined third party against which the dyad was asked to discuss the best shared argumentation strategy. The competition instruction 1 was to argue, on the basis of each participant’s list of arguments, as convincingly as possible against the position of the interaction partner; the second competition instruction was asymmetrical, where one participant received a longer list of five strong arguments, and the other a list of two weaker arguments. The fifth interaction was a ‘fun task’ with the instruction, “Please design a five-course meal composed of dishes and drinks that both you and your interaction partner dislike.” The fun task was adapted from Chovil (1991). Our motivation for including the fun task was to investigate the relationship between affect and synchrony not only in themed discussions, but also in the engaging and humorous atmosphere reliably created by it; this assumption had been supported in the pilot phase of the present experiment. The fun-task type of cooperation was also more similar to previous work employing cooperative and com- petitive interactions (e.g., Bernieri et al., 1994). Given that the fun FIGURE 1 | Depiction of the experimental setup. Top left and right, digital task deviated from the previous four, more highly structured, topi- video cameras. cal interactions, we always included it at the end of the experiment outside of the randomization scheme. The sequence of instruc- tions was randomized and balanced, with 50% of dyads receiving available persons, with the rule that interactants in a dyad had the cooperation 1 – cooperation 2 – competition 1 – competition same sex and the same linguistic background, since the Swiss, Ger- 2 – fun task, and 50% receiving competition 1 – competition 2 – man, and Austrian variants of spoken German vary considerably. cooperation 1 – cooperation 2 – fun task. Participants’ education levels were generally high, 82% had high- All five interactions (duration 5 min each) of a dyad were school levels (“Matura” degree in the Swiss schooling system), and recorded using two cameras joined into a split-screen image. 57% had received higher education degrees (technical or peda- Prior to the experiment, it was ascertained that participants did gogic college 12%, university graduation 45%). Participants were not know each other, were fluent speakers of German, and had paid 30 Swiss francs for their time. no current or previous psychiatric diagnosis. Participants were unaware of the specific hypotheses of the study, especially of the MOTION ENERGY ANALYSIS (MEA) AND MEASUREMENT OF fact that the synchrony of movement was a variable of interest. NONVERBAL SYNCHRONY Thus, the goals of the experiment were not disclosed in detail until The idea of interactional synchrony was originally introduced the end of the experiment. Instead, participants were informed by Condon and Ogston (1966), who manually coded movement that the experiment sought to analyze processes taking place in changes occurring between consecutive frames of film recordings. verbal negotiations between unacquainted persons. The filming Other methods have relied on trained judges’ evaluations (Bernieri and audio recording of interactions was explained as being a pre- and Rosenthal, 1991). Technical advances in digital video process- requisite for subsequent evaluation of discussion performance. ing have since greatly facilitated quantification of movement based The interactions were conducted in a studio-like setting with on video recordings (Grammer et al., 1997; Grammer et al., 1999; standardized seating arrangements (Figure 1). The audio–visual Ramseyer and Tschacher, 2006, 2011; Nagaoka and Komori, 2008; recording of interactions was openly declared in the recruitment Kupper et al., 2010; Altmann, 2011; Paxton and Dale, 2013a,b). description and participants gave informed consent complying Motion energy analysis (MEA), an objective method to deter- with Swiss ethical regulation policies. When participants arrived mine changes in movement, relies on the same principle of at the lab, a research assistant introduced them to each other and frame-by-frame change introduced almost five decades ago. Yet, explained the sequence of events. Each person individually com- MEA provides a cost- and time-efficient alternative to manual pleted a battery of psychological measures prior to the interaction observer ratings because it is automated to continuously monitor sequences. the amount of change occurring in pre-defined regions of inter- est (Figure 2). The technical prerequisites for the recordings are PARTICIPANTS a static camera position and stable light conditions with constant Participants (N = 168) were 84 women and 84 men (mean age shutter-speed and aperture. Regions of interest should not overlap, 27.8 years, SD = 4.8), with men on average 3.37 years older and people should not occlude one another. [t(166) = 4.81; p < 0.0001). Participants were predominantly stu- Motion energy analysis involves several processing steps: Digi- dents or university graduates, and further persons recruited by the tized sequences (10 frames/s) of dyadic interactions were analyzed investigators. Fluency in German was an inclusion criterion; 87% with commercial video-analysis software (‘softVNS’ Rokeby, 2006; were Swiss citizens, other participants were from Germany (9%) ‘MaxMsp’ cycling ’74, 2006) that was customized for MEA and other European countries; all participants were Caucasian. (Ramseyer, 2008; see www.psync.ch for details). Motion energy Participants were assigned to dyads randomly from the pool of was defined as differences in gray-scale pixels between consecutive www.frontiersin.org November 2014 | Volume 5 | Article 1323 | 3 Tschacher et al. Synchrony and affect FIGURE 2 | Motion energy analysis (MEA). (A; top row), consecutive and threshold-adjusted motion energy values; abszissa: time in frames 1, 2, and 3 of split-screen video recordings. (A; bottom row), 1/10 s). The regions of interest (ROI) are shown as boxes in image corresponding images of motion energy 1*, 2*, and 3*. (B) time 1 and 1*. In images 1* and 2*, episodes of nonverbal synchrony series of individual motion energies (ordinate: smoothed, z-standardized occur. video-frames (Grammer et al., 1999), with differences indicating by step-wise shifting one time-series in relation to the other (50 body motion of the respective participant. One region of inter- steps in each direction, i.e., positive and negative lags). Cross- est (ROI) was adapted to each participant, covering the entire correlations were then standardized (Fisher’s Z) and their absolute body including the head and legs (see Figure 2A). Time-series of values were aggregated over the entire 5-min interval of an interac- these raw pixel-changes in each ROI were then smoothed with a tion, yielding one global value of nonverbal synchrony for each of moving average of 0.5 s, which reduces fluctuations due to signal- the five interactions of each dyad. The use of absolute values means distortion present in most videos. In order to account for different that both positive and negative cross-correlations contributed size regions of interest, data were z-transformed and a threshold positively to the 5-min synchrony measure. This strategy yields for minimal movement was individually calculated for each inter- synchrony values representative of the movement coordination action and each participant. Data filtered and corrected in this of an interacting dyad. These values were used as the synchrony manner were submitted for quantification of nonverbal synchrony variable in testing hypotheses 2, 3, and 4. (see below). The objectivity of this kind of automatic movement In order to evaluate the significance of synchrony values analysis is high, i.e., MEA is observer-independent once the pro- (hypothesis 1), a control for coincidental synchrony is needed. cedure is established. MEA provides objective and unobtrusive Bernieri et al. (1988) worked with so-called pseudointeractions “... quantitative measures of movement dynamics. by isolating the video image of each interactant and then pair- The quantification of nonverbal synchrony was based on ing them with the video images of other interactants recorded cross-correlations of participants’ movement time-series. In every in other interactions” (p. 245). Bernieri et al. (1988) were able 5-min interaction, motion energies of both participants were to show significantly higher synchrony in genuine mother–child cross-correlated (Boker et al., 2002) in window segments of 30 s interactions than in pseudointeractions. We implemented a simi- duration. In contrast to previous work with MEA in psychother- lar technique and generated pseudointeractions using automated apy dyads (Ramseyer and Tschacher, 2011), the window size of surrogate algorithms (Ramseyer and Tschacher, 2010). Surrogate 30 s was chosen to account for the relatively more dynamic and datasets (n = 100 out of each genuine dataset) were produced shorter turn-taking latencies in argumentative interaction com- by segment-wise (30 s segments) shuffling of the original data of pared to a psychotherapy setting. Segmentation into windows was interactants X and Y of a dyad, then aligning movement segments chosen in order to take into consideration the non-stationary of X with movement segments of Y that never actually occurred at nature of movement behaviors. Cross-correlations for positive the same time. This procedure kept the window-wise, individual and negative time-lags up to 5 s in steps of 0.1 s were computed progressive time structure of the real data intact but permuted the Frontiers in Psychology | Personality and Social Psychology November 2014 | Volume 5 | Article 1323 | 4 Tschacher et al. Synchrony and affect temporal location of window segments. Synchrony in pseudoint- global severity index of the SCL (Franke, 1994; Klaghofer and eractions (i.e., pseudosynchrony) was finally calculated identically Brähler, 2001), a measure used to assess general symptom dis- to the synchrony of the original data as described above. For the tress, such as worries, emotional instability, being nervous. Using statistical comparison of synchrony versus pseudosynchrony, the a five-point Likert scale, participants indicate to what extent they synchrony values of all 100 surrogate datasets of a dyad were com- experienced each of nine distress symptoms in the past week. We puted in each of the five interaction types, yielding a distribution used the mean of the items as an assessment of overall current of pseudosynchronies. Each dyad’s five genuine synchrony values symptomatology. For the original scales, internal consistencies were expressed as z values on the basis of the respective distribution between 0.63 and 0.85 with a test–retest reliability between 0.73 of pseudosynchronies. and 0.92 over a 1-week period were reported. The Saarbrücker Persönlichkeitsfragebogen SPF (Saarbrücken PROCESS MEASURE OF AFFECT personality questionnaire) is a reworked German version of the Prior to and subsequent to each of the five interaction tasks, all Interpersonal Reactivity Index (IRI: Davis, 1983), a questionnaire participants rated their own momentary affective state using the for the measurement of empathy. The SPF has four scales, per- Positive and Negative Affect Scale, PANAS (Krohne et al., 1996). spective taking, fantasy, empathic concern, and personal distress. The PANAS is a short standard instrument for the self-assessment Paulus (2009) reported internal consistencies of 0.66 to 0.71 and of emotional states. It consists of twenty emotion adjectives (e.g., good test–retest reliability. ‘active,’ ‘interested,’ ‘upset,’ ‘afraid’) that are rated on five-point In the present study, factor analysis was used to construct factors scales ranging from ‘very slightly or not at all’ to ‘extremely.’ The that represent these personality-based and clinical trait measures. 20 items load on two factors, positive affect and negative affect. The intention was to define strictly orthogonal factors that were Internal consistencies of the factors are usually found to be high, suited as predictors in later regression analyses (hypothesis 4), exceeding 0.80. The PANAS is a change-sensitive instrument and thereby avoiding collinearity of predictors. The small number of thus has low test-retest reliability. The positive and negative affect factors also reduces the problem of alpha inflation of statistical factors were used in hypotheses 2, 3, and 4. modeling, since the complete set of trait data comprised 22 sub- scales of the available instruments (five subscales of the NEO-FFI, MEASURES OF PERSONALITY TRAITS eight subscales of the IIP, four of the MAQ, one SCL-K-9 scale, four All participants filled out questionnaires covering different areas subscales of the SPF). Factor analysis with Maximum likelihood such as interactional style, interpersonal problems, personality estimation (JMP Pro 10 statistical software) yielded six factors traits, and psychological symptoms, to be used as predictors in with an Eigenvalue > 1. Using Varimax rotation, six orthogonal hypothesis 4. All instruments were based on participants’ self- factors were obtained by the linear composites of the subscales. reports. These questionnaires were given at baseline, prior to the These factors explained 57.4% of total variance. Factor loadings interactions. are given in Table 1. The factors (with explained variances) can be The Five Factor Personality Inventory (NEO-FFI, Borkenau and described as follows: Ostendorf, 1991; Costa and McCrae, 1992) is a multidimensional Factor 1 selfish-domineering (12.4%): This factor describes questionnaire for the self-assessment of the fundamental dimen- domineering, vindictive, and intrusive IIP styles with negative sions of personality (the postulated “Big Five” are neuroticism, NEO agreeableness and SPF perspective taking. extraversion, openness for new experience, agreeableness, consci- Factor 2 non-assertive-accommodating (10.6%): High loadings entiousness). The five dimensions are measured on the basis of 60 on several IIP scales, especially on overly accommodating, self- items with five-point scales. Cronbach’s alpha of the dimensions sacrificing, non-assertive, and intrusive/needy. in the German version ranged between 0.71 (openness) and 0.85, Factor 3 cold-avoidant (10.4%): This factor loads on IIP and test–retest reliabilities are commonly reported as adequate. scales cold/distant and socially inhibited, with MAQ avoidant and The Inventory of Interpersonal Problems (IIP, Horowitz et al., insecure attachment. Lower loadings are on NEO extraversion 1988, 1994) is a measure of current difficulties in interpersonal (negative). functioning. Apart from a total score indicative of the overall Factor 4 ambivalent-troubled (8.7%): This factor loads highest level of interpersonal problems, eight subscales pertaining to the on MAQ ambivalent-worried and ambivalent-merging attach- circumplex model of interpersonal behavior are assessed using a ment, with loadings on NEO neuroticism and SCL-K-9 symptoms five-point Likert scale (64 items ranging from 0 to 4). Horowitz of psychopathology. et al. (1994) reported an internal consistency ranging from 0.82 to Factor 5 introverted-distressed (8.3%): This factor loads on IIP 0.94 with a 10-week test–retest reliability of 0.80 to 0.90. non-assertive and socially inhibited, SPF personal distress, NEO The patients’ adult attachment style was measured with the neuroticism and, negatively, NEO extraversion. Measure of Attachment Qualities (MAQ, Carver, 1997). The 14 Factor 6 open-empathic (7.1%): This factor loads on SPF fan- items are scored on a four-point Likert scale ranging from 1 tasy, NEO openness, and SPF empathic concern and perspective to 4. The measure provides four scales of attachment types: taking. security, avoidance, ambivalence-worry, and ambivalence-merger. Carver (1997) reported internal consistencies of 0.69 to 0.76 and HIERARCHICAL LINEAR MODELING a test–retest reliability of 0.61 to 0.80 over a 6-week period. For the sake of testing hypothesis 3, we construed a two-level The SCL-K-9 is a short version of the Symptom Checklist SCL- model with interactions (Level 1) nested within dyads (Level 2). 90-R, composed of the nine items correlating highest with the The dependent variable was nonverbal synchrony of dyads, and www.frontiersin.org November 2014 | Volume 5 | Article 1323 | 5 Tschacher et al. Synchrony and affect Table 1 | Factor loadings after factor analysis with Varimax rotation of 22 questionnaire scales. Scale Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 selfish- non-assertive- cold-avoidant ambivalent- introverted- open-empathic domineering accommodating troubled distressed NEO neuroticism 0.32 0.15 0.04 0.52 0.56 0.12 NEO extraversion −0.01 0.02 −0.49 −0.07 −0.52 0.15 NEO openness −0.06 −0.07 0.00 −0.19 −0.05 0.61 NEO agreeableness −0.60 0.25 −0.33 −0.09 0.03 0.10 NEO conscientiousness −0.36 −0.07 0.02 0.03 −0.11 −0.05 MAQ security −0.04 0.05 −0.49 0.11 0.05 0.11 MAQ avoidance 0.10 0.13 0.69 0.06 0.09 −0.03 MAQ ambivalence-worry 0.01 0.08 0.08 0.77 0.10 −0.01 MAQ ambivalence-merger 0.09 0.11 0.04 0.50 0.09 −0.08 IIP domineering/controlling 0.77 0.13 0.21 0.16 0.03 0.03 IIP vindictive/self-centered 0.73 0.04 0.31 0.16 0.20 −0.12 IIP cold/distant 0.34 0.19 0.71 0.20 0.12 0.01 IIP socially inhibited 0.17 0.30 0.60 0.23 0.47 0.02 IIP non-assertive 0.02 0.55 0.32 0.11 0.59 0.12 IIP overly accommodating 0.00 0.88 0.12 0.08 0.21 0.01 IIP self-sacrificing 0.19 0.77 0.08 0.28 0.05 0.11 IIP intrusive/needy 0.65 0.45 −0.01 0.32 −0.06 0.16 SPF perspective taking −0.43 0.13 0.11 0.00 −0.17 0.50 SPF fantasy 0.16 0.03 −0.13 0.05 0.20 0.64 SPF empathic concern 0.04 0.23 −0.21 0.21 0.06 0.56 SPF personal distress 0.25 0.19 0.00 0.33 0.58 0.12 SCL-9 global 0.31 0.26 −0.05 0.48 0.33 0.24 Factor loadings exceeding 0.40 are in boldface. NEO, Five Factor Personality Inventory; MAQ, Measure of Attachment Qualities; IIP, Inventory of Interpersonal Problems; SPF, Saarbrücker Persönlichkeitsfragebogen; SCL-9, Symptom Checklist short version. the two predictors of synchrony were PANAS-defined affect of participants’ positive and negative affect. We used a step-up proce- interactants in a dyad, either measured before the interaction dure, exploring different combinations of predictors (i.e., models). or after the interaction. The direction of causality was esti- We evaluated all models in terms of Akaike’s information criterion mated on the basis of temporal sequence, using the significance (AIC) to select the best model; we used the software package JMP of these predictors. Thus, we were testing competing mod- Pro 10 (SAS Institute Inc, Cary, NC, USA) for computation of the els of temporal relationships between synchrony and affect: If corrected AIC (AICc) and for all further statistical analyses. We affect prior to the interaction was significant whereas affect applied mixed-effects analysis to explain the variance of the depen- after the interaction was not, this would speak for Granger- dent variable ‘positive (negative) affect’ by the following fixed causality‘affect causing synchrony.’ If affect prior to the interaction effects (i.e., predictors): ‘Synchrony,’‘Interaction type,’‘Interaction was insignificant whereas affect after the interaction was signifi- type × Synchrony,’ ‘Age,’ ‘Sex,’ ‘Sex × Synchrony,’ ‘Factor 1, 2,..., cant, this would speak for Granger-causality ‘synchrony causing 6.’ In all models, ‘Participant’ and ‘Dyad’ were entered as random affect.’ effects, which defined the dependency structure inherent to this For hypotheses 2 and 4, we modeled interactions (Level 1) hierarchical dataset (Raudenbush and Bryk, 2002). The best-fitting as nested within participants (Level 2) and nested within dyads and most parsimonious model was selected with the following pro- (Level 3). The dependent variables were participants’ positive cedure: We incrementally entered the predictors (fixed effects) in (negative) affect as measured by the PANAS. n = 840 measure- the sequence of the list above. Statistical significance (p < 0.05) ments were planned (168 participants × 5 interactions), with of the entered predictor was applied as a criterion to either keep three PANAS measurements missing due to data loss (hence, the current predictor and add the following predictor, or skip n = 837 in Tables 2 and 3). For the exploratory assessments of the current predictor and enter the following predictor. In this hypothesis 4 we tested different explanatory variables as predic- manner, nine models were computed for the dependent variable tors of affect in a systematic modeling procedure, separately for ‘Positive affect’ and the same number for ‘Negative affect.’ Finally, Frontiers in Psychology | Personality and Social Psychology November 2014 | Volume 5 | Article 1323 | 6  Tschacher et al. www.frontiersin.org Table 2 | Mixed effects models of N = 168 participants interacting in 84 dyads. Dependent variable, PANAS positive affect. Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) Fixed Effects Synchrony t = 3.65*** t = 3.29** t = 3.52*** t = 3.51*** t = 3.47*** t = 3.73*** t = 3.75*** t = 3.71*** Interaction type F = 10.3**** F = 13.3**** F = 13.3**** F = 13.3**** F = 14.1**** F = 14.1**** F = 14.1**** Interaction type × Synchrony F = 4.86** F = 4.86** F = 4.88** F = 4.41* F = 4.48* F = 4.47* Age t = 0.18 Sex[male] t = 1.72 t = 1.72 t = 1.89 t = 1.28 Sex[male] × Synchrony t = −3.62*** t = −3.56*** t = −3.60*** Factor 1 selfish-domineering t = .71 Factor 2 non-assertive-accommodating t = −.99 Factor 3 cold-avoidant t = −1.63 Factor 4 ambivalent-troubled t = −.95 Factor 5 introverted-distressed t = −2.42* t = −2.84** Factor 6 open-empathic t = 1.57 Random Effects Participant [Dyad] (% variance) 48.85 49.41 50.04 50.58 50.89 51.11 51.63 51.63 51.68 Dyad (% variance) 16.27 15.95 16.06 15.54 15.33 14.65 14.67 14.47 13.28 Whole model variance (%) 71.23 71.76 72.64 72.98 72.99 72.98 73.49 73.49 73.47 AIC 1385.4 1373.6 1369.1 1361.0 1370.3 1364.2 1352.6 1374.0 1350.6 Factor, orthogonal factors of participants’ traits; AIC, Akaike’s Information Criterion. PANAS, Positive and Negative Affect Scale. AIC minimum printed in boldface. For each model, fixed effects estimates, random November 2014 | Volume 5 | Article 1323 | 7 effects estimates, whole model variance and AIC are listed (top to bottom). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Synchrony and affect  Tschacher et al. Frontiers in Psychology | Personality and Social Psychology Table 3 | Mixed effects models of N = 168 participants interacting in 84 dyads. Dependent variable, PANAS negative affect. Model Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 1(n = 837) (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) (n = 837) Fixed Effects Synchrony t = −2.28* t = −0.62 t = −0.60 t = −0.61 t = −0.62 t = −0.57 Interaction type F = 50.9**** F = 46.6**** F = 50.9**** F = 50.9**** F = 51.1**** F = 53.73**** F = 53.70**** Interaction type × Synchrony F = 0.46 Age t = 0.66 Sex[male] t = 0.17 t = 0.16 Sex[male] × Synchrony t = −0.65 Factor 1 selfish-domineering t = 2.56* t = 2.61* Factor 2 non-assertive-accommodating t = 3.79*** t = 3.96*** Factor 3 cold-avoidant t = 0.00 Factor 4 ambivalent-troubled t = 4.13**** t = 4.37**** Factor 5 introverted-distressed t = 1.14 Factor 6 open-empathic t = 1.58 Random Effects Participant [Dyad] (% variance) 39.99 40.28 44.28 44.22 44.61 44.12 44.06 33.00 31.83 Dyad (% variance) 3.13 2.79 3.26 3.30 2.98 3.61 3.73 7.92 9.40 Whole model (% variance) 52.03 52.32 58.95 59.00 58.97 58.97 59.02 58.57 58.53 AIC 385.9 383.1 306.3 309.3 316.9 314.3 316.5 308.5 288.3 Factor, orthogonal factors of participants’ traits; AIC, Akaike’s Information Criterion; PANAS, Positive and Negative Affect Scale. AIC minimum printed in boldface. For each model, fixed effects estimates, random November 2014 | Volume 5 | Article 1323 | 8 effects estimates, whole model variance and AIC are listed (top to bottom). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Synchrony and affect Tschacher et al. Synchrony and affect AIC was used, a common approximation to model evidence. The Table 4 | Mixed effects models of N = 84 dyads in five interaction AIC includes both an accuracy and complexity term, in other conditions. Dependent variable, nonverbal synchrony. words, it identifies the most accurate model that can also provide a parsimonious explanation for observed data. Smaller AIC indi- Model a (n = 420) Model b (n = 420) cates the better model. The respective AIC-optimal models are Fixed Effects printed bold in the resulting tables below. PANAS positive before t = −0.16 RESULTS PANAS positive after t = 2.52* Nonverbal synchrony did not vary significantly across the four PANAS negative before t = 0.08 debate conditions that realized cooperation or competition PANAS negative after t = −1.98* [F(1,83) = 0.082; p = 0.775], but it significantly increased across Random Effect all five interactions [it was higher in the final interaction, the fun Dyad (% variance) 6.30 5.72 task; F(1,83) = 17.993; p < 0.0001]. Within the 5-minute tasks, Whole model variance (%) 12.66 11.15 there was no effect of time on synchrony across all 30-second windows [F(9,3986) = 0.933; p = 0.495]. AIC −1705.9 −1705.3 For both models, fixed effects estimates, the random effect estimate, whole HYPOTHESIS 1 model variance, and AIC are listed (top to bottom). AIC, Akaike’s Information The comparison of synchrony with pseudosynchronies derived Criterion; PANAS, Positive and Negative Affect Scale. *p < 0.05. from shuffled data demonstrated that nonverbal synchrony was significantly present at an above-chance level in this sample of 84 dyads and in all of the five interaction conditions. The mean significant effect in the model (t = 2.52, p < 0.05) whereas positive z values for genuine synchrony were 0.41 (cooperation 1), 0.71 affect prior to the interactions was not (t = −0.16, p = 0.87). Neg- (cooperation 2), 0.74 (competition 1), 0.78 (competition 2), and ative affect after the interactions was likewise significant in Model 1.11 (fun task). These mean z values are identical to effect sizes b (t = −1.98, p < 0.05) whereas negative affect prior to the inter- (Cohen’s d), i.e., they demonstrate moderate to strong effects. actions was not predictive (t = 0.08, p = 0.93). This supported the When the five interaction conditions are collapsed in three condi- assumption that, in the present sample, positive or negative affect tions, mean z values were 0.56 (cooperation), 0.76 (competition), may have been caused by synchrony rather than that affect caused and 1.11 (fun task). The fun task showed a significantly higher synchrony. synchrony than both competition (p < 0.05; d = 0.30 for the difference) and cooperation (p < 0.001; d = 0.49). HYPOTHESIS 4 The mean genuine synchrony values were 0.168 (cooperation The preferred direction of causality suggested by the test of 1), 0.170 (cooperation 2), 0.174 (competition 1), 0.178 (com- hypothesis 3 advised to treat positive and negative affect as the petition 2), and 0.195 (fun task). In three categories, genuine dependent variables in further explorations. As in the test of synchrony ranged from 0.169 (cooperation), to 0.176 (competi- hypothesis 2, individual affect ratings were used. Hierarchical tion), and 0.195 (fun task); the three categories of genuine syn- linear models with all predictors entered in a sequential fashion chrony were all significantly different from each other (p < 0.0001; are shown in Tables 2 and 3 (to simplify the tables, F values are d = 0.63 and 0.86). There was no general increase or decrease of given for the categorical predictor‘Interaction type,’ for the specific synchrony over time in the initial four interactions whose sequence t values see the following text). was randomized. Participants’ positive affect was best explained by Model 9 of Table 2, by the predictors ‘Synchrony,’‘Interaction type’ and, nega- HYPOTHESIS 2 tively, the personality factor 5‘introverted-distressed.’ As for‘Inter- Hierarchical linear models were computed for positive affect action type,’ the competition condition produced significantly (Table 2) and negative affect (Table 3). Individual affect ratings higher positive affect (t = 4.97, p < 0.0001), and the cooperation were used. In both tables, model 2 refers to the modulation of affect condition significantly lower positive affect (t = −2.79, p < 0.01), by the dyad’s synchrony alone. As expected, it was found that non- both relative to the fun task. Additionally, the interactions ‘Inter- verbal synchrony positively predicted positive affect (t = 3.65, action type × Synchrony’ and ‘Sex × Synchrony’ were significant p < 0.001) and predicted reduced negative affect (t = −2.28, predictors: The single tests of the former interaction showed that p < 0.05). in the cooperation condition, synchrony was linked less with pos- itive affect as compared to the fun task condition (t = −2.94, HYPOTHESIS 3 p < 0.01), whereas competition and fun task resulted in similar Hierarchical linear models were computed for nonverbal syn- linkages of synchrony and positive affect (t = 0.61, p = 0.55). The chrony (dependent variable) explained by positive affect prior to latter interaction ‘Sex × Synchrony’ means that in male dyads, syn- the respective interaction and by positive affect after this interac- chrony was less closely linked with positive affect than in female tion; affect ratings were the averages of the two dyad members in dyads (t = −3.60, p < 0.001, see Table 2). the interaction (Table 4, Model a). The procedure was repeated Negative affect of interactants was predicted by a different with negative affect prior to, and after, an interaction (Table 4, combination of independent variables (Model 9 in Table 3): Model b). We found that positive affect after the interactions was a Again, interaction type was a significant predictor. Interaction www.frontiersin.org November 2014 | Volume 5 | Article 1323 | 9 Tschacher et al. Synchrony and affect type fully mediated the influence of synchrony on negative affect, In line with their findings, our fun task ranked highest in syn- since synchrony was (negatively) predictive in Model 2, but lost chrony and lowest in negative affect (d = −0.66 in comparison significance as soon as interaction type was also entered. The to competition). These differences suggest that task affordances competition condition of ‘Interaction type’ was associated with were a driving force for synchrony and also influenced affect in significantly higher negative affect than the fun task (t = 10.05, this experiment. p < 0.0001), the cooperation condition and the fun task were not In general, we found objective evidence of nonverbal syn- statistically different (t = −0.49, p = 0.62). Three trait factors con- chrony in dyadic interactions among healthy individuals previ- tributed to negative affect: Factor 1 (selfish-domineering), Factor ously unknown to each other, and were able to quantify the 2 (non-assertive-accommodating) and Factor 4 (ambivalent- (considerable) magnitude of synchrony using effect sizes. Nonver- troubled). In addition to the predictive value of interaction type, bal synchrony in this project was obtained inconspicuously, i.e., we found a significant decrease of negative affect over the course participants were blind to our focus on the social synchroniza- of the experiment. tion of their body movement. This finding was possible on the basis of MEA methodology. It is a merit of this and similar frame- DISCUSSION differencing approaches (e.g., Delaherche et al., 2012) that they In the present study, we explored nonverbal synchrony in same- are independent of ratings by researchers, and can be applied in sex dyads that engaged in discussions of prescribed topics. The settings where participants may move without restrictions. Hence, method used was motion energy analysis (MEA), an objective the use of computerized video analysis offers a tool to improve the and automatized movement analysis of video recordings. MEA reproducibility of social psychology findings (Yong, 2012; Nelson has been previously applied only to clinical samples (movement et al., 2014). deficits in schizophrenia: Kupper et al., 2010; patient–therapist Affect was associated with nonverbal synchrony in the expected synchrony in psychotherapy sessions: Ramseyer and Tschacher, way: more synchronous interactions were entailed by higher pos- 2011). Here, we implemented the same measure with a differ- itive affect and lower negative affect. Comparing the association ent design and context, the first prospective application of MEA between synchrony and affect directly before and affect directly in healthy adults. Based on the analogous methodological setup, after a conversation, we found that the present data favored an we showed here that nonverbal synchrony was present to an interpretation that synchrony entailed affect and not vice versa. even higher extent than in the psychotherapy sample, with most This may indicate the predominant direction of causality between effect sizes exceeding the moderate range. Effect sizes (Cohen’s d) synchrony and affect in the present sample. Synchrony was not between d = 0.50 and 0.59 were reported in psychotherapy dyads merely an expression of participants’ affective states. Instead, our (Ramseyer and Tschacher, 2011). The effect sizes in the present findings are rather consistent with the interpretation that syn- sample were between 0.56 for interactions in the cooperation con- chrony caused affect, with a moderate to large effect size. To dition, to 0.76 in competitive discussions, and 1.11 in a fun task. our knowledge, this is the first project where the two directions Hence it was found that, other than might have been expected of causality were compared within the same dataset. Albeit not (Bernieri et al., 1994; Paxton and Dale, 2013b), (mildly) compet- resulting from an experimental design, this finding provides new itive conversations were more synchronized than conversations information on the generative mechanisms responsible for the with cooperation instructions. However, the comparison with synchrony-affect correlation. previous studies assessing both cooperative and competitive inter- Nonverbal synchrony measured by MEA has the great advan- actions should be made with caution: In our experiment, the tage of being objective, reproducible, and independent of the instructions for cooperative and competitive debates were of a (supposed) meanings of nonverbal gestures and postures – we similar nature: The activity in both conditions was identical, only created a content-independent measure in the spirit of non- the focus (“convince your partner” versus“convince a third party”) linear systems theory. As psychologists, however, we wished to was modified by our instructions. The available eight topics in our assess what the measurements ‘mean’: The exploratory analyses study were used for both cooperative and competitive discussions. (hypothesis 4) corroborated the strong link between synchrony This kind of similarity is clearly different from Bernieri et al.’s and positive affect in dyadic interactions (hypothesis 2) when, (1994) vacation-trip planning (cooperative) versus debate (com- importantly, all participants are unaware of the measurement petitive) instructions, or from Paxton and Dale’s (2013b) affiliative of nonverbal synchrony. This indicates that the embodiment of (“find and discuss media that both enjoy”) and argumentative dyadic interactions, as represented by correlated body movement, (“try to convince each other of your opinion”) instructions. is an important contributor to positive affect originating from Competition realized in the present project was linked with the these interactions, and hence to the emotional quality of the dyadic highest ratings for both positive and negative affect, thus compe- relationship. We showed that the direct link between bodily vari- tition was affectively arousing and not perceived entirely negative. ables and positive affect was not explained by personality traits, sex, On the other hand, cooperation realized in this project was linked or age of the interacting individuals. The influence of synchrony with the lowest ratings for positive affect (d = −0.22 in compar- was partially moderated by sex of participants and the type of ison to competition), which could be interpreted as a sign that the interactions, because the connection between synchrony and our implementation of a cooperative debate was less prone to positive affect was especially enhanced in female dyads, as well as elicit arousal in participants. Our fun task was more similar to in competitive and humorous/affiliative conversations. Yet none the trip-planning and media-discussion activities implemented in of the various explored predictors and interactions of predictors Bernieri et al.’s (1994) and Paxton and Dale’s (2013b) experiments: reduced the direct linkage between synchrony and positive affect. Frontiers in Psychology | Personality and Social Psychology November 2014 | Volume 5 | Article 1323 | 10 Tschacher et al. Synchrony and affect LIMITATIONS independent of interactants’ intentions or imitation goals. This As described in the Section “Materials and Methods,” MEA does notion is supported by the fact that synchrony in social contexts not assess the qualitative aspects of nonverbal behavior, which occurs spontaneously and usually unnoticed by interactants. Based means that MEA does not take into account qualitative features on theoretical considerations of complex systems theory applied to such as participants’ facial expression (e.g., smiling) or posture psychology (Kyselo and Tschacher, 2014), we would theorize that (e.g., approaching vs. disengaging postures). Future analyses based it is the individually varying affordance, or motivational valence, on observer codings will shed more light on the association of these of the respective discussion topics that may account for the not qualitative features of nonverbal behavior and their relation to syn- further explicated variance components of the random effect ‘par- chrony and affect. Further analytic avenues could be opened up ticipant’ (cf. Tables 2 and 3). Future research should therefore by using Actor-Partner-Interdependence Models (APIM; Kenny increasingly incorporate qualitative and even narrative analyses et al., 2006). The present study has limitations in the causality of nonverbal behavior and its associations with the context and assessment (hypothesis 3) since its design rests on temporal, not content of conversations. experimental, inference (i.e., causality inferred from temporal In both practical and scientific applications, we believe that sequence). We interpreted that nonverbal synchrony Granger- MEA methodology is suitable for the exploration of a wide range caused positive and inhibited negative affect, yet in principle third of social encounters where coordinative processes such as non- variables such as the motivational valence of the tasks and top- verbal synchrony occur. It offers objective measures for analyzing ics may have influenced the association between synchrony and individual movement as well as for assessing synchrony in dyads or affect. Nevertheless, Granger causality constitutes the only design more complex social aggregates. Given today’s high availability of that allows checking both causal directions in the same dataset, video recordings, MEA greatly facilitates research into nonverbally which appears preferable to conventional designs such as com- embodied aspects of emotion and communication. paring two experimental groups. The included sample has certain additional limitations as we recruited an academically advanced ACKNOWLEDGMENTS sample not representative of the general population. Also, we This study was conducted in the context of the European Cooper- formed only same-sex dyads owing to statistical power consid- ation in Science and Technology, COST action 2102, and funded erations and to previous research indicating erratic findings in by the Swiss Federal Department of Home Affairs (project number opposite-sex dyads – sex complicates interaction. The restric- C07.0036). We thank Eric Keller for important support, Yvonne tion to same-sex dyads obviously narrows the generalizability of Delevoye Turrell for methodological suggestions, and Christine the findings. Finally, we did not fully randomize the temporal Adamus for help in preparing the manuscript. sequence of interaction type in the experimental design because we expected the fun task to have extremely divergent implica- REFERENCES tions when positioned as first versus final task. Post hoc analyses, Alloa, E., Bedorf, T., Grüny, C., and Klass, T. N. (eds). (2012). Leiblichkeit. Geschichte however, showed that negative affect may have significantly dwin- und Aktualität eines Konzepts [Corporeality: History and actuality of a concept]. dled over time. Thus, the predictors ‘temporal sequence’ and Tübingen: Mohr-Siebeck. ‘interaction type’ were not fully disentangled, and in the mod- Altmann, U. (2011). “Investigation of movement synchrony using windowed cross- eling of negative affect the temporal sequence would probably be lagged regression,” in Analysis of Verbal and Nonverbal Communication and Enactment. Lecture Notes in Computer Science, Vol. 6800, eds A. Esposito, A. Vin- an important fixed effect. A minor problem was the inclusion ciarelli, K. Vicsi, C. Pelachaud, and A. Nijholt (Berlin: Springer-Verlag), 335–345. of somewhat older male than female participants, since age was doi: 10.1007/978-3-642-25775-9-31 explicitly controlled for and found to be an insignificant predic- Bavelas, J. B., Black, A., Lemery, C. R., Mullett, J., and Eisenberg, N. (1987). “Motor tor in all models. The experimental context in the project implied mimicry as primitive empathy,” in Empathy and its Development, ed. J. Strayer (New York: Cambridge University Press), 317–338. that the monetary compensation was given irrespective of a par- Bernieri, F. J., Davis, J. M., Rosenthal, R., and Knee, C. R. (1994). Interactional ticipant’s performance in the interactions. Thus, competitive and synchrony and rapport: measuring synchrony in displays devoid of sound and cooperative behavior was not specifically rewarded, which proba- facial affect. Pers. Soc. Psychol. Bull. 20, 303–311. doi: 10.1177/0146167294 bly affected the fidelity of these two interaction types, and may in 203008 part explain the unexpected synchrony we found in competitive Bernieri, F. J., Gillis, J. S., Davis, J. M., and Grahe, J. E. (1996). Dyad rapport and the accuracy of its judgment across situations: a lens model analysis. J. Pers. Soc. conversations. Psychol. 71, 110–129. doi: 10.1037/0022-3514.71.1.110 Bernieri, F. J., Reznick, S., and Rosenthal, R. (1988). Synchrony, pseudosyn- CONCLUSIONS chrony, and dissynchrony: measuring the entrainment process in mother-infant Concerning conceptualization of the main construct of this study, interactions. J. Pers. Soc. Psychol. 54, 243–253. doi: 10.1037/0022-3514.54. ‘synchrony’ is an appropriate and neutral concept, and in our opin- 2.243 Bernieri, F. J., and Rosenthal, R. (1991). “Interpersonal coordination: behavior ion preferable to concepts such as ‘mimicry’ or ‘imitation’: Like in matching and interactional synchrony,” in Fundamentals of Nonverbal Behav- the present study, synchrony is commonly found to occur uninten- ior. Studies in Emotion and Social Interaction, eds R. S. Feldman and B. Rime tionally, without the awareness of interactants, hence quite unlike (New York: Cambridge University Press), 401–432. what a ‘mime’ in ‘mimicking’ or ‘imitating’ would do. As stated Blake, R., and Shiffrar, M. (2007). Perception of human motion. Annu. Rev. Psychol. in the introduction, we believe that a tendency toward synchro- 58, 47–73. doi: 10.1146/annurev.psych.57.102904.19015 Boker, S. M., Xu, M., Rotondo, J. L., and King, K. (2002). Windowed nization may be derived from general assumptions and even laws cross-correlation and peak picking for the analysis of variability in the asso- concerning the self-organization of complex systems (Tschacher ciation between behavioral time series. Psychol. Methods 7, 338–355. doi: and Haken, 2007; Haken and Tschacher, 2010). Synchrony emerges 10.1037//1082-989X.7.3.338 www.frontiersin.org November 2014 | Volume 5 | Article 1323 | 11 Tschacher et al. Synchrony and affect Borkenau, P., and Ostendorf, F. (1991). Ein Fragebogen zur Erfassung fünf robuster Hatfield, E., Cacioppo, J. T., and Rapson, R. L. (1994). Emotional Contagion. Persönlichkeitsfaktoren [A questionnaire to assess five robust personality factors]. (Cambridge, MA: Cambridge University Press). Diagnostica 37, 29–41. Hatfield, E., Rapson, R. L., and Le, Y. C. L. (2009). “Emotional contagion Burgoon, J. K., Stern, L. A., and Dillman, L. (1995). Interpersonal Adaptation: and empathy,” in The Social Neuroscience of Empathy, eds J. Decety and W. Dyadic Interaction Patterns. Cambridge, MA: Cambridge University Press. doi: Ickes (Boston, MA: MIT Press), 19–30. doi: 10.7551/mitpress/9780262012973. 10.1017/CBO9780511720314 003.0003 Carver, C. S. (1997). Adult attachment and personality: converging evidence and Horowitz, L. M., Rosenberg, S. E., Baer, B. A., Ureño, G., and Villaseñor, V. S. a new measure. Pers. Soc. Psychol. Bull. 23, 865–883. doi: 10.1177/01461672 (1988). Inventory of interpersonal problems: psychometric properties and clinical 97238007 applications. J. Consult. Clin. Psychol. 56, 885–892. doi: 10.1037/0022-006X.56. Chartrand, T. L., and Bargh, J. A. (1999). The chameleon effect: the perception- 6.885 behavior link and social interaction. J. Pers. Soc. Psychol. 76, 893–910. doi: Horowitz, L. M., Strauss, B., and Kordy, H. (1994). IIP-D. Inventar zur Erfassung 10.1037/0022-3514.76.6.893 Interpersonaler Probleme. Deutsche Version. Weinheim: Beltz. Chartrand, T. L., and Lakin, J. L. (2013). The antecedents and consequences of Hove, M. J., and Risen, J. L. (2009). It’s all in the timing: interpersonal synchrony human behavioral mimicry. Annu. Rev. Psychol. 64, 285–308. doi: 10.1146/ increases affiliation. Soc. Cogn. 27, 949–960. doi: 10.1521/soco.2009.27.6.949 annurev-psych-113011-143754 Iacoboni, M. (2009). Imitation, empathy, and mirror neurons. Annu. Rev. Psychol. Chovil, N. (1991). Discourse-oriented facial displays in conversation. Res. Lang. Soc. 60, 653–670. doi: 10.1017/s0140525x07003123 Interact. 25, 163–194. doi: 10.1080/08351819109389361 Kenny, D. A., Kashy, D. A., and Cook, W. L. (2006). Dyadic Data Analysis. New York: Condon, W. S., and Ogston, W. D. (1966). Sound film analysis of normal and Guilford Press. pathological behavior patterns. J. Nerv. Ment. Dis. 143, 338–457. doi: 10.1097/ Klaghofer, R., and Brähler, E. (2001). Konstruktion und teststatistische Prüfung einer 00005053-196610000-00005 Kurzform der SCL-90–R [Construction and test statistical evaluation of a short Costa, P. T., and McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO PI- version of the SCL-90–R]. Z. Klin. Psychol. Psychiatr. Psychother. 49, 115–124. R) and NEO Five Factor Inventory. Professional Manual. Odessa, FL: Psychological Krohne, H. W., Egloff, B., Kohlmann, C.-W., and Tausch, A. (1996). Untersuchungen Assessment Resources. mit einer deutschen Version der“Positive and Negative Affect Schedule” (PANAS). Davis, M. (ed.). (1982). Interaction Rhythms. Periodicity in Communicative Behavior. Diagnostica 42, 139–156. New York: Human Sciences Press. Kupper, Z., Ramseyer, F., Hoffmann, H., Kalbermatten, S., and Tschacher, W. (2010). Davis, M. (1983). Measuring individual differences in empathy: evidence for a Video-based quantification of body movement during social interaction indicates multidimensional approach. J. Pers. Soc. Psychol. 44, 113–126. doi: 10.1037/0022- the severity of negative symptoms in patients with schizophrenia. Schizophr. Res. 3514.44.1.113 121, 90–100. doi: 10.1016/j.schres.2010.03.032 De Dreu, C. K., Weingart, L. R., and Kwon, S. (2000). Influence of social motives on Kyselo, M., and Tschacher, W. (2014). An enactive and dynamical systems the- integrative negotiation: a meta-analytic review and test of two theories. J. Pers. ory account of dyadic relationships. Front. Psychol. 5:452. doi: 10.3389/fpsyg. Soc. Psychol. 78, 889–905. doi: 10.1037/0022-3514.78.5.889 2014.00452 de Waal, F. B. (2007). “The “Russian doll” model of empathy and imitation,” in On Lumsden, J., Miles, L. K., and Macrae, C. N. (2014). Sync or sink? Interpersonal Being Moved: From Mirror Neurons to Empathy, ed. S. Braten (Amsterdam: John synchrony impacts self-esteem. Front. Psychol. 5:1064. doi: 10.3389/fpsyg.2014. Benjamins Publishing Company), 49–69. 01064 Delaherche, E., Chetouani, M., Mahdaoui, A., Saint-Georges, C., Viaux, S., and Miles, L. K., Griffiths, J. L., Richardson, M. J., and Macrae, C. N. (2010). Too late Cohen, D. (2012). Interpersonal synchrony: a survey of evaluation methods to coordinate: contextual influences on behavioral synchrony. Eur. J. Soc. Psychol. across disciplines. IEEE Trans. Affect. Comput. 3, 349–365. doi: 10.1109/T-AFFC. 40, 52–60. doi: 10.1002/ejsp.721 2012.1 Nagaoka, C., and Komori, M. (2008). Body movement synchrony in psychothera- Dunbar, N. E., and Abra, G. (2010). Observations of dyadic power in interpersonal peutic counseling: a study using the video-based quantification method. IEICE interaction. Commu. Monogr. 77, 657–684. doi: 10.1080/03637751.2010.520018 Trans. Inform. Syst. E91, 1634–1640. doi: 10.1093/ietisy/e91-d.6.1634 Franke, G. (1994). SCL-90-R. Die Symptom-Checkliste von Derogatis, deutsche Nelson, A., Grahe, J., Ramseyer, F., and Serier, K. (2014). Psychological data from Version. Weinheim: Beltz. an exploration of the rapport / synchroy interplay using motion energy analysis. Fuchs, T., and Jaegher, H. (2009). Enactive intersubjectivity: participatory sense- J. Open Psychol. Data 2, e5. doi: 10.5334/jopd.ae making and mutual incorporation. Phenomenol. Cogn. Sci. 8, 465–486. doi: Nicolis, G., and Prigogine, I. (1977). Self-Organization in Nonequilibrium Sys- 10.1007/s11097-009-9136-4 tems: From Dissipative Structures to Order through Fluctuations. New York: Fuchs, T., and Koch, S. C. (2014). Embodied affectivity: on moving and being Wiley-Interscience. moved. Front. Psychol. 5:508. doi: 10.3389/fpsyg.2014.00508 Paulus, C. (2009). Der Saarbrücker Persönlichkeitsfragebogen. Available at: Gallese, V. (2005). Embodied simulation: from neurons to phenomenal experience. http://psydok.sulb.uni-saarland.de/volltexte/2009/2363/pdf/SPF_Artikel.pdf Phenomenol. Cogn. Sci. 4, 23–48. doi: 10.1007/s11097-005-4737-z Paxton, A., and Dale, R. (2013a). Argument disrupts interpersonal synchrony. Q. J. Gottman, J. M., and Notarius, C. I. (2000). Decade review: observing marital Exp. Psychol. 66, 2092–2102. doi: 10.1080/17470218.2013.85308 interaction. J. Marriage Fam. 62, 927–947. doi: 10.1111/j.1741-3737.2000.00927.x Paxton, A., and Dale, R. (2013b). Frame-differencing methods for measuring Grammer, K., Filova, V., and Fieder, M. (1997). “The communication paradox bodily synchrony in conversation. Behav. Res. Methods 45, 329–343. doi: and possible solutions,” in New Aspects of Human Ethology, eds A. Schmitt, K. 10.3758/s13428-012-0249-2 Atzwanger, K. Grammer, and K. Schaefer (London: Plenum Press), 91–120. doi: Ramseyer, F. (2008). Synchronisation nonverbaler Interaktion in der Psychothera- 10.1007/978-0-585-34289-4_6 pie. [Synchrony of Nonverbal Interaction in Psychotherapy]. Doctoral Dissertation, Grammer, K., Honda, R., Schmitt, A., and Jütte, A. (1999). Fuzziness of nonverbal Institute of Psychology, Bern. Available at: http://www.psync.ch courtship communication unblurred by motion energy detection. J. Pers. Soc. Ramseyer, F., and Tschacher, W. (2006). Synchrony: a core concept for a Psychol. 77, 487–508. doi: 10.1037/0022-3514.77.3.487 constructivist approach to psychotherapy. Constructivism Hum. Sci. 11, 150–171. Graziano, W. G., Jensen-Campbell, L. A., and Hair, E. C. (1996). Perceiving inter- Ramseyer, F., and Tschacher, W. (2010). “Nonverbal synchrony or random coinci- personal conflict and reacting to it: the case for agreeableness. J. Pers. Soc. Psychol. dence? How to tell the difference,” in Development of Multimodal Interfaces: Active 70, 820. doi: 10.1037/0022-3514.70.4.820 Listening and Synchrony, eds A. Esposito, N. Campbell, C. Vogel, A. Hussain, and Haken, H. (1977). Synergetics–An Introduction. Nonequilibrium Phase-Transitions A. Nijholt (Berlin: Springer), 182–196. and Self-Organization in Physics, Chemistry and Biology. Berlin: Springer. Ramseyer, F., and Tschacher, W. (2011). Nonverbal synchrony in psychother- Haken, H., and Tschacher, W. (2010). A theoretical model of intentionality with an apy: coordinated body-movement reflects relationship quality and outcome. application to neural dynamics. Mind and Matter 8, 7–18. J. Consult. Clin. Psychol. 79, 284–295. doi: 10.1037/a0023419 Hall, N. R., Millings, A., and Boucas, S. B. (2012). Adult attachment orienta- Ramseyer, F., and Tschacher, W. (2014). Nonverbal synchrony of head- and body- tion and implicit behavioral mimicry. J. Nonverbal Behav. 36, 235–247. doi: movement in psychotherapy: different signals have different associations with 10.1007/s10919-012-0136-7 outcome. Front. Psychol. 5:979. doi: 10.3389/fpsyg.2014.00979 Frontiers in Psychology | Personality and Social Psychology November 2014 | Volume 5 | Article 1323 | 12 Tschacher et al. Synchrony and affect Raudenbush, W., and Bryk, A. S. (2002). Hierarchical Linear Models. Applications Tschacher, W., and Haken, H. (2007). Intentionality in non-equilibrium systems? and Data Analysis Methods. London: Sage. The functional aspects of self-organized pattern formation. New Ideas Psychol. Richardson, D., Dale, R., and Shockley, K. (2008). “Synchrony and swing in con- 25, 1–15. doi: 10.1016/j.newideapsych.2006.09.002 versation: coordination, temporal dynamics, and communication,” in Embodied Vallacher, R. R., and Nowak, A. (2009). “The dynamics of human experience: Communication in Humans and Machines, eds I. Wachsmuth, M. Lenzen, and G. fundamentals of dynamical social psychology,” in Chaos and Complexity in Knoblich (New York: Oxford University Pres), 75–94. Psychology: The Theory of Nonlinear Dynamical Systems, eds S. J. Guastello, Rodriguez, E., George, N., Lachaux, J. P., Martinerie, J., Renault, B., and Varela, F. M. Koopmans, and D. Pincus (Cambridge, MA: Cambridge University Press), J. (1999). Perception’s shadow: long-distance synchronization of human brain 370–401. activity. Nature 397, 430–433. doi: 10.1038/17120 Wallbott, H. G. (1996). “Congruence, contagion, and motor mimicry: mutualities in Rokeby, D. (2006). softVNS 2.17 [Computer Software (http://www.softVNS.com)]. nonverbal exchange,” in Mutualities in Dialogue, eds I. Markova, C. F. Graumann, Toronto, Canada. and K. Foppa (Cambridge, MA: Cambridge University Press), 82–98. Salvatore, S., and Tschacher, W. (2012). Time dependency of psychotherapeu- Wiltermuth, S. S., and Heath, C. (2009). Synchrony and cooperation. Psychol. Sci. tic exchanges: the contribution of the theory of dynamic systems in analyzing 20, 1–5. doi: 10.1111/j.1467-9280.2008.02253.x process. Front. Psychol. 3:253. doi: 10.3389/fpsyg.2012.00253 Yong, E. (2012). Replication studies: bad copy. Nature 485, 298–300. doi: Scheflen, A. E. (1964). The significance of posture in communication systems. 10.1038/485298a Psychiatry 27, 316–331. Schmidt, R. C., Morr, S., Fitzpatrick, P., and Richardson, M. J. (2012). Measuring Conflict of Interest Statement: The authors declare that the research was conducted the dynamics of interactional synchrony. J. Nonverbal Behav. 36, 263–279. doi: in the absence of any commercial or financial relationships that could be construed 10.1007/s10919-012-0138-5 as a potential conflict of interest. Seiter, J. S., Weger, H., Kinzer, H. J., and Jensen, A. S. (2009). Impression man- agement in televised debates: the effect of background nonverbal behavior on Received: 05 September 2014; accepted: 31 October 2014; published online: 24 audience perceptions of debaters’ likeability. Commu. Res. Rep. 26, 1–11. doi: November 2014. 10.1080/08824090802636959 Citation: Tschacher W, Rees GM and Ramseyer F (2014) Nonverbal synchrony and Stel, M., and Vonk, R. (2010). Mimicry in social interaction: benefits for mim- affect in dyadic interactions. Front. Psychol. 5:1323. doi: 10.3389/fpsyg.2014.01323 ickers, mimickees, and their interaction. Br. J. Psychol. 101, 311–323. doi: This article was submitted to Personality and Social Psychology, a section of the journal 10.1348/000712609X46542 Frontiers in Psychology. Thompson, L. (1990). Negotiation behavior and outcomes: empirical evidence and Copyright © 2014 Tschacher, Rees and Ramseyer. This is an open-access article dis- theoretical issues. Psychol. Bull. 108, 515–532. doi: 10.1037/0033-2909.108.3.515 tributed under the terms of the Creative Commons Attribution License (CC BY). The Tickle-Degnen, L., and Rosenthal, R. (1990). The nature of rapport and its nonverbal use, distribution or reproduction in other forums is permitted, provided the original correlates. Psychol. Inquiry 1, 285–293. doi: 10.1207/s15327965pli0104-1 author(s) or licensor are credited and that the original publication in this journal is cited, Tschacher, W., and Bergomi, C. (eds). (2011). The Implications of Embodiment: in accordance with accepted academic practice. No use, distribution or reproduction is Cognition and Communication. Exeter: Imprint Academic. permitted which does not comply with these terms. www.frontiersin.org November 2014 | Volume 5 | Article 1323 | 13