Dance Like Someone is Watching : A Social Relations Model Study of Music-Induced Movement

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC 4.0 https://creativecommons.org/licenses/by-nc/4.0/ Dance Like Someone is Watching : A Social Relations Model Study of Music-Induced Movement © The Author(s) 2018. Published version Carlson, Emily; Burger, Birgitta; Toiviainen, Petri Carlson, E., Burger, B., & Toiviainen, P. (2018). Dance Like Someone is Watching : A Social Relations Model Study of Music-Induced Movement. Music and Science, 1, 1-16. https://doi.org/10.1177/2059204318807846 2018 Article Dance Like Someone is Watching: A Social Relations Model Study of Music-Induced Movement Emily Carlson, Birgitta Burger and Petri Toiviainen Abstract Although dancing often takes place in social contexts such as a club or party, previous study of such music-induced movement has focused mainly on individuals. The current study explores music-induced movement in a naturalistic dyadic context, focusing on the influence of personality, using five-factor model (FFM) traits, and trait empathy on participants’ responses to their partners. Fifty-four participants were recorded using motion capture while dancing to music excerpts alone and in dyads with three different partners, using a round-robin approach. Analysis using the Social Relations Model (SRM) suggested that the unique combination of each pair caused more variation in participants’ amount of movement than did individual factors. Comparison with self-reported personality and empathy measures provided some preliminary insights into the role of individual differences in such interaction. Self-reported empathy was linked to greater differences in amount of movement in responses to different partners. When looking at males only, this effect persisted for the whole body, head, and hands. For females, there was a significant relationship between participants’ Agreeableness (an FFM trait) and their partners’ head movements, suggesting that head movement may function socially to indicate affiliation in a dance context. Although consisting of modest effect sizes resulting from multiple comparisons, these results align with current theory and suggest possible ways that social context may affect music-induced movement and provide some direction for future study of the topic. Keywords Dance, dyadic movement, empathy, motion capture, personality Submission date: 20 September 2017; Acceptance date: 26 September 2018 Despite the ubiquity of posters, t-shirts, and internet memes urging us to “dance like no one is watching,” dance often takes place in social contexts such as clubs, concerts, or parties, where being seen by others is almost inevitable. Being seen may even be part of the point of dance; recent studies have suggested that synchronizing with others to music can promote social bonding (Quiroga Murcia, Kreutz, Clift, & Bongard, 2010; Rabinowitch et al., 2015; Vicary, Sperling, Von Zimmermann, Richardson, & Orgs, 2017) and even increase pain tolerance (Tarr, Launay, & Dunbar, 2016), supporting evolutionary theories that music and dance developed to support social cooperation necessary for human survival, in contexts such as group chorusing or sexual selection (Hodges, 2009; Huron, 2001; PhillipsSilver, Aktipis, & Bryant, 2010). Factors such as personality, felt and perceived emotion, music preference, and even sexual attractiveness have been related to qualities of free dance movements (Burger, 2013; Burger, Saarikallio, Luck, Thompson, & Toiviainen, 2013; Luck, Saarikallio, Burger, Thompson, & Toiviainen, 2010; Saarikallio, Luck, Burger, Thompson, & Toiviainen, 2013), all of which could possibly be decoded by observers, allowing dance to function as a kind of social signaling. Previous studies of dance have tended to focus on individual participants, and have shown free dance movement to be reflective of individual qualities such as personality, or felt or perceived emotion (e.g., Burger, 2013; Carlson, Department of Music, Arts and Culture, University of Jyva ¨skyla ¨, Finland Corresponding author: Emily Carlson, Department of Music, Arts and Culture, Jyvaskylan Yliopisto, P.O. Box 35, Jyvaskyla 40014, Finland. Email: [email protected] Music & Science Volume 1: 1–16 ªThe Author(s) 2018 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/2059204318807846 journals.sagepub.com/home/mns Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License (http://www.creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). l Burger, London, Thompson, & Toiviainen, 2016; van Dyck, Maes, Hargreaves, Lesaffre, & Leman, 2013; Luck et al., 2010). A few recent studies suggest that social context plays an important role in such music-induced movement. Solberg and Jensenius (2017) used a naturalistic electronic dance music (EDM) setting to show that the presence of and increased movement with other dancers increased subjective enjoyment of the dance experience. De Bruyn, Leman, and Moelants (2008) found that, in nine-year-old children, movement intensity as measured by Wii-remotes was increased in a social compared with individual setting, while van Dyck, Moelants, et al. (2013) found evidence for group entrainment in that there were greater correlations between dancers’ tempos and activity within, rather than between, groups. Using choreographed movement exercises, von Zimmermann, Vicary, Sperling, Orgs, and Richardson (2018) found that movement similarity between dyads in a group predicted group affiliation better than synchronization of the full group. However, while these studies provide valuable information about interpersonal coordination and the behaviors of a group, they are unable to provide information about how an individual’s improvised, spontaneous dance movements, such as have been shown to reflect personality (Luck et al., 2010), might be influenced by the presence of another dancer in a naturalistic setting. Do we, in fact, dance differently when someone is watching, and, moreover, when we are watching someone else dance? There is reason to believe that we do. Although there is a wealth of evidence that personality is generally stable across condition and time and indeed may be biologically based (Digman, 1990; Jang, Livesley, & Vemon, 1996; Letzring & Adamcik, 2015; Schaefer, Heinze, & Rotte, 2012; Soldz & Vaillant, 1999), there is similar evidence from social psychology research on that social context is a major determinant of behavior (Holtgraves, 2011; Malloy, Barcelos, Arruda, DeRosa, & Fonseca, 2005; Webster & Ward, 2011). It is thus imperative to consider both an individuals’ own tendencies and the influence of other individuals (and their own natural tendencies) when exploring social behavior (Griffin & Gonzalez, 2003; Wagerman & Funder, 2009). The widely-used five-factor model (FFM) of personality provides a useful measure of behavioral tendency through the measurement of five bipolar traits: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. Of these, Agreeableness and Extraversion are considered to be primarily interpersonal and therefore most relevant to social functioning, while Openness to Experience, Neuroticism and Conscientiousness are primarily intrapsychic (Ansell & Pincus, 2004). Agreeableness, which has also been labeled “likeability” and “friendliness,” is characterized by tact, kindness, warmth, conformity, and compliance (Graziano & Tobin, 2002) and has been linked to pro-social behavior (Graziano, Habashi, Sheese, & Tobin, 2007; Jensen-Campbell et al., 2002). Extraversion is characterized by positive affect, interest in social engagement and sensationseeking (Ashton, Lee, & Paunonen, 2002; Digman, 1990; Gray, 1970) and has been related to peer acceptance, goaloriented behavior, and modest advantages in decoding nonverbal behavior (Berry & Hansen, 2000; JensenCampbell et al., 2002; McCabe & Fleeson, 2012). Across cultures, females report higher levels of both Extraversion and Agreeableness than males (Costa, Terracciano, & McCrae, 2001; Schmitt, Realo, Voracek, & Allik, 2008). There is evidence that high levels of both Extraversion and Agreeableness provide advantages in social interactions, although this can depend on the particular combination of personalities in a given dyad (Berry & Hansen, 2000; Cuperman & Ickes, 2009; Isbister & Nass, 2000). Dyadic interactions may also be influenced by empathy. Empathy may be broadly defined as a complex psychological process, including both cognitive and affective components, which allows for the understanding of others’ emotions and perceptions (Decety & Jackson, 2004; Harari, Shamay-Tsoory, Ravid, & Levkovitz, 2010; ShamayTsoory, Tomer, Goldsher, Berger, & Aharon-Peretz, 2004; Zahavi, 2010). As with Agreeableness and Extraversion, males tend to report lower levels of trait empathy than females, while neuroimaging has shown differences between males and females in brain networks recruited for empathy (Schulte-Ru ¨ther, Markowitsch, Shah, Fink, & Piefke, 2008). Empathy is particularly important to study in the context of music and dance, as deficiencies and abnormalities in empathic function, such as autism and schizophrenia, have been associated with various serious mental disorders currently being treated in clinical-music and dance-therapeutic settings (Koch, Mehl, Sobanski, Sieber, & Fuchs, 2015; LaGasse, 2017; Lee, Jang, Lee, & Hwang, 2015), Promisingly, some studies have suggested a relationship between engaging in rhythmic entrainment, such as joint drumming activities or being swung in synchrony with a partner, and increased empathy or pro-social behavior (Kirschner & Tomasello, 2009; Rabinowitch et al., 2015; Rabinowitch, Cross, & Burnard, 2013; Rabinowitch & Meltzoff, 2017). Regarding free dance movement, Bamford and Davidson (2017) found that trait empathy was associated with better adjustment to abrupt tempo changes, while Carlson, Burger, London, Thompson, and Toiviainen (2016) found no relationship between empathy and adjustment to small tempo differences across stimuli. Both studies included individual dancing only; it may be that the effect of empathy on dance and other music-induced movement is clearer in an overtly social context. In the area of music performance, Novembre, Ticini, Schu ¨tz-Bosbach, and Keller (2014), found that more empathic participants appeared to rely more on motor simulations when adjusting their piano playing to a partner, supporting the importance of a social context in studying empathy in a dyadic movement context. Taken together, the previous work discussed above suggest that examining how individuals respond to a social setting in the context of 2Music & Science free dance, taking both individual and social factors into account, is likely to provide new insights into musicinduced movement in general. Using movement variables gathered simultaneously from two dyad members to investigate the influence of one dancer on the other (and vice versa) raises unique analytical complications that do not come up in individual dance research. Parametric statistical tests assume independence (Field, 2009), violations of this assumption can have serious consequences for tests of significance, with marked increases in the likelihood of both Type I and Type II errors (Field, 2009; Kenny & La Voie, 1984; Nimon, 2012; Wiedermann & von Eye, 2013), and may take the form of partner effects, the influence of one participant on another, mutual influence between partners, or common fate, where both partners are exposed to the same conditions resulting in similar responses (Kenny, 1996). For example, Dancer B might wave her hands while dancing and thereby encourage Dancer A to dance more vigorously than he would otherwise; this would be a partner effect. Dancer A and Dancer B may try to outdo each other in who can jump up and down the most, so the more Dancer A jumps the more Dancer B jumps and vice versa; this is an example of mutual influence. Finally, Dancer A and Dancer B may both be pretending to be chickens because they are both listening to the “Chicken Dance”; this is common fate. Untangling these influences on behavior is one of the main tasks of dyadic data analysis. One approach to this problem has historically been the use of confederates whose behavior is constrained (Griffin & Gonzalez, 2003). Although this simplifies matters statistically, multiple researchers have pointed out that to capture truly naturalistic behavior, both actors should be free to respond to the other as they wish (Cupeman & Ickes, 2009). Various statistical models have been developed to cope with, explore, and understand such non-independence, such as the Actor–Partner Interdependence Model (APIM) (Kenny et al., 2006) which considers the causal contribution of members of unique dyads while correcting for nonindependence, or the latent dyadic model which assesses shared variance between members (Griffin & Gonzalez, 2003). However, considering a person within the context of only one dyad, as opposed to multiple dyads, entails some notable limitations for generalization (Back & Kenny, 2010; Malloy et al., 2005). Imagine, for example, that the reason Dancer A dances more vigorously when Dancer B waves her hands is that Dancer A wants to please Dancer B because they are friends. We might erroneously conclude that hand-waving is related to vigorous dancing, whereas Dancer A may dance less vigorously if Dancer C waves her hands, because he does not particularly like Dancer C. These are examples of relationship effects, knowledge of which are crucial in understanding dyadic effects, but which, as can be seen in the example, can be mathematically determined only if each participant has more than one partner. The Social Relations Model (SRM) provides a structure for determining general knowledge about dyadic phenomena by comparing an individuals’ behavior with multiple partners (Back & Kenny, 2010; Gill & Swartz, 2001; Kenny et al., 2006). The aim of SRM is to separate causality of a given behavior as it takes place in a dyad (vigorous dancing, for example) into actor effects (the degree to which an individual tends to dance very vigorously), partner effects (the degree to which an individual tends to cause their partners to dance vigorously) and relationship effects (the degree to which an individual and a given partner have unique effects on the vigor of each other’s dancing when compared with their behaviors with other partners). Across a sample, these are expressed in terms of variances. If we wished to investigate whether the presence of a partner influences handwaving in dance and found a very large actor variance and a very small partner and relationship variances, we might conclude that the amount of hand-waving in dance depends chiefly by the person doing the handwaving, regardless of their partner. On the other hand, if we found a very large partner variance, we could conclude that the amount of hand-waving would vary depending on who a person’s partner is for a given interaction. 1 Similarly, a large relationship variance would indicate that the amount of hand-waving in determined by unique characteristics of a given dyad, such as friendship, liking, attraction, personality, and so on (Back & Kenny, 2010; Kenny et al., 2006; Kenny, Mannetti, Pierro, Livi, & Kashy, 2002; Kenny & Cook, 1999). The aim of the current study is to explore the relative influence of actor, partner, and relationship effects using full-body motion capture in a naturalistic, free dance movement context using the SRM, taking into account individual differences of personality and empathy. The study poses two research questions: 1. Does the presence of a partner moving to the same music affect music-induced movements of the individual? 2. Do characteristics of an individual, specifically Agreeableness, Extraversion, and trait empathy, relate to responsiveness to a partner in a dance setting? While previous work has been limited in its ability to extract movement from more than one body part per dancer (e.g., De Bruyn, Leman, & Moelants, 2008; Solberg & Jensenius, 2017), the current study employs full-body motion capture, making it is possible here to consider where specifically in the body social behavior might manifest in dance. While the whole body may be considered globally in dance (Carlson et al., 2016), in a social context we may also expect the hands to be important as hand gestures are particularly associated with communication (Bernardis & Gentilucci, 2006; Goldin-Meadow, 2006; Carlson et al. 3 Krauss, Chen, & Chawla, 1996) . Head movements have also been shown to be important in communication, particularly in non-verbally communicating rapport (Beck, Daughtridge, & Sloane, 2002; Helweg-Larsen, Cunningham, Carrico, & Pergram, 2004; TickleDegnen & Rosenthal, 1990). Both may be implicated in musical contexts as well (Davidson, 2001; Luck & Thompson, 2010; Thompson & Luck, 2012). An eyetracking study found that observers focused relatively little on dancers’ feet and core body, focusing more on the head. Therefore, in addition to the body as a whole, head and hand movement are considered separately in the current study. Kenny and Malloy (1988) reviewed SRM literature and found that across samples, against their expectations, partner effects (the degree to which an individual elicits consistent responses from all of their partners) were weak in affective and cognitive domains and virtually nonexistent in behavioral domains, except for being slightly more apparent in nonverbal communication. It is therefore reasonable to assume there may be a similar pattern in the current context. Kenny and Malloy suggest that this may be due to individual differences as well as experimental context. Given these observations, as well as previous findings and theoretical considerations regarding personality and empathy, we make the following predictions: H1: Participants will respond to the presence of a partner in a free dance context by changing aspects of their movement, specifically the overall amount of movement in the whole body, the head, and hands (e.g., GoldinMeadow, 2006; Helweg-Larsen et al., 2004; Kenny & Malloy, 1988). H2: In line with previous behavioral research, participants’ movements will be affected by the presence of a partner and their individual characteristics such that SRM analysis will show moderate actor and relationship effects and weak partner effects for movement of the whole body, hands, and head, in dyadic condition (Kenny & Malloy, 1988). H3: Participants will respond to the presence of a partner in a free dance context by changing aspects of their movement, specifically the overall amount of movement in the whole body, the head, and hands (e.g. GoldinMeadow, 2006; Helweg-Larsen et al., 2004; Kenny & Malloy, 1988). H4: Participants who are high in trait empathy, Agreeableness or Extraversion will vary their movement quality more in response to their partners, leading them to have smaller actor effects; as empathy and personality differ by sex, these correlations may also differ by sex (e.g., Baron-Cohen, 2009; Graziano et al., 2007; Schmitt et al., 2008). Methods Stimuli Since music preference and genre have previously been related to qualities of music-induced movement (e.g., Burger, 2013), a stimuli set including multiple genres was considered desirable, both to allow for nonindependencerelatedtocommonfate,andtoensurethat dyadic effects could not be attributed to the characteristics of a single genre. As genre in music is notably difficult to define (Pachet & Cazaly, 2000), to avoid researcher bias in stimuli selection a data-driven approach was devised using the methods described by Carlson, Saari, Burger, and Toiviainen (2017). A total of 2,407 tracks were collected from online music service Last.fm from those tagged by users as “danceable,” “dancing,” “head banging,” or “headbanging,” and which had been tagged with only one genre label (e.g., “Country” or “Jazz”). Tracks were retained only if they had a non-zero danceability score according to Echo Nest (the.echonest.com, an online music and data intelligence service where music categorization is determined by computational analysis of a given track’s acoustic features, including beat strength, tempo, and loudness), and only if the track’s tempo fell between 118–132 beats per minute (BPM). Four randomly selected excerpts from each genre were checked for tempo and stylistic consistency by the researchers, leaving 48 stimuli from 12 genres: Blues, Country, Dance, Funk, Jazz, Metal, Oldies, Pop, Rap, Reggae, Rock, and Soul. For a complete description of this stimuli-selection methodology, see Carlson et al. (2017). Participants (n¼210) were recruited using University student and departmental email lists and social media to rate their preference for these 48 excerpts in an online listening experiment using Survey Gizmo (www.surveygizmo.eu). Participants were entered into a lottery to win one of ten movie ticket vouchers, and were given feedback about their music preferences and personality upon completing the survey. Participants who completed the survey were also given the chance to sign up for the motion capture study. For the motion capture study, the number of genres was reduced from 12 to eight, and the number of stimuli per genre from four to two, in order to keep the experiment sufficiently short and limit the effects of fatigue. From each genre, two stimuli with the highest variability in preference ratings were chosen. This resulted in a final set of 16 from the following eight genres: Blues, Country, Dance, Jazz, Metal, Pop, Rap, and Reggae. Funk, Oldies, Rock, and Soul had the least variability in preference ratings and were therefore eliminated. Stimuli were 35 seconds in duration, including a 2.5-second fade-in and 2.5-second fade-out, as well as a sinusoidal beep at the start of each excerpt to mark the beginning for later synchronization with the motion capture data. 4Music & Science Participants A total of 73 participants (54 females) completed the motion capture experiment. However, due to several cancelations and “no-shows,” only 52 (38 female) completed the experiment in groups of four. Since the SRM requires a minimum of four participants (Kenny, Kashy & Cook, 2006), only data from these groups were included in the current analysis. Thus, each group consisted of four participants, resulting in six dyads per group. Participants ranged in age from 19 to 40 years (M¼25.74, SD ¼4.72). Thirty held bachelor’s degrees while 16 held master’s degrees. Thirty-three participants reported having received some formal musical training; 7 reported 1–3 years, 10 reported 7–10 years, while 16 reported 10 or more years of training. Seventeen participants reported having received some formal dance training; 10 reported 1–3 years, five reported 4–6 years, while two reported 7–10 years. Participants were of 24 different nationalities, with Finland, the United States, and Vietnam being the most represented. Participants received two movie ticket vouchers each for attending the experiment. All participants spoke and received instructions in English. Participant grouping Previous work has shown small but fairly consistent relationships between personality and music preference (e.g., Greenberg et al., 2016; Rawlings & Ciancarelli, 1997; Rentfrow, Goldberg, & Levitin, 2011). While it is not known how music preference affects music-induced movement in a dyadic setting, it is known that people make social judgements based on the music preferences of others (Rentfrow & Gosling, 2007; Rentfrow & Gosling, 2006; Rentfrow, McDonald, & Oldmeadow, 2009, Scha¨fer et al., 2015). Therefore, groups with evenly varied musical preferences were sought, such that effects were not confounded by unusual similarity or unusual difference in preference between participants in a given group. To achieve this, principal component analysis (PCA) was performed on the participants’ preference ratings of the 16 stimuli. The first component accounted for 22.6% of variance and included high negative loadings for both Metal genre excerpts and moderately high positive loadings for Reggae, Rap, and Pop excerpts, while loadings for other excerpts were small, suggesting a preference for upbeat, contemporary, danceable music and a dislike for Metal. The second component accounted for 22.1%of variance and included high positive loadings for both Jazz excerpts and moderately high positive loadings for Metal, suggesting a preference for one may relate to a dislike of the other. Scores for these first two components were subjected to a median-split, and participants were subsequently divided into four categories: high in both components, low in both components, or high in one and low in the other, respectively. Participants were grouped such that there was one member of each category in each group, limiting the possibility that movement effects could be attributed to unexpected convergence or lack of convergence in the dancers’ music preferences. This approach allowed for the use of multiple genres while still allowing for participants to have varied music preferences. Although an effort was made to prevent participants who knew each other well from being in the same group (for example, not granting requests from participants to be grouped with friends), a minority of participants (n 12) were acquainted before the experiment. Personality measures FFM personality dimensions were measured using the Big Five Inventory (BFI), a 44-item self-report measure in which participants rank their agreement on a seven-point Likert scale with statements such as “I see myself as someone who is talkative” or “ ...tends to be lazy” (Pervin & John, 1999). Only the Agreeableness (A) and Extraversion (E) scales were used in analysis, as these are considered interpersonal traits, most relevant for social functioning (Ansell & Pincus, 2004). In addition to personality, trait empathy was also measured. The Empathy Quotient (EQ), developed by Baron-Cohen and Wheelwright (2004), measures trait empathy as a whole, including both cognitive and affective aspects. For the current study, trait empathizing was measured using the shortform (22-item) version of the EQ, developed and validated by Wakabayashi et al. (2006). Apparatus The SRM dictates that to calculate actor and partner effects each individual must act with a minimum of three different partners. It was therefore necessary for participants to attend the experiment in groups of four, allowing for the creation of six unique dyads, and to capture not only multiple dancers but multiple dyads at once. Participants’ movements were recorded using a 12-camera optical motion capture system (Qualisys Oqus 5þ,Go¨teborg, Sweden) that tracked, at a frame rate of 120 Hz, the threedimensional positions of 21 reflective markers attached to each participant. Eight cameras were mounted on the ceiling, and four were placed near the wall of the capture space (see Figure 1). The locations of the numbered markers were as follows (where L ¼left, R ¼right, F ¼front, and B ¼ back): 1. LF head; 2. RF head; 3. B head; 4. L shoulder; 5. R shoulder; 6. sternum; 7. stomach; 8. LB hip; 9. RB hip; 10. L elbow; 11. R elbow; 12. L wrist; 13. R wrist; 14. L middle finger; 15. R middle finger; 16. L knee; 17. R knee; 18. L ankle; 19. R ankle; 20. L toe; 21. R tow. These can be seen in Figure 1(a). As multiple dancers in a motion capture space may also be difficult to differentiate once captured (Haugen & Nymoen, 2016), each participant was given either one, two, three or four extra markers Carlson et al. 5 attached to their leg. These markers were not used in data analysis. The musical stimuli were played in a random order in each condition via four Genelec 8030A loudspeakersandasub-wooferusingaMaxpatch(Cycling‘74, SanFrancisco,CA)runningonanAppleMaccomputer. The direct (line-in) audio signal of the playback and the synchronization pulse transmitted by the Qualisys cameras when recording were recorded using ProTools software (Avid Technology, Burlington, MA) in order to synchronize the motion capture data with the musical stimulus afterwards. To keep the experiment sufficiently short, it was necessary to capture multiple dancers and multiple dyads at once without their seeing one another. To facilitate this, a wall was installed that divided the visible capture space in half. An additional screen stood between the researchers and the capture space during motion capture to provide the participants with privacy from immediate observation so as to increase their comfort level. To minimize missing data, the capture space visible to the cameras was marked off on the floor using tape, and four of the cameras were set up on tripods on either side of the wall mitigate marker occlusion (Haugen & Nymoen, 2016). Additionally, Arabic numerals 1 through 4 were marked on the floor in order to guide participants where to be during dyadic conditions. The capture space set-up can be seen in Figure 2. Procedure For each group of four, participants were labeled A, B, C, or D, and wore a badge displaying their letter to enable easy identification amongst themselves and by the researchers. Participants were told to imagine that they were dancing in a social setting such as a club or party, and that they would hear a wide variety of music. They were asked to listen to the music and move as freely as they desired, but staying within the marked capture space. The aim of these instructions was to create a naturalistic paradigm, such that participants would feel free to behave as they might in a real-world situation. Stimuli were presented in a randomized order. In the first condition, participants moved alone in one half of the capture space. As only two participants could be motion captured at once in this way, this condition was repeated such that two participants completed their “individual” condition while the other two participants left the laboratory and completed personality questionnaires. In the remaining conditions, participants were organized into dyads on either side of the wall, such that all six possible combinations were recorded over three conditions: AB, AC, AD, BC, BD, and CD. This design is referred to as Round Robin in SRM research; see the section on statistical analysis for more detail (Back & Kenny, 2010). Participants were told that they could interact or not interact with their partner, as they felt comfortable, but were asked not to hold hands or switch places in the capture space, to avoid undue difficulty in labeling the data. To limit the effects of fatigue, participants were given 3to 10-minute breaks between each condition and were offered water, juice, and biscuits as light refreshment. Participants were informed that they were free to ask for a break or to stop the experiment at any time. After all conditions were complete, participants filled out a form providing demographic information, were debriefed about the experiment, and given the opportunity to ask questions and share feedback. The experiment lasted approximately two hours. Movement data processing Using the Motion Capture (MoCap) Toolbox (Burger & Toiviainen, 2013) in MATLAB [Version R2016b], movement data of the 21 markers were first trimmed to match the exact duration of the musical excerpts. Gaps in the data were linearly filled. Following this, the data were transformed into a set of 20 secondary markers—subsequently referred to as joints. The locations of these 20 joints are depicted in Figure 1(b). The locations of joints B, C, D, E, F, G, H, I, M, N, O, P, Q, R, S, and T are, in each case, identical to the locations of one of the original markers, while the locations of the remaining joints were obtained by averaging the locations of two or more markers: Joint A is the midpoint of the two back hip markers, joint J is the midpoint of the shoulder and hip markers, joint K is the midpoint of shoulder markers, and joint L is the midpoint of the three head markers. Acceleration data were chosen to assess participants’ overall amount of movement, and to reflect participants’ global complex movement in response to stimuli. Acceleration has been identified as a key movement feature in allowing musicians to synchronize to conductors’ gestures Figure 1. Marker and joint locations: (a) anterior view of the marker locations a stick figure illustration; (b) anterior view of the locations of the secondary markers/joints used in the analysis. 6Music & Science (Luck & Toiviainen, 2006), and has previously been used to give broad information about overall amount of movement within whole performances (Carlson et al., 2016). This approach allows for wide variety to exist between dancers’ free, improvised movements, as individuals may choose to embody the music in many different ways while still responding to and interacting with each other. Using overall acceleration also mitigates potential differences in dancers’ movements related to culture, while still providing broad information on essential aspects of their movements. Acceleration in three dimensions of the joints was calculated using numerical differentiation and a Butterworth smoothing filter (second order zero-phase digital filter). For each participant, the instantaneous magnitudes of acceleration were estimated for each joint and stimulus, and subsequently temporally averaged over each stimulus. Previous research has shown that rhythmic, timbral, and structural features, as well as perceived emotional content of music, can affect music-induced movement (Burger, Saarikallio, et al., 2013; Burger, Thompson, Luck, Saarikallio, & Toiviainen, 2013; Luck, Saarikallio, Burger, Thompson, & Toiviainen, 2014; Solberg & Jensenius, 2017). Since such features may vary significantly between genres (Lidy & Rauber, 2005; Pachet & Cazaly, 2000; Sordo, Celma, Blech, & Guaus, 2008), acceleration was averaged across genre rather than condition (condition in this case refers to all stimuli danced to with a given partner) to avoid the potential confound. Thus, for each participant, for each genre in each condition, mean acceleration across all 20 joints was obtained, as well as mean acceleration for the head and across the hands. These variables are hereafter referred to as Mean movement, Head movement, and Hand movement, respectively. These variables were analyzed using the SRM. Statistical analysis SRM analysis was implemented in MATLAB using a Round Robin design. An example of the Round Robin design can be seen in Table 1. In the hypothetical data presented in Table 1, participant A’s score when dancing with participant B is 4, while participant B’s score while dancing with participant A is 3. Participant A’s actor effect can be determined via scores in Table 1. Example of round robin design. Partner ABCD Actor A_434 B3_75 C23_ D632_ Figure 2. Motion capture space and divider wall. Carlson et al. 7 row A, as each of these represent participant A’s scores in various conditions (with various partners). Participant A’s partner effect can be determined via the scores in column A, as these represent the scores of each partner when they were acting with participant A. Were participant A to score a six, regardless of who their partner was, this would indicate a strong actor effect. Similarly, were each of participant A’s partners to score a six when acting with participant A, this would indicate a strong partner effect, that is, participant A would have a similar effect on all of his partners (see Kenny et al., 2006, pp. 194–198 for a thorough discussion of the estimation of SRM effects). Using the SRM, the score of a given participant dancing with a given partner, for example A dancing with B, is modeled using the following equation: XAB ¼mþaAþbBþgAB þE where mis the mean of all scores, a A is participant A’s actor effect (i.e., A’s level of consistency across interactions), b B is participant B’s partner effect (i.e., the consistency of responses of B’s partners to B), g AB is the unique response of A and B after controlling for each other’s actor and partner effects respectively, and Eis random error. To take an example from the context of dyadic dancing, the amount that Dancer A waves his hands while dancing with Dancer B is estimated as the group mean amount of hand waving plus Dancer A’s tendency to wave his hands, plus Dancer B’s tendency to elicit hand-waving from her dance partners, plus Dancer A’s unique response to Dancer B, plus random error. Participant A’s actor effect can be estimated as: aA¼ðn1Þ2 nðn2ÞMaAþn1 nðn2ÞMrAþn1 n2M where nis the group size, M aA is the participant’s actor scores (row A), M pA is the mean of their partners’ scores (column A), and Mis the mean of all observations. Similarly, participant A’s partner effect can be estimated as: bA¼ðn1Þ2 nðn2ÞMrAþn1 nðn2ÞMaAþn1 n2M The relationship effect of Dancer A with Dancer B, or the degree to which Dancer A’s response to Dancer B is unique given A’s actor effect and B’s partner effect, is estimated using the above terms as follows: gAB ¼XAB aAaBM Actor, partner and relationship variances are used to indicate the degree to which effects vary across individuals. In our example, the actor variance would indicate the degree to which some participants tend to wave their hands in dancing a lot while some tend to wave their hands very little. Variances are calculated using the mean squares of scores within and between dyads. The mean of variances is taken across groups. For further details of the estimation of the SRM can be found in Kenny et al. (2006), and well as Appendix B of Kenny (1994). Results To assess overall differences between individual and dyadic conditions, and to check for significant differences between musical genres, a two-way repeated-measures ANOVA was run for each of the three movement features using condition (individual or the mean of dyadic conditions) and genre as within-subject factors. For genre, Mauchly’s Test of Sphericity indicated that the assumption of sphericity had been violated for Mean movement (w 2 (27) ¼85.97, p< .001), Head movement (w 2 (27) ¼208.30, p< .001), and Hand movement (w 2 (27) ¼148.23, p< .001). Mauchly’s Test of Sphericity was also significant for genre and condition for Mean movement (w 2 (27) ¼64.76, p< .001), Head movement (w 2 (27) ¼210.06, p< .001), and Hand movement (w 2 (27) ¼190.55, p< .001). Therefore, a Greenhouse-Geisser correction was used in these cases. Results showed that there was a significant effect of conditiononHandmovement(F(1,51) ¼182.46, p< .05), but not on Mean movement or Head movement. There was a significant effect of genre on Mean movement (F(7,243) ¼22.91, p< .001), Head movement (F(7,137) ¼18.83, p< .001), and Hand movement (F(7,176) ¼14.67, p< .001). There was a significant interaction of genre*condition for Hand movement only (F(2.8,144) ¼3.97, p< .01). Bonferroni-corrected pairwise comparisons revealed a number of significant differences in movement features that differed per genre. Results are summarized in Figure 3, which shows that movement patterns per genre are similar across individual compared with dyadic conditions, with dyadic conditions showing more movement. Figures indicate that Metal stimuli resulted in the most head movement while Jazz stimuli resulted in the most movement overall. Results can be viewed in detail in Appendix A. To learn whether these movement differences between genres affected SRM results, actor and partner variances were calculated for each genre separately. The mean of these was taken to obtain an overall measure of variances. Actor variances (AV) and relationship variances (RV) by genre can be viewed in Table 2. As all partner variances were very close to zero or slightly negative (a statistical anomaly that can occur in the SRM, see Kenny et al. (2006) for an explanation), these variances are considered to be zero and not reported here. Actor variance ranged from .19 to .86, while relationship variances ranged from .42 to .89, suggesting that, between all dyads, movement features were differently influenced by characteristics of the individual, by characteristics of each given dyad, and that this differed somewhat by genre. Partner effects did not vary noticeably within any movement feature, suggesting that individual dancers did not tend to reliably elicit similar movement 8Music & Science Pachet, F., & Cazaly, D. (2000). A taxonomy of musical genres. In J. Mariani & D. Harman (Eds.), RIAO ’00: Content-based multimedia information access (pp. 1238–1245). Paris, France: Le Centre De Hautes Etudes Internationales D’Informatique Documentaire. Pecenka, N., & Keller, P. E. (2011). The role of temporal prediction abilities in interpersonal sensorimotor synchronization. Experimental Brain Research,211, 505–515. Pervin, L. A., & John, O. P. (1999). The Big-Five Trait Taxonomy: History, measurement and theoretical perspective. In O. P. John & S. Srivastava (Eds.), Handbook of personality: Theory and research (2nd ed., pp. 102–138). New York, NY: Guilford Press. Phillips-Silver, A. J., Aktipis, C. A., & Bryant, G. A. (2010). Entrainment: Foundations of rhythmic movement. Music Perception,28, 3–14. Quiroga Murcia, C., Kreutz, G., Clift, S., & Bongard, S. (2010). Shall we dance? An exploration of the perceived benefits of dancing on well-being. Arts & Health,2, 149–163. Rabinowitch, T.-C., Cross, I., & Burnard, P. (2013). Long-term musical group interaction has a positive influence on empathy in children. Psychology of Music,41, 484–498. Rabinowitch, T.-C., Knafo-Noam, A., Wheatley, T., Kang, O., Parkinson, C., Looser, C., ...Lawrence, J. (2015). Synchronous rhythmic interaction enhances children’s perceived similarity and closeness towards each other. PLoS One,10, e0120878. Rabinowitch, T.-C., & Meltzoff, A. N. (2017). Synchronized movement experience enhances peer cooperation in preschool children. Journal of Experimental Child Psychology,160, 21–32. Rawlings, D., & Ciancarelli, V. (1997). Music preference and the five-factor model of the NEO personality inventory. Psychology of Music,25, 120–132. Rentfrow, P. J., & Gosling, S. D. (2006). Message in a ballad: The role of music preferences in interpersonal perception. Psychological Science,17, 236–242. Rentfrow, P. J., Goldberg, L. R., & Levitin, D. J. (2011). The structure of musical preferences: A five-factor model. Journal of Personality and Social Psychology,100, 1139–1157. Rentfrow, P. J., & Gosling, S. D. (2007). The content and validity of music-genre stereotypes among college students. Psychology of Music,35, 306–326. Rentfrow, P. J., McDonald, J. A., & Oldmeadow, J. A. (2009). You are what you listen to: Young people’s stereotypes about music fans. Group Processes & Intergroup Relations,12, 329–344. Saarikallio, S., Luck, G., Burger, B., Thompson, M., & Toiviainen, P. (2013). Dance moves reflect current affective state illustrative of approach–avoidance motivation. Psychology of Aesthetics, Creativity, and the Arts,7, 296. Schaefer, M., Heinze, H.-J., & Rotte, M. (2012). Touch and personality: Extraversion predicts somatosensory brain response. NeuroImage,62, 432–438. Scha¨fer, T., Auerswald, F., Bajorat, I. K., Ergemlidze, N., Frille, K., Gehrigk, J., ...Wilker, T. (2015). The effect of social feedback on music preference. Musicae Scientiae, 20,1–6. Schmitt, D. P., Realo, A., Voracek, M., & Allik, J. (2008). Why can’t a man be more like a woman? Sex differences in Big Five personality traits across 55 cultures. Journal of Personality and Social Psychology,94, 168. Schulte-Ru ¨ther, M., Markowitsch, H. J., Shah, N. J., Fink, G. R., & Piefke, M. (2008). Gender differences in brain networks supporting empathy. NeuroImage,42, 393–403. Shamay-Tsoory, S. G., Tomer, R., Goldsher, D., Berger, B. D., & Aharon-Peretz, J. (2004). Impairment in cognitive and affective empathy in patients with brain lesions: Anatomical and cognitive correlates. Journal of Clinical and Experimental Neuropsychology,26, 1113–1127. Solberg, R. T., & Jensenius, A. R. (2017). Pleasurable and intersubjectively embodied experiences of electronic dance music. Empirical Musicology Review,11, 301–318. Soldz, S., & Vaillant, G. E. (1999). The Big Five personality traits and the life course: A 45-year longitudinal study. Journal of Research in Personality,33, 208–232. Sonnby-Borgstro¨m,M.,Jo¨nsson, P., & Svensson, O. (2003). Emotional empathy as related to mimicry reactions at different levels of information processing. Journal of Nonverbal Behavior,27, 3–23. Sonnby–Borgstro¨m, M. (2002). Automatic mimicry reactions as relatedtodifferencesin emotional empathy. Scandinavian Journal of Psychology,43, 433–443. Sordo, M., Celma, O., Blech, M., & Guaus, E. (2008). The quest for musical genres: Do the experts and the wisdom of crowds agree? In J. P. Bello, E. Chew, & D. Turnbull (Eds.), ISMIR 2008: Proceedings of the 9th international conference on music information retrieval (pp. 255–260). Philadelphia, PA: Drexel University. Tarr, B., Launay, J., & Dunbar, R. I. M. (2016). Silent disco: Dancing in synchrony leads to elevated pain thresholds and social closeness. Evolution and Human Behavior,37, 343–349. Thompson, M. R., & Luck, G. (2012). Exploring relationships between pianists’ body movements, their expressive intentions, and structural elements of the music. Musicae Scientiae, 16, 19–40. Tickle-Degnen, L., & Rosenthal, R. (1990). The nature of rapport and its nonverbal correlates. Psychological Inquiry,1, 285–293. Tomei, A., & Grivel, J. (2014). Body posture and the feeling of social closeness: An exploratory study in a naturalistic Setting. Current Psychology,33, 35–46. Vicary, S., Sperling, M., Von Zimmermann, J., Richardson, D. C., & Orgs, G. (2017). Joint action aesthetics. PLoS One,12, 1–21. Van Dyck, E., Maes, P. J., Hargreaves, J., Lesaffre, M., & Leman, M. (2013). Expressing induced emotions through free dance movement. Journal of Nonverbal Behavior,37, 175–190. Van Dyck, E., Moelants, D., Demey, M., Deweppe, A., Coussement, P., & Leman, M. (2013). The impact of the bass Carlson et al. 15 drum on human dance movement. Music Perception,30, 349–359. Von Zimmermann, J., Vicary, S., Sperling, M., Orgs, G., & Richardson, D. C. (2018). The choreography of group affiliation. Topics in Cognitive Science,10, 80–94. Wagerman, S. A., & Funder, D. C. (2009). Personality psychology of situations. In P. J. Corr & G. Matthews (Eds.), The Cambridge handbook of personality psychology (pp. 27–42). New York, NY: Cambridge University Press. Wakabayashi, A., Baron-Cohen, S., Wheelwright, S., Goldenfeld, N., Delaney, J., Fine, D., ...Weil, L. (2006). Development of short forms of the Empathy Quotient (EQ-Short) and the Systemizing Quotient (SQ-Short). Personality and Individual Differences,41, 929–940. Webster, M. M., & Ward, A. J. (2011). Personality and social context. Biological Reviews,86, 759–773. Wiedermann, W., & von Eye, A. (2013). Robustness and power of the parametric t test and the nonparametric Wilcoxon test under non-independence of observations. Psychological Test and Assessment Modeling,55, 39–61. Zahavi, D. (2010). Empathy, embodiment and interpersonal understanding: From Lipps to Schutz. Inquiry,53, 285–306. Appendix A Results of post-hoc pairwise comparisons comparing genre Tables show differences in mean between genres in Mean movement using a Bonferroni correction for multiple comparisons. Values are rounded to the nearest whole number. Negative values indicate that the column genre has a lower mean than the row (e.g., Blues is lower than Country). Non-significant differences are shown in gray text. 2) Head movement Blues Country Dance Jazz Metal Pop Rap Country 164 – Dance 591* 426* – Jazz 661* 497* 70 – Metal 927* 762* 336 266 – Pop 489* 324 102 172 438 – Rap 659* 494* 68 2 268 169 Reggae 196 31 394* 465* 731 293* 463* *p<.001. 3) Hand movement Blues Country Dance Jazz Metal Pop Rap Country 465 – Dance 1074* 610 – Jazz 1521* 1056* 446 – Metal 1051 586 24 470 – Pop 928* 463 146 592 122 – Rap 1198* 734* 124 322 148 240 Reggae 334 130 740* 1186* 716 594 *p< .001. 1) Mean movement Blues Country Dance Jazz Metal Pop Rap Country 484 – Dance 972* 489 – Jazz 1405* 921* 433 – Metal 285 199 688* 1120* – Pop 799* 315* 173 606* 514 – Rap 1041* 558 69 363 757* 242 Reggae 378 105 594 1027* 93 421* 663* *p<.001. 16 Music & Science