cacioppo et al ns of sync

7/18/2019 Cacioppo Et Al Ns of Sync

1/17

1

2 YOU ARE IN SYNC WITH ME: NEURAL CORRELATES3 OF INTERPERSONAL SYNCHRONY WITH A PARTNER

4 S. CACIOPPO, a,bQ1 H. ZHOU, c G. MONTELEONE, a

5 E. A. MAJKA, c K. A. QUINN, d,e A. B. BALL, a

6 G. J. NORMAN, c G. R. SEMIN f,g AND J. T. CACIOPPO a,b,c*

7 a Center for Cognitive and Social Neuroscience, HPEN8 laboratory, University of Chicago, 940 East 57th Street, Chicago,9 IL 60637, United StatesQ2

10 b Department of Psychiatry and Behavior Neuroscience, University of11 Chicago, A27 South Maryland Avenue, Chicago, IL 60637, United12 States

13 c Department of Psychology, University of Chicago, 584814 South University Avenue, Chicago, IL 60637, United States

15 d School of Psychology, University of Birmingham,16 Edgbaston, Birmingham B15 2TT, United KingdomQ3

17 e Department of Psychology, DePaul University, 2219 North18 Kenmore Avenue, Chicago, IL 60657, United States

19 fFaculty of Social and Behavioral Sciences, Utrecht20 University, Heidelberglaan 1, 3584 CS Utrecht, Netherlands

21 g Department of Psychology, Koc University, Istanbul, Turkey

22 AbstractInterpersonal synchrony is characterized by a

temporary alignment of periodic behaviors with another per-

son. ThisQ4 process requires that at least one of the two indi-

viduals monitors and adjusts his/her movements to

maintain alignment with the other individual (the referent).

Interestingly, recent research on interpersonal synchronyhas found that people who are motivated to befriend an

unfamiliar social referent tend to automatically synchronize

with their social referents, raising the possibility that syn-

chrony may be employed as an affiliation tool. It is unknown,

however, whether the opposite is true; that is, whether the

person serving as the referent of interpersonal synchrony

perceives synchrony with his/her partner or experiences

affiliative feelings toward the partner. To address this

question, we performed a series of studies on interpersonal

synchrony with a total of 103 participants. In all studies, par-

ticipants served as the referent with no requirement to mon-

itor or align their behavior with their partners. Unbeknown to

the participants, the timings of their partners movements

were actually determined by a computer program based on

the participants (i.e., referents) behavior. Overall, our

behavioral results showed that the referent of a synchrony

task expressed greater perceived synchrony and greater

social affiliation toward a synchronous partner (i.e., one dis-

playing low mean asynchrony and/or a narrow asynchrony

range) than with an asynchronous partner (i.e., one display-

ing high mean asynchrony and/or high asynchrony range).

Our neuroimaging study extended these results by demon-

strating involvement of brain areas implicated in social cog-

nition, embodied cognition, self-other expansion, and action

observation as correlates of interpersonal synchrony (vs.

asynchrony). These findings have practical implications

for social interaction and theoretical implications for under-

standing interpersonal synchrony and social coordination.

2014 Published by Elsevier Ltd. on behalf of IBRO.

Key words: social neuroscience, fMRI, interpersonal

synchrony, dyads, shared representations.23

24INTRODUCTION

25Early studies of synchrony focused on mechanisms

26underlying a persons ability to synchronize movements27with some referent, such as a metronome (cf, Repp,

282005). Interpersonal synchrony, the alignment in time of

29the periodic movements of two or more individuals, has

30also been investigated because of its putative social con-

31sequences. Interpersonal synchrony promotes an array of

32positive interpersonal outcomes, such as affiliation (Hove

33and Risen, 2009), liking (Miles et al., 2009), rapport

34(Vacharkulksemsuk and Fredrickson, 2012), and emo-

35tional support satisfaction (Jones and Wirtz, 2007). Inter-

36personal synchrony also leads to outcomes that extend

37beyond individuals to promote groups, including coopera-

38tion (Wiltermuth and Heath, 2009) and compassion

39(Valdesolo and DeSteno, 2011). Functionalist accounts

40of synchrony posit that the primary purpose of synchrony41is to foster social bonds (Semin, 2007; Semin and

42Cacioppo, 2008) and strengthen the collective (McNeill,

431995; Ehrenreich, 2006; Haidt et al., 2008; Haidt, 2012).

44McNeill (1995) argued that synchrony played an important

45role in the ascension of our species, and previous inves-

46tigations have documented motivational factors that pro-

47mote interpersonal synchrony and various social

48consequences of synchrony (Bernieri, 1988; Cappella,

491997; Lakin and Chartrand, 2003; Hove and Risen,

502009; Marsh et al., 2009; Wiltermuth and Heath, 2009;

51Miles et al., 2010; Paladino et al., 2010; Valdesolo and

http://dx.doi.org/10.1016/j.neuroscience.2014.07.051

0306-4522/ 2014 Published by Elsevier Ltd. on behalf of IBRO.

*Correspondence to: J. T. Cacioppo, The University of Chicago,

Department of Psychology, 5848 South University Avenue, Chicago,

IL 60637, United States.

E-mail address: [email protected] (J. T. Cacioppo).

Abbreviations: ACC, anterior cingulate cortex; ANOVA, analysis ofvariance; BOLD, blood oxygenation level-dependent; dACC, dorsalanterior cingulate; DLPFC, dorsolateral prefrontal cortex; dmPFC,dorsomedial prefrontal cortex; fMRI, functional magnetic resonanceimaging; FOV, field of view; IPL, inferior parietal lobule; M1, primarymotor cortex; pre-SMA, pre-supplementary motor area; PMC, premotorcortex; S1, primary somatosensory cortex; SMA, supplementary motorcortex; SMS, sensorimotor synchronization; vmPFC, ventromedialprefrontal cortexQ5 .

Neuroscience xxx (2014) xxxxxx

Please cite this article in press as: Cacioppo S et al. You are in sync with me: Neural correlates of interpersonal synchrony with a partner. Neuroscience

(2014),http://dx.doi.org/10.1016/j.neuroscience.2014.07.051

1

NSC 15591 No. of Pages 17

5 August 2014
http://-/?-http://dx.doi.org/10.1016/j.neuroscience.2014.07.051mailto:[email protected]://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051mailto:[email protected]://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-http://-/?-


2/17

52 DeSteno, 2011; Vacharkulksemsuk and Fredrickson,

53 2012).

54 The research to date has focused on a particular type

55 of interpersonal synchrony, in which the participants

56 share the goal of synchronizing (either directly with their

57 fellow participants, or with some other cue that results in

58 their synchronization with each other). Interpersonal59 synchrony can take other forms, however, and

60 individuals may find themselves being the referents for

61 others synchronization goals without sharing those

62 goals for themselves. In the current research, we

63 investigated experimentally whether being the referent

64 for a partner who responds in a more or less

65 synchronous fashion (rather than an intentional

66 contributor to the synchrony produced by a partner)

67 affects the referents perceived synchrony with and

68 affiliative response toward the partner. Second, we

69 investigated the neural correlates of interpersonal

70 synchrony (vs. asynchrony) in this referent.

71 Three processes underlying the emergence of

72 interpersonal synchrony

73 The temporal relation between the movements of two or

74 more individuals determines the degree of interpersonal

75 synchrony. However, the same state of synchrony may

76 be the outcome of any of three distinct production

77 processes, which we refer to as orchestration, reciprocal

78 entrainment, and unilateral entrainment. In orchestration,

79 synchrony is achieved when two or more individuals

80 entrain their movements to an external pacesetter (e.g.,

81 the pacing sound of a metronome) that directs the

82 shared movement pattern, much like a conductor

83 leading scores of musicians. For example, Hove and84 Risen (2009) manipulated interpersonal synchrony by

85 having participants tap to beats created by a metronome.

86 In reciprocal entrainment, synchrony is achieved

87 through a give-and-take process in which individuals

88 within a system (e.g., dyad) monitor each other and

89 adjust their own movement in a mutual fashion. For

90 example, Oullier et al. (2008)found that dyadic interper-

91 sonal synchrony reflected movements that were distinct

92 from individuals movements prior to the interaction, sug-

93 gesting that participants shifted their movement in

94 response to their partners movement.

95 Finally, in unilateral entrainment, one individual within

96 a dyad (the synchronizer) unilaterally adjusts his or her

97 movements to entrain to the movements of the other

98 individual (the referent) within the dyad an individual

99 who moves periodically but does not adjust his or her

100 movements in reciprocation to promote synchrony.

101 Previous work has focused on interpersonal synchrony

102 achieved through orchestration or reciprocal entrainment

103 (e.g.,Delaherche et al., 2012; Repp and Su, 2013). Our

104 focus here is on the social effects and neural correlates

105 of unilateral entrainment. In a pilot study (Study I) and

106 Study II, we sought to establish the extent to which a

107 referent, who is subjected to a partner who behaves in a

108 relatively synchronous or asynchronous fashion,

109 perceives the former partners movements to be more

110 synchronous than the latter partners movements, and

111feels greater affiliation toward the former than latter part-

112ner. In other words, we sought to ascertain whether (or

113not) unilateral synchrony promoted a sense of liking and

114rapport, thereby extending previous investigations

115centered on assessing the relative movement of those

116with a heightened motivation to socially connect with a tar-

117get (e.g.,Miles et al., 2010, 2011). In Study III, we inves-118tigated the neural correlates of perceived synchrony in the

119referent.

120Social functions of synchrony

121Over the past decades, two main bodies of literature have

122developed to better understand interpersonal synchrony.

123The literature on sensorimotor synchronization (SMS)

124focuses on an action that leads to synchrony by means

125of temporary coordination with a predictable external

126event (the referent). Among the findings in this field are

127that error correction is required to maintain SMS (see

128review by Repp, 2005), and stability is greater for syn-

129chronous than asynchronous inter-limb (e.g., arm or leg)

130movements within an individual (e.g., Yamanishi et al.,

1311980; Kelso, 1984) and between individuals (e.g.,

132Schmidt et al., 1990; Richardson et al., 2005), with the

133result being an increased likelihood of entrainment (e.g.,

134Engstro m et al., 1996; Schmidt and OBrien, 1997).

135A second literature focuses on the socialfunctions of

136interpersonal synchrony. Hatfield et al. (1994) hypothe-

137sized that interpersonal synchrony enhances the

138moment-by-moment tracking of other peoples feelings

139(even when individuals are not explicitly attending to this

140information Q6), thereby promoting emotional alignment

141between interacting individuals. Relatedly, as described

142above,McNeill (1995) posited that synchrony contributes143to group solidarity. Since the 1990s, a large number of

144studies have reinforced these hypotheses and showed

145that performing actions that are similar to, and coordi-

146nated with, those of an interacting partner enhances feel-

147ings of connectedness, affiliation, interpersonal rapport,

148and a blurring of self-other boundaries (Bernieri, 1988;

149Tickle-Degnen and Rosenthal, 1990; Bernieri et al.,

1501994; Cappella, 1997; Lakin and Chartrand, 2003; Hove

151and Risen, 2009; Miles et al., 2010, 2011; Paladino

152et al., 2010; Vacharkulksemsuk and Fredrickson, 2012),

153liking (e.g., Hove and Risen, 2009; Miles et al., 2009), per-

154ceived similarity and compassion (Valdesolo and

155DeSteno, 2011), joint action (Valdesolo et al., 2010),

156cooperation and enhanced altruistic behavior

157(Wiltermuth and Heath, 2009; Valdesolo and DeSteno,

1582011), better negotiation outcomes (Maddux et al.,

1592008), emotional empathy (Chartrand and Bargh, 1999;

160Sonnby-Borgstro m, 2002; Marzoli et al., 2011), person

161memory (Macrae et al., 2008; Miles et al., 2010), group

162cohesion (McNeill, 1995), and prosocial behavior (van

163Baaren et al., 2003a, 2004; Marsh et al., 2009;

164Valdesolo and DeSteno, 2011; Mu ller et al., 2012). In

165sum, interpersonal synchrony is a foundation for effective

166social interaction and enhanced sociality (Miles et al.,

1672009; Delaherche et al., 2012; Lumsden et al., 2012). Lit-

168tle is known, however, about the social consequences of

169synchrony by unilateral entrainment.

2 S. Cacioppo et al. / Neuroscience xxx (2014) xxxxxx


5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-http://-/?-http://-/?-


3/17

170 The neural correlates of interpersonal synchrony

171 There is an extensive body of research on the underlying

172 brain mechanisms for SMS with an external stimulus.

173 Briefly, brain areas known to be involved in movement

174 timing, temporal prediction, error correction and internal

175 modeling of sensorimotor dynamics (such as the basal176 ganglia, cerebellum, and prefrontal regions; e.g., Strick

177 et al., 1993; Rao and Georgeff, 1997; Salman, 2002;

178 Krause et al., 2010; Bijsterbosch et al., 2011; cf. also

179 reviews by Rao and Georgeff, 1997; Lewis et al., 2004;

180 Repp, 2005) are activated during synchrony. This brain

181 network highlights the importance of temporary coordina-

182 tion with a predictable external event (the referent) during

183 synchrony. For instance,Lewis et al. (2004) investigated

184 the neural correlates of rhythmic movement complexity

185 to investigate error monitoring and correction. Among

186 the brain regions that varied with movement complexity

187 during SMS (but not during similar self-paced move-

188 ments) were the premotor cortex (PMC), supplementary

189 motor cortex (SMA), and right dorsolateral prefrontal cor-190 tex (DLPFC) (cf.Rao and Georgeff, 1997).

191 The literature on the neural correlates of the

192 perception and social consequences of interpersonal

193 synchrony is smaller (Tognoli et al., 2007; Kelso et al.,

194 2009; Konvalinka et al., 2010; Fairhurst et al., 2012Q7 ). To

195 date, the social consequences of behavioral interpersonal

196 synchrony have been mostly documented following both

197 the mimicry of discrete bodily movements (e.g., foot shak-

198 ing, face touching;van Baaren et al., 2003a) and the syn-

199 chronization of more continuous sequences of action

200 (e.g., postural movements, facial expressions, gestures;

201 Bernieri, 1988; Cappella, 1997; for review cf. Miles

202

et al., 2009). For instance, a meta-analysis of studies of203 a related social motor actionimitationindicates activa-

204 tion of parietal and frontal regions including the superior

205 parietal lobule, inferior parietal lobule (IPL), and dorsal

206 PMC (Molenberghs et al., 2009). Guionnet et al. (2011)

207 extended this work in an functional magnetic resonance

208 imaging (fMRI) study of participants as they imitated or

209 were imitated by another person. Results revealed activa-

210 tion in the primary sensorimotor cortex, premotor and

211 supplementary motor areas, left inferior frontal gyrus, left

212 IPL, and left insula, whether imitating or being imitated. In

213 addition, activation was found in the dorsal anterior cingu-

214 late (dACC), pre-supplementary motor area (pre-SMA),

215 and a rostral part of the DLPFC in all conditions except

216 during instructed imitation. The contrast of imitating or217 being imitated revealed that being imitated by another

218 person led to greater activation in the dACC, pre-SMA,

219 and DLPFC, and the dorsal region of the left anterior insu-

220 lar cortex, whereas imitating led to greater activation in

221 the visual cortex, medial frontal cortex, posterior cingulate

222 gyrus, precuneus, bilateral IPL, para-hippocampus, and

223 hippocampus than being imitated.

224 Many of these regions constitute the default mode

225 network (DMN;Raichle et al., 2001; Fox et al., 2005), a

226 network that is more active during self-referential, social,

227 and affective processing (Raichle and Snyder, 2007;

228 van Overwalle and Baetens, 2009). Fairhurst et al.

229 (2012) performed an fMRI study of SMS with a virtual

230 partner using a finger-tapping paradigm in which the

231virtual partner varied in adaptivity, which also corre-

232sponded to differing degrees of coupling between the vir-

233tual partner and participant. Participants were instructed

234to synchronize with the virtual partner while also maintain-

235ing the initial tempo, thereby establishing the goals of

236maintaining the periodicity of the finger tapping and mini-

237mizing the phase differences in the finger-tapping task.238Objective synchrony was operationalized in terms of

239phase relations, whereas the feeling of being synchro-

240nized was operationalized as (lower) perceived task diffi-

241culty. Regression analyses identified different networks

242whether the participants were objectively in synchrony

243with the virtual partner (positive correlation with increased

244midline activation of structures including the ventromedial

245prefrontal cortex, vmPFC; hippocampus, supplementary

246motor area, SMA; primary somatosensory cortex, S1

247extending into primary motor cortex, M1; posterior cingu-

248late; and precuneus) or subjective perception of syn-

249chrony (i.e., reduced task difficulty was correlated with

250

greater activation of the right IFG, right anterior insula,251posterior dorsomedial prefrontal cortex (dmPFC), bilateral

252ventrolateral prefrontal cortex, superior frontal gyrus, and

253inferior parietal activity in the region of the temporo-parie-

254tal junction for perceived synchronization difficulty, and

255SMA, S1/M1, vmPFC and hippocampus; Fairhurst et al.,

2562013).

257Although this body of research is on the perception

258and social consequences of interpersonal synchrony

259(e.g.,Tognoli et al., 2007; Kelso et al., 2009; Konvalinka

260et al., 2010; Fairhurst et al., 2012), these studies have

261focused primarily on the neural correlates of ones syn-

262chronizing their behavior with a referent. Little is known

263about the neural bases of interpersonal synchrony from

264the perspective of the referent. Thus, in the present study,265we used fMRI to investigate how regional brain activity

266was modulated by differences in synchronous stimuli dur-

267ing a tapping-based interactive task compared to asyn-

268chronous stimuli with a synchronizer. Moreover, little is

269known about the neural regions that might be correlated

270with subjective perceptions of synchrony and correspond-

271ing feelings of affiliation between a referent and a syn-

272chronizer. Therefore, we also ran correlational analyses

273to explore this relationship (see Method section for

274details). To the best of our knowledge, this is the first

275empirical investigation of the neural correlates of the par-

276ticipants perception of interpersonal synchrony and their

277feelings of affiliation with a virtual co-acting partner when

278the participant is the referent (rather than the

279synchronizer).

280GENERAL EXPERIMENTAL PROCEDURES

281Participants

282All participants were native English speakers with normal

283or corrected-to-normal vision, and were not taking

284antidepressant medication. As ascertained by an

285anamnesis, none of the participants reported prior or

286current neurological or psychiatric disorders (e.g.,

287traumatic brain injury with loss of consciousness,

288epilepsy, neurological impairment or degenerative

289neurological illness). All participants provided written

S. Cacioppo et al. / Neuroscience xxx (2014) xxxxxx 3


5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-http://-/?-http://-/?-http://-/?-


4/17

290 informed consent to participate in the experiment, which

291 was approved by the University of Chicago Health

292 Sciences Institutional Review Board. All participants

293 received monetary compensation for their participation.

294 General experimental task

295 The experimental task was presented to participants as a

296 computer-mediated communication task that involved

297 simple back-and-forth keyboard tapping between

298 members of a dyad. Specifically, the task was described

299 as an abstract simulation of cell-phone texting, where a

300 beat (i.e., a single tap on the computer keyboard)

301 replaced actual textactions described as bexting,

302 short for beat-based texting (Fig. 1).

303 Throughout the session, the message board at the top

304 of the screen displayed various information and

305 instructions about the task. Participants were informed

306 that the circle labeled I was their own avatar, which

307 would immediately pulse each time they would send a

308 beat (i.e., pressed the keyboard once). The pulse was

309 visually represented by a short animation of the circle

310 transforming into a square and then back into the circle.

311 Participants were also told that the central server would

312 pair them up with randomly a selected fellow participant

313 in the room, one of whom would be represented by an

314 avatar labeled A or B. It was emphasized to the

315 participants that the specific avatar (i.e., A or B)

316 chosen to represent their partner on the screen was

317 randomly determined after the partner was selected,

318 thus bearing no relationship to the partners true identity.

319 Once the dyad was formed, participants avatar and the

320 partners avatar entered the bexting zone represented

321 by the rectangular box surrounding the two avatars322 (Fig. 1).

323 The participant was told that their task was simply to

324 generate a series of beats at a designated frequency

325 (e.g., 1 beat/s), regardless of their partners beat

326 frequency (i.e., unencumbered by any need to

327 coordinate their beats with their partners beats).

328 Participants were also informed that the task of their

329 partners was to respond to each one of their beats with

330 another beatwith no time constraint to respond except

331 that they had to send a beat back to each beat prior the

332 occurrence of the referents n + 1st beat. Although the

333 participant served as the referent, no mention was

334 made of this and no mention was made of synchrony.

335 The two dyadic members bextedQ8 with each other for an336 extended period of time, called a bexting round

337 (described below), which consisted of multiple equal-

338 length trials separated by short breaks.

339 At the end of a bexting round, the participants reported

340 their impression of their partners by answering a short

341 questionnaire displayed on the message board. After

342 completing the questionnaire, participants were led to

343 believe that the server would form a new pairing

344 between themselves and another randomly selected

345 fellow participant and that a new bexting round would

346 then ensue. This made it possible to manipulate partner

347 synchrony using a within-subjects design, which is

348 especially important if the paradigm is also to be used349 to investigate the neural correlates of perceived

350interpersonal synchrony. Due to the presence of at least

351three other fellow participants, the use of cubicles,

352rubber keyboards to ensure key presses could not be

353heard, and the supposedly random pairing scheme

354implemented by the central server, it was impossible for

355the participants to map their ostensible partners to any

356particular individual in the room. As a result, the only357reliable information about a given partner accessible to

358the participants was the timings of that partners beat

359series. Objective synchronicity by definition is contingent

360on the alignment of timing per se, so this feature of the

361paradigm allowed us to examine whether timing

362information was sufficient to influence perceived

363synchrony and social affiliation.

364General manipulation of unilateral entrainment

365The participant and his/her partner correspond,

366respectively, to the referent and synchronizer involved in

367unilateral entrainment. Unbeknown to the participants,

368the partners beat series were generated by a

369computer program, which made it possible to

370experimentally manipulate the degree to which the

371partners beats were entrained to the referents beats.

372More precisely, the partners beat latency (i.e., the

373interval between the referents beat and the partners

374beat) was sampled from a uniform distribution with pre-

375determined mean and range (described below).

376Because prior research has manipulated synchrony

377using latency ranges varying between 0 and 90, beat

378latencies in the present research were manipulated

379within the same range. By manipulating the means and

380range of the distribution of partners response latency,

381different levels of synchrony could be produced. This382feature ensured that the variation in synchrony was

383determined solely by the unilateral entrainment on the

384part of the ostensible synchronizer rather than through

385mutual entrainment or orchestration.

386STUDY I (PILOT STUDY)

387Because the present experimental tapping task differs

388from existing paradigms, we first conducted a pilot study

389to test whether the cover story for the tapping task was

390believable and whether the task instructions were easy

391for participants to understand.

392Participants

393Forty-seven community residents (19 women)

394participated in this pilot study. Participants ranged from

39519 to 52 years of age (M= 25.10, SD= 7.19). No

396participants were excluded from the analyses. Data

397collection started at a beginning of an academic quarter

398and stopped at the end of that academic quarter.

399Experimental procedure

400Participants were tested in groups of four in the same

401testing room. This procedure was used to ensure that

402participants did not know with whom they would be

403bexting during any given task period. Each participant404was seated in a separate cubicle, which was equipped



5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


5/17

405 with one computer. Participants were free to adjust the

406 position of their chairs to their utmost comfort level.

407 Participants were told that all the four computers in the

408 room were connected to a central server. Each bexting

409 round consisted of six 12-s trials. The asynchrony (i.e.,

410 response latency) distributions of the partners were

411 experimentally manipulated such that the interaction with

412 one partner was more synchronous than the other.

413 Specifically, the mean asynchrony and asynchrony range

414were 220 ms and 110 ms for the low-synchronous

415partner and 110 ms and 10 ms for the high-

416synchronous partner. The order in which participants

417bexted with a synchronous or asynchronous partner was

418counterbalanced across participants. The bexting

419program was coded in Adobe ActionScript 3 and ran

420through Adobe Flash Player.

421Participants instruction was the following: Using the

422spacebar, tap at a slow rate (approximately 1 beat per 2

Fig. 1. Screenshot of the computer interface of the bexting task.



5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-


6/17

423 second) [a moderate rate (approximately 1 beat per

424 second)/a fast rate (approximately 2 beats per second)].

425 In each experimental block, all three suggested tempos for

426 beat generation appeared twice (thus six trials in total),

427 with the order randomly determined. The variation of the

428 suggested tempos was to investigate generalizability.

429 At the end of each tapping experimental block,430 participants answered six items concerning the degree

431 of social affiliation they felt toward the ostensible partner

432 in that tapping experimental block. Specifically,

433 participants were to indicate on a seven-point scale

434 anchored by 1 (not at all) and 7 (very much), (1) How

435 much rapport they felt with the partner, (2) How much

436 they trusted the partner, (3) How much they liked the

437 partner, (4) How much they would like to work with

438 the partner, (5) How much they would like to confide in

439 the partner, and (6) how close they felt to the partner.

440 These six items showed high internal consistency

441 across both conditions (as > .92) and were thus

442

averaged to yield asocial affiliation score.443 Embedded among these affiliation items was a

444 perceived synchrony item, which asked participants to

445 indicate how synchronized they were with the partner on

446 the same seven-point scale (How synchronized was the

447 communication between you and Partner A?). The

448 inclusion of this measure was motivated primarily by

449 one main consideration. Although our experimental

450 manipulation objectively created two levels of synchrony,

451 it was unclear whether participants would subjectively

452 map the difference in their experiences with the two

453 partners on the dimension of synchronicity. Given the

454 apparent non-rhythmic nature of the synchronizers task,

455 the participants might have parsed the partners

456 behaviors into a series of independent local events (i.e.,457 whether the partner responded in time on a given trial)

458 instead of integrating these local events across the

459 temporal span of the tapping experimental block. Thus,

460 the participants might not perceive the synchronizer as

461 engaging in periodic movement and thereby might not

462 construe their interaction in terms of synchronicity.

463 Results

464 Participants feedback about the task instruc-

465 tion. Results from this pilot study revealed that none of

466 the participants reported being confused regarding the

467 task instruction. Furthermore, none of the participants468 suspected that their partners were actually a computer

469 program rather than two of their fellow participants.

470 Participants behavioral performance. To determine

471 whether the participants performance was influenced by

472 the experimental manipulation, their performance was

473 subjected to a 2 (Partners type: low-synchrony or high-

474 synchrony) 2 (Order) 2 (Gender) 3 (Tapping pace:

475 2/s, 1/s, .5/s) mixed analysis of variance (ANOVA).

476 Neither the main effect of synchrony manipulation nor

477 any of the interactive effects involving synchrony

478 manipulation was significant. Of all the interactive

479 effects, the one with the largest effect size was the480 interaction between synchrony manipulation and tapping

481pace (F(2,86) = 1.72, p= .02, g2partial= .04). As for the

482main effect of synchrony manipulation, we found no483evidence of our manipulation influencing tap-to-tap

484variability (F(1,46) = 0.24, p = .63, g2partial= .01).

485Participants perceived synchrony. The perceived

486synchrony scores were subjected to a 2 (Partners type:

487low-synchrony or high-synchrony) 2 (Order) 2

488(Gender) mixed ANOVA. No significant results involving

489gender, order or tapping were observed, so we collapsed

490across these factors. Results showed that participants

491rated their interaction with the high-synchronous partner

492as being more synchronized (M= 5.91, SD= 1.47)

493than their interaction with the low-synchronous (i.e.,

494asynchronous) partner (M= 5.13, SD= 1.81; F(1,46)

495= 6.45,p = .02,g2partial= .03;Table 1).

496Participants social affiliation. The social affiliation

497scores were also subjected to a 2 (Partner type: low-

498synchrony or high-synchrony) 2 (Order) 2 (Gender)499mixed ANOVA. No effects involving gender, order were

500found, so we collapsed across these factors. Results

501showed that participants felt greater social affiliation with

502the high-synchronous partner (M= 4.91, SD = 1.59)

503than the low-synchronous (asynchronous) partner

504(M= 4.54, SD = 1.67; F(1,46) = 4.32, p = .004,

50g2partial= .02;Table 2).

506Interim conclusion

507The results from this pilot study suggest that the tapping

508task is a viable paradigm for studying interpersonal

509synchrony achieved through unilateral entrainment. The

510cover story is believable and the instructions are easy to

511understand. The difference in perceived synchrony

512across the two conditions suggests that participants

513were influenced by interpersonal synchrony achieved

514through unilateral entrainment even though the

515participants played no role in the production of the

516synchrony1 and the synchrony was unrelated to their task

517performance. We nevertheless found a significant effect518on perceived synchrony and a stronger affiliative519response toward the synchronous than asynchronous

520partner. This suggests the effects of interpersonal

521synchrony are not dependent on the synchrony being522task-relevant or to the participant actually contributing to

523the observed synchrony.

524STUDY II (BEHAVIORAL STUDY)

525Participants

526Forty community residents (20 women) participated in this

527behavioral study. Participants ranged in age from 19 to 43

528years (M= 23.9,SD = 6.87) and were tested in a similar

529setting as the pilot study. No participants were excluded in

530the analyses. Data collection stopped at the end of an

531academic quarter.

1 The computer algorithm used to manipulate the degree of

synchrony ensured that the experimental manipulation of synchronywas orthogonal to the participants beat series.



5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


7/17

532 Experimental procedure

533 A similar procedure to that used in the pilot study (Study I)

534 was used in Study II. Each participant played a one-tapping535 experimental block with each of four ostensible partners.

536 Each experimental block consisted of eight 12-s trials

537 followed by the series of questions on perceived

538 synchrony and affiliation. The suggested tapping tempo

539 for the referent (i.e., the participant) was kept the same

540 throughout the experimental session at one beat per

541 second. The asynchronies of the four ostensible partners

542 were sampled respectively from four uniform distributions

543 with unique mean-range combinations obtained by

544 crossing two levels of asynchrony mean (120 ms versus

545 220 ms) with two levels of response latency ranges

546 (10 ms versus 110 ms). The order in which

547 participants bexted with the four partners was

548 manipulated using a Latin Square design, yielding 10

549 different orders. As in the pilot study, the six items

550 measuring social affiliation exhibited a high level of

551 internal consistency across all four conditions (as > .97)

552 and hence were combined.

553 Results

554 Participants behavioral performance. A 2 (Mean

555 response latency: 120 ms or 220 ms) 2 (Response

556 latency range: 10 ms or 110 ms) 2 (Gender)

557 ANOVA was performed to determine whether the558 participants (i.e., referents) responses were influenced

559 by their partners behavior. No significant differences

560 involving gender were observed, so we collapsed across

561 this factor. The ANOVA revealed no significant

562 interaction (F(1,39) = 0.002, p= .97, g2partial= 0), and563 no main effect for mean response latency564 (F(1,39) = 0.006, p= .94, g2partial= 0). The response

565 latency range manipulation, however, did affect the

566 referents tap-to-tap variability. Specifically, participants567 tap-to-tap variability was smaller when interacting with568 narrow-range partners (+10 ms) than with broad-range

569 partners (+110 ms) (Ms = 14.29 ms and 82.34 ms,570 respectively; F(1,39) = 162.1, p < .001, g2partial= .81).

571Participants perceived synchrony. The perceived

572synchrony ratings were subjected to a 2 (Mean

573response latency: 120 ms or 220 ms) 2 (Response

574latency range: 10 ms or 110 ms) 2 (Gender)

575ANOVA. No significant differences involving gender

576were observed, so we collapsed across this factor.

577Results indicated that both main effects were significant.578Partners who responded with short (120 ms) mean lags

579were rated as being more synchronized (M= 5.10) than

580partners who responded with long (220 ms) mean lags

581(M= 4.59; F(1, 39) = 3.79, p= .06, g2partial= .09;

582Table 3), and partners with narrow (+10 ms) ranges583were rated as being more synchronized than partners584with broad (+110 ms) ranges (Ms = 5.20 and 4.59,

585respectively; F(1,39) = 8.21, p= .007, g2partial= .17;586Table 3). The two-way interaction was not significant587(F(1,39) = 0.02, p = .88, g2partial= .01).

588Participants social affiliation. The social affiliation589scores were subject to a 2 (Mean response latency:

590120 ms or 220 ms) 2 (Response latency range:

59110 ms or 110 ms) 2 (Gender) ANOVA. No tests

592involving gender were significant, so we also collapsed

593across this factor. The main effects for both aspects of

594partners timing were significant: The participants

595expressed more social affiliation with the partners

596characterized by mean response latencies of 120 ms

597rather than 220 ms (Ms = 4.53 and 4.06, respectively;

598F(1,39) = 5.02,p = .03,g2partial= .10;Table 4), and with599partners characterized by 10-ms than 110-ms response600latency ranges (Ms = 4.61 and 3.98, respectively;601F(1,39) = 9.54, p= .01, g2partial= .20; Table 4). The two-602way interaction did not reach significance (F(1,39) =6030.65,p = .42, g2partial= .02).

604Interim conclusion

605Participants serving as referents in the current study

606perceived partners as more synchronous when they

607showed relatively short response latencies (i.e.,

608relatively small phase shifts) and when the variability of

609these response latencies was relatively small.

610Furthermore, and as in the pilot study, the mean

611response latency manipulation of interpersonal

612synchrony did not influence the referents tap-to-tap613variability. Although the range (variability) of response

614latencies did affect the referents tap-to-tap variability,

615the participants tapping responses were not correlated

616with the perceived synchrony (r(40) = 0.22, p= .17)

617or affiliative responses toward the partner (r(40) =

6180.25, p= .12), suggesting that the social affiliation

619effect cannot be explained by the effect of the

620experimental manipulation on the referents tapping

621behavior. These findings are generally consistent with

622prior research (Miles et al., 2009) in which observers per-

623ceived higher levels of rapport between members of a

624dyad when the mean temporal difference between their

625strides while they were walking decreased.

Table 1. Feelings of perceived synchrony with an adaptive partner

Condition Mean STD SE 95% CI

lower bound

95% CI

higher bound

Synchrony 5.91 1.47 .22 5.48 6.35

Asynchrony 5.13 1.81 .26 4.60 5.66

Table 2. Feelings of social affiliation with an adaptive partner


lower bound

95% CI

higher bound

Synchrony 4.91 1.59 .23 4.45 5.38

Asynchrony 4.54 1.67 .24 4.05 5.03



5 August 2014




8/17

626 STUDY III (NEUROIMAGING STUDY)

627 Participants

628 A total of 16 volunteers (seven women) were recruited via

629 e-mail and subsequently screened and qualified with a

630 follow-up telephone interview. All participants were right-

631 handed, ranging from 19 to 25 years old (M= 21.44,

632 SD = 1.63), and were healthy with no medical history of

633 neurological, psychiatric or psychological disorders as

634 ascertained by an anamnesis. Data from three

635 volunteers out of the 16 could not be included in the

636 analyses because the volunteers did not complete

637 entirely the task as they were too slow and took too

638 long during the instruction periods in between bexting639 rounds. The design was self-paced, and those subjects

640 appeared to have trouble with the task and did not

641 complete it before the set scanning time was complete

642 (the scanner had a finite period in which it could run for

643 each scan). The final fMRI results, thus, include 13

644 subjects.

645 Experimental procedure

646 A similar procedure to the one described in the above

647 behavioral study was used in this neuroimaging study.

648 The main difference was that stimuli were presented

649 while the participants were lying down in the scanner.

650 Visual stimuli were projected from a PC located in the651 experimenter room to a back projection screen located

652 in the scanner room. Stimuli were viewed using

653 binocular goggles mounted on the head coil

654 approximately two inches above the participants eyes.

655 The entire task consisted of five blocks. Four of the

656 experimental blocks involved the participant tapping at

657 1 Hz with an ostensible partner, and one block involved

658 the participant tapping at 1 Hz with no partner (self-

659 pacing). This latter block was included in order to

660 evaluate participants motor movements per se. Each

661 experimental block consisted of eight 12-s trials. The

662 order of the experimental conditions was varied across

663 participants using a Latin Square design. Button-press664 responses were made with the index finger on an

665fMRI-compatible response box. As in the behavioral

666studies, a tap of the button during an experimental block

667caused the I avatar to pulse momentarily from a circle

668to a square, and the partners response beat was

669depicted likewise.

670After each one of the four experimental tapping

671blocks, the participants were also asked to answer the

672series of questions on perceived synchrony and

673affiliation with their ostensible partner. As in the previous

674two behavioral studies described above, these seven

675questions included one question about perceived

676synchronization and six questions about affiliation with

677their partners. Answers to the other six questions were

678again averaged into one composite index of679interpersonal affiliation because of the high Cronbach

680alpha (>.8). Answers were navigated using the middle

681finger (moving to the left, selecting lower values) and

682ring finger (moving to the right, selecting higher values),

683and the answer selection was done using the index finger.

684Before performing the actual behavioral experimental

685task, the participants and confederates (research

686assistants who did not participate in the study)

687performed a practice block in which they were asked to

688interact with a computer (rather than with a human). In

689contrast to the actual experimental task, the computers

690response during practice lacked variability and had a

691constant inter-beat interval of 100 ms. This was

692intended to not only allow participants to familiarize

693themselves with the task, but also to enhance their

694perception that the beats they would then see during the

695experimental task were actually generated by a human

696partner. Following the practice block, the participant was

697prepared for fMRI scanning, where they performed the

698experimental task.

699Magnetic resonance imaging recordings

700Imaging was performed on a 3-T Philips Achieva Quasar

701Dual 16 Ch scanner with quadrature head coil used for

702spin excitation and signal reception. High-resolution

703volumetric T1-weighted spoiled gradient-recalled (SPGR)704images were obtained for each participant in one hundred

Table 3. Feelings of perceived synchrony with an adaptive partner


lower bound

95% CI

higher bound

Small range + small lag 5.48 1.89 .30 4.87 6.08

Large range + small lag 4.73 2.10 .33 4.05 5.40

Large range + large lag 4.25 2.21 .35 3.54 4.96Small range + large lag 4.93 2.06 .33 4.27 5.58

Table 4. Feelings of social affiliation with an adaptive partner


lower bound

95% CI

higher bound

Small range + small lag 4.92 1.76 .29 4.35 5.48

large range + small lag 4.15 1.76 .28 3.58 4.71

large range + large lag 3.81 1.89 .30 3.21 4.42

Small range + large lag 4.30 1.79 .28 3.73 4.88



5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-http://-/?-http://-/?-http://-/?-


9/17

705 eighty-one 1.0-mm sagittal slices with an 8 flip angle and a

706 24-cm field of view (FOV) for use as anatomical images.

707 Functional images were acquired using an echo-planar

708 acquisition with Z-Shimming with 32 4-mm coronal

709 slices with an inter-slice gap of 0.5 mm spanning the

710 whole brain (TR = 2 s, TE = 30 ms, flip angle = 80,

711 FOV = 22 cm, 64 64 matrix size, fat suppressed).

712 Functional image processing and analyses

713 Image pre-processing and analyses were performed

714 using Analysis of Functional NeuroImages software

715 (AFNI version AFNI_2011_12_21_1014, Medical

716 College of Wisconsin). For each subject, motion

717 detection and correction were undertaken using a six-

718 parameter, rigid-body transformation. Functional images

719 were co-registered and spatially smoothed using a

720 5-mm full width at half maximum Gaussian filter.

721 Individual-subject analyses were conducted using the

722 general linear model to generate estimates of blood

723 oxygenation level-dependent (BOLD) signal on a

724 voxelwise basis (Ward, 2002). Stimulus timing vectors

725 for each of the four experimental conditions were con-

726 volved with a gamma-variate waveform using the AFNI

727 program Waver, and the resulting model was fit voxelwise

728 to preprocessed time-series data with a linear least-

729 squares model using the AFNI program 3dDeconvolve,

730 generating a map consisting of beta coefficients (fit

731 values) at each voxel for each modeled condition

732 short-lag/synchronous variance; long-lag/synchronous

733 variance; short-lag/asynchronous variance; and long-lag/

734 asynchronous varianceas well as a baseline coefficient.

735 Output from the deconvolution analysis for each subject

736 was scaled voxelwise to percent signal change from the737 baseline, and each subjects data were spatially trans-

738 formed to Talairach and Tournoux (1988) stereotaxic

739 coordinate space and interpolated to 3-mm3 isometric

740 voxels for group analysis.

741 Our fMRI analysis aimed to identify how regional brain

742 activity was modulated by differences in synchronous

743 stimuli during a tapping-based interactive task compared

744 to asynchronous stimuli with a synchronizer. To this

745 purpose, we first identified the brain regions sensitive to

746 differences in synchrony and asynchrony using a

747 voxelwise 2 (task/response period) 2 (small/large

748 range) 2 (small/large lag) factorial ANOVA. Then, to

749 assess the relationship of these regions to corresponding

750 perceptions of synchrony and feelings of social affiliation,751 we correlated BOLD activity in each identified cluster with

752 each respective behavioral measure. The self-pacing

753 blocks were modeled in the fMRI GLM and were not

754 treated as residuals. In terms of our contrasts, they were

755 treated as regressors of non-interest. The cluster

756 threshold was atp < .01 corrected to alpha < .05.

757 Finally, because little is also known about the overall

758 network of neural regions that might be correlated with

759 subjective perceptions of synchrony and corresponding

760 feelings of social affiliation between a referent and a

761 synchronizer, we ran voxelwise correlation analyses in

762 the same respect. To further elucidate what was driving

763 voxelwise correlation effects, BOLD activity within764 voxelwise correlation regions was assessed according

765to a median split of behavioral measures. Voxelwise

766fMRI analyses were performed at the group level, the

767results of which were corrected for multiple comparisons

768using a Monte Carlo simulation to determine minimum

769cluster sizes corresponding to an alpha value of .05 for

770voxelwise threshold of p < .01 (729ll) for the ANOVA

771analysis (Nichols, 2012). An additional corrected voxel-772wise threshold of p< .025 (1080ll), was also used for

773the BOLD:behavior analysis, as p< .01 yielded no

774results for BOLD:Affiliation and limited results for

775BOLD:Synchrony.

776Difference scores (synchrony minus asynchrony) of

777BOLD signal and the corresponding behavioral data

778were also calculated for each subject, and these values

779were entered into a group-level, whole-brain voxelwise

780Pearson correlation to identify regions in which

781differential BOLD activity in response to the stimulus

782conditions was associated with the same contrast

783patterns in the behavioral responses.

784Results

785Behavioral results. The participants ratings of

786perceived synchrony and affiliative responses were

787subjected to a 2 (Mean response latency: 120 ms or

788220 ms) 2 (Response latency range: 10 ms or

789110 ms) 2 (Gender) ANOVA. A gender effect was

790observed in this sample for ratings of perceived

791synchrony (Mmale = 5.25,Mfemale = 3.50,F(1,11) = 7.08,

792p= .02, g2 = .39), and a marginal effect was observed

793for ratings of social affiliation (Mmale = 4.98, Mfemale =

7943.59, F(1,11) = 4.40, p = .06, g2 = .29). However,

795neither gender effects showed a significant interaction796with mean response latency or latency range, so we

797collapsed across the gender factor. Analyses revealed

798that participants perceived greater interpersonal

799synchrony (M10 ms = 5.19, M110 ms = 3.42, F(1,12) =

80013.45, p= .004, g2 = .40) and greater social affiliation

801(M10 ms = 4.74, M110 ms = 3.72, F(1,12) = 7.46,

802p= .02,g2 = .16) when the response latency range was

803small than large. No other tests approached statistical

804significance. No behavioral interaction effects were

805statistically significant for measures of perceived

806synchrony (Gender Mean response latency: F(1,11) =

807.011, p= .92, g2 = .0001; Gender Response latency

808range,F(1,11) = 4.04,p = .07,g2 = .05; Mean response

809latency Var: F(1,11) = 2.07, p= .18, g2 = .01;810Gender Mean response latency Response latency

811range: F(1,11) = 1.15,p = .31,g2 = .007) or for feelings

812of social affiliation. No behavioral interaction effects were

813statistically significant for measures of perceived

814synchrony (Gender Mean response latency: F(1,11)

815= 3.17,p= .10, g2 = .011; Gender Response latency

816range:F(1,11) = 2.99,p = .11,g2 = .04; Mean response

817latency Response latency range: F(1,11) = .43,

818p= .52, g2 = .002; Gender Mean response latency

819Response latency range: F(1,11) = .06, p = .81,

82g2 = .0002).

821Functional neuroimaging results. Synchrony vs.822asynchrony contrast. Based on the above results we



5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-http://-/?-


10/17

823 collapsed across the Mean Response Latency factor to

824 investigate the neural effects of variations in a partners

825 perceived synchrony with ones responding. Fig. 2 and

826 Table 5 display the main effects for Response Latency

827 Range during the experimental tapping task. The

828 synchrony minus asynchrony contrast revealed a

829 significant main effect of synchrony, which was830 characterized by a greater response in three brain

831 regions: (i) left IPL extending to the angular gyrus,

832 portions of the left (ii) parahippocampal gyrus extending

833 to the amygdala and (iii) the vMPFC and anterior

834 cingulate cortex (ACC; Table 5). No significant results

835 were found for which asynchronous stimuli elicited a

836 larger BOLD response than synchronous stimuli.

837 Correlation analyses. Correlational analyses were first

838 performed between the participants ratings and each of

839 the three areas depicted inFig. 2. The BOLD differential

840 synchrony scores (dBOLD for synchrony minus

841 asynchrony) in the vmPFC was the only region to be842 significantly correlated with the comparable difference in

843 the ratings of perceived synchrony, t(11) = 2.84;

844 p= 0.016; R = 0.65, and feelings of social affiliation,

845 t(11) = 2.44, p = 0.03; R= 0.59;Fig. 3).

846 Next, whole-brain voxelwise correlation analyses were

847 performed between the dBOLD and the corresponding

848 differences between conditions in perceived synchrony.

849 Results revealed a positive correlation in the right cere-

850 bellar tonsil, and negative correlations in the right

851 anterior prefrontal cortex/lateral prefrontal cortex (BA

852 46), left dMPFC, right lingual gyrus and right middle

853 occipital gyrus (Fig. 4A, B; Table 6). To better

854understand this effect, we calculated a median split of

855our groups based on the rating difference and then

856analyzed the percent signal change of the synchronous

857and asynchronous conditions separately for the two

858groups (seeFig. 4C).

859Similar negative correlations were observed for the

860feelings of affiliation in the right lingual gyrus, and right861IPL (Fig. 5A, B,Table 7). We again calculated a median

862split and analyzed the percent signal change of the

863synchronous and asynchronous conditions separately

864for the two groups (Fig. 5C).

865DISCUSSION

866In the present series of three studies, we first sought to

867experimentally investigate an individuals social

868perceptions ofa partner who responds in a more or less

869synchronous fashion in a unilateral entrainment

870paradigm. Behavioral results across all three studies871revealed that synchrony by the partner enhanced a

872participants ratings of perceived interpersonal

873synchrony of and social affiliation with the partner.

874Specifically, the participants felt greater synchrony

875toward a synchronous partner than with an

876asynchronous partner. These results indicate that

877neither the perception of interpersonal synchrony nor

878the affiliative consequences of synchrony are contingent

879on an individuals behavioral intentions or explicit goal to

880synchronize. In all three studies, referent participants

881felt more social affiliation with partners who responded

882synchronously rather than asynchronously, even though

Fig. 2. BOLD responses obtained for synchrony compared to asynchrony. (A) Synchrony > asynchrony is shown in yellow on lateral views of the

fiducial left side of the brain (A). Brain activities were mapped on the AFNI Colin brain using Caret 5.65 software (Van Essen, 2005). (B). Plots of

percent (%) signal change were extracted for the three significant regions (IPL, left; parahippocampal region, center; and vmPFC, right) between

synchrony (orange) and asynchrony (blue). All clusters were significant at p < .01 corrected to alpha < .05. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)



5 August 2014




11/17

883 all partners (actually, a programed series of responses)

884 performed the assigned experimental task equally well.885 The current findings suggest that interpersonal

886 synchrony achieved through unilateral entrainment may

887 produce the same array of social consequences as has

888 been found previously in studies using orchestration or

889 reciprocal synchrony paradigms (cf. Bernieri, 1988;

890 Tickle-Degnen and Rosenthal, 1990; Hatfield et al.,

891 1994; Bernieri et al., 1994; Cappella, 1997; Lakin and

892 Chartrand, 2003; Hove and Risen, 2009; Miles et al.,

893 2009, 2010, 2011; Paladino et al., 2010;

894 Vacharkulksemsuk and Fredrickson, 2012) or in studies

895 using mimicry (e.g., van Baaren et al., 2003a, 2009;

896 Maddux et al., 2008; Stel et al., 2010; Mu ller et al.,

897 2012). One possible interpretation for such social conse-

898 quences may rely on the automatic (or nonconscious)899 human tendency to act in synchrony with others even

900 when they are not aware of it. Like mimicry, interpersonal

901 synchrony increases the social connection felt between

902 individuals through an automatic process of mimicry

903 that is described in the literature as a by-product in inter-

904 action (e.g., Chartrand and Bargh, 1999; van Baaren

905 et al., 2009). This process is in line with a large body of

906 evidence suggesting that the affiliative effects are not

907 dependent on an individuals awareness of the interper-

908 sonal synchrony (e.g., see review by Hatfield et al.,

909 1994; Cacioppo and Cacioppo, 2012). Another possible

910 interpretation, which is related to the latter, is an interac-

911

tion between feelings of liking and the activation of shared912 motor representations between the self and the other in

913several tasks, as it has been reported in interpersonal

914somatic mimicry (Sonnby-Borgstro m, 2002; Marzoli915et al., 2011). Although interpersonal synchrony refers to

916the coordination of movement that occurs between indi-

917viduals and interpersonal mimicry refers to the similarity

918in form of the actions between individuals, they both fea-

919ture similarities in the temporal alignment of the actions

920and in their social consequences (Semin and Cacioppo,

9212009; Cacioppo and Cacioppo, 2012). As illustrated by

922the social cognition model (from Semin and Cacioppo,

9232009), synchronization and mimicry are time-locked to

924the observed stimulus. Like mimicry, interpersonal syn-

925chrony also increases the social connection felt between

926individuals.

927Our fMRI results extend these behavioral results by

928revealing the recruitment of brain areas involved in929social cognition, embodied cognition, self-other

930information processing, and action observation as

931correlates of interpersonal synchrony (vs. asynchrony).

932More precisely, the synchrony minus asynchrony

933contrast revealed greater response in three brain

934regions: (i) left IPL (BA 40) extending to the angular

935gyrus, (ii) portions of the left parahippocampal gyrus

936(BA 38) extending to the amygdala; and (iii) the ventro-

937medial prefrontal cortex (vmPFC; BA 32) extending to

938the ACC. No significant results were found for which

939asynchronous stimuli elicited a larger BOLD response

940than synchronous stimuli.

941

The recruitment of BA 40 is consistent with previous942studies showing the recruitment of this brain region

Table 5. Variance range main effect results of the whole-brain factorial ANOVA. All clusters were significant at p < .01 corrected to alpha < .05.

Regions are indexed with MNI coordinates; Brodmann areas are indicated for appropriately located clusters

Vol (ll) x y z t

Left Inferior parietal lobule, IPL (BA 40) 2565 48 57 38 3.87

Supramarginal gyrus

Angular gyrusLeft Parahippocampal gyrus (BA 38) 945 28 3 19 3.52

Amygdala

Left vmPFC/anterior cingulate (BA 32) 918 3 38 2 3.45

Fig. 3. Correlations between neural activity and behavioral measures in the ventromedial prefrontal cortex (vmPFC). The BOLD effect for

synchrony found in vmPFC (see Fig. 2) significantly correlated with measures of perceived synchrony and feelings of affiliation. The ordinate

indicates behavioral difference scores for synchronousasynchronous trials; the abscissa indicates difference scores in BOLD activity (dBOLD).

Participants reporting higher perception of synchrony and feelings of affiliation for synchronous items also showed greater corresponding vmPFC

activity. Results were obtained with a voxelwise cluster threshold ofp < .025, corrected for multiple comparisons to alpha < .05.



5 August 2014




12/17

943 while participants integrate visuo-motor information

944 during observation and evaluation of actions (Grafton

945 et al., 1996; Rizzolatti and Craighero, 2004; Desmurget

946 et al., 2009; Grafton, 2009; Ortigue et al., 2009, 2010)

947 and perception of elementary mechanical causality

948 events (Blakemore et al., 2001). This action observation

949 brain system is also known to sustain embodied cognitive

950 mechanisms, meta-representation of the bodily self,

951 detection of movements of others, self-other expansion,

952monitoring of others intentions, perspective taking, and

953perception of a synchrony between visual and propriocep-

954tive feedbacks, as well as observed and imagined actions

955(e.g., IPL; Grafton et al., 1996; Shimada et al., 2005;

956Rizzolatti and Sinigaglia, 2007; Ortigue et al., 2009; van

957Overwalle and Baetens, 2009; Fairhurst et al., 2013).

958The recruitment of this brain network is in line with theo-

959ries of embodied cognition and simulation, which suggest

960that people may understand actions of others, without any

Fig. 4. (A) Results of voxelwise correlation analyses between the BOLD differential synchrony scores (dBOLD: synchrony minus asynchrony) andreported feelings of perceived synchrony projected onto a slice from the MNI atlas (left, z= 42) and mapped on the Caret AFNI Colin brain right

hemisphere, lateral view (center) and medial view (right). (B) Scatter plots for each respective cluster, from left to right: cerebellar tonsil, right middle

occipital gyrus (BA 19), right lateral prefrontal cortex (BA 46), dorsomedial prefrontal cortex (BA 9), and right lingual gyrus (BA 18/19). (C) Median

split plots indicating each clusters BOLD activity in each condition for the subsamples above and below the behavioral median. Results wereobtained with a voxelwise cluster threshold ofp < .025, corrected for multiple comparisons to alpha < .05.

Table 6. Clusters resulting from the voxelwise analysis correlating BOLD signal during task period and behavioral ratings of perceived synchrony.

Results were obtained with a voxelwise cluster threshold ofp< .025, corrected for multiple comparisons to alpha < .05. Regions are indexed with MNI

coordinates; Brodmann areas are indicated for appropriately located clusters

Vol (ll) x y z r

Positive correlation

Right Cerebellar tonsil 1269 27 55 49 .64

Negative correlations

Right Anterior prefrontal cortex (BA 10) 2808 38 46 10 ..69

Lateral prefrontal cortex (BA 46)

Dorsomedial prefrontal cortex (BA 9) 1566 1 53 33 ..60

Right Lingual gyrus (BA 18/19) 1107 10 89 14 ..66

Right Middle occipital gyrus (BA 19) 1080 29 89 9 ..68



5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051


13/17

961 inferential reasoning, through a direct matching process

962 that occurs via an automatic mapping between observed

963 and performed actions, and via the reactivation of the

964 bodily states that were originally active during past self-

965 related experiences (Grafton, 2009; Niedenthal, 2007;

966 Niedenthal, et al., 2005; Rizzolatti and Craighero, 2004;

967 Rizzolatti et al., 2001). Although embodied mechanisms

968 are not a pre-requisite to act, connect or understand oth-

969 ers, embodied behaviors offer new ways to investigate

970 social perception, cognition, and behavior (e.g., Semin

971 and Smith, 2002; Semin and Cacioppo, 2009; Cacioppo

972 and Cacioppo, 2012). In line with Aron and Arons

973 (1986) self-expansion model which posits that others

974 toward whom one feels a close social bond can be

975incorporated into the representation of ones self, and

976the relational model of communal sharing and cognitive

977interdependence (see Fiske, 2004; Smith, 2007;

978IJzerman and Semin, 2010; Cacioppo and Cacioppo,

9792012).

980Differences in activation were also found in the

981parahippocampal areaa region shown previously to be

982involved in temporal discrimination and interval

983comparison (Harrington et al., 2002), and learning of

984adaptive events (Fairhurst et al., 2013; Grossberg,

9852013). These findings are in line with adaptive resonance

986theory, a cognitive and neural theory of how the brain

987automatically learns to identify, categorize, and predict

988events in a changing world (Grossberg, 2013).

Fig. 5. Results of correlation analyses between the BOLD differential synchrony scores (dBOLD between synchrony minus asynchrony) andreported feelings of affiliation. (A) Correlation clusters mapped onto the Caret AFNI Colin brain right hemisphere, lateral view (left) and medial view

(right). (B) Scatter plots for each respective cluster from left to right: inferior parietal/supramarginal gyrus (BA 40), lingual gyrus (BA 19). (C) Median

split plots indicating each clusters BOLD activity in each condition for the subsamples above and below the behavioral median. Results were

obtained with a voxelwise cluster threshold ofp < .025, corrected for multiple comparisons to alpha < .05.

Table 7. Clusters resulting from the voxelwise analysis correlating BOLD signal during task period and behavioral ratings of feelings of affiliation.

Results were obtained with a voxelwise cluster threshold ofp< .025, corrected for multiple comparisons to alpha < .05. Regions are indexed with MNI

coordinates; Brodmann areas are indicated for appropriately located clusters

Vol (ll) x y z r

Negative correlations (no positive correlations found)

Right Lingual gyrus (BA 19) 1431 9 87 1.5 .68

Right Inferior parietal lobule/supramarginal gyrus (BA 40) 1323 59 49 34 .6



5 August 2014


http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://dx.doi.org/10.1016/j.neuroscience.2014.07.051http://-/?-


14/17

989 Finally, several investigators have found the ventral

990 part of the medial PFC is relatively activated when

991 processing information about the self or similar others,

992 whereas the dorsal part of the medial PFC is relatively

993 activated when processing information about others

994 (Mitchell et al., 2005; Amodio and Frith, 2006; Keysers

995 and Gazzola, 2007; Epley et al., 2009). Consistent with996 synchrony increasing the perception of similarity,

997 Fairhurst et al. (2013)found greater activity in the vmPFC

998 region when participants were in relative synchrony with a

999 virtual partner. We also found greater activity in the

000 vmPFC in the synchronous than asynchronous condition,

001 and correlational analyses further revealed that the

002 greater the difference in the BOLD signal in the vmPFC

003 between the synchronous and asynchronous conditions,

004 the greater the corresponding difference in the ratings of

005 perceived synchrony and affiliation.

006 Correlational analyses involvingthe dmPFC showed the

007 opposite pattern, as might be expected if interpersonal

008

synchrony increases self-other overlap or egocentric009 information processing about the partner. To further

010 investigate this result, a median split was performed to

011 create two groups of participants, those who rated the

012 synchronous partner as much more synchronous than

013 they rated the asynchronous partner, and those who rated

014 the synchronous and asynchronous partner relatively

015 similarly on perceived synchrony. Analyses of the dmPFC

016 showed the lowest levels of activation when the

017 participants who most distinguished between the

018 conditions were performing with a synchronous partner

019 and the highest levels of activation when the participants

020 who distinguished most between the conditions were

021 performing with an asynchronous partner. This pattern

022 was reversed and weaker in participants who perceived023 relatively little difference in synchrony between their

024 synchronous and asynchronous partners.

025 In sum, the analyses of the mPFC regions suggest

026 that the participants, who most distinguished between

027 the synchronous and asynchronous partners, thought

028 about the synchronous partner as being more similar to

029 themselves and thought about the asynchronous partner

030 as being more dissimilar to themselves, than the

031 participants who less distinguished between the

032 synchronous and asynchronous partners. When a

033 synchronous, relative to an asynchronous, partner is

034 assimilated to the self, it is the asynchronous partner

035 who requires the most attention and mentalizing to

036 understand and predict. In contrast, for participants who

037 show relatively little difference in the perceived

038 synchrony of the synchronous and the asynchronous

039 partners (and who show little difference in the activation

040 of the vmPFC region; seeFig. 3), it is the (synchronous)

041 partner whose temporal behavior is reflective of the

042 participants behavior but is not rated as being

043 synchronous who may evoke greater attention and

044 mentalizing to understand and predict. Consistent with

045 this reasoning, the correlational analyses between the

046 BOLD differential synchrony scores (dBOLD: synchrony

047 minus asynchrony) and reported feelings of perceived

048 synchrony revealed negative correlations for the right

049 lateral prefrontal cortex (BA 46), the right lingual gyrus

1050(BA 18/19), and the right middle occipital gyrus (BA 19;

1051see Fig. 4). The former is involved in control-related

1052processes (Hare et al., 2009), the lingual gyrus has been

1053involved in third-person perspective-taking (Jackson

1054et al., 2006), and the middle occipital gyrus has been

1055involved in visual attention and discrimination (Tu et al.,

10562013). Exploratory analyses based on median splits1057further indicated the lowest levels of activation when the

1058participants whose ratings of perceived synchrony most

1059distinguished between the conditions were performing

1060with a synchronous partner and the highest levels of

1061activation when these participants were performing with

1062an asynchronous partner, whereas this pattern was

1063reversed in participants who reported relatively little differ-

1064ence in perceived synchrony between their synchronous

1065and asynchronous partners. In short, for participants

1066who perceive large differences between their synchro-

1067nous and asynchronous partners and show evidence of

1068relative vmPFC activation and self-other overlap with

1069

the synchronous partner, it is the asynchronous partner1070who activates brain regions involved in attention, visual

1071discrimination, and cognitive control, whereas for partici-

1072pants who see relatively little difference between these

1073partners in terms of perceived synchrony and show little

1074difference in vmPFC activation and little self-other overlap

1075with the synchronous partner, it is the synchronous part-

1076ner who activates these regions more than the asynchro-

1077nous partner.

1078For the participants who show relatively large

1079differences in perceived synchrony across conditions (and

1080relatively large differences in vmPFC activity), the assimi-

1081lation of the synchronous (in contrast to the asyn-

1082chronous) partner to the self should result in the

1083application of an abstract trait representation of the self to1084the synchronous partner, thereby diminishing the need for

1085continued attention and mentalizing. For the participants

1086who show relatively little difference in perceived

1087synchrony across conditions (and relatively small

1088differences in vmPFC activity), both the synchronous and

1089the asynchronous partner may be regarded as dissimilar

1090others; as such, the temporal aspects of the

1091asynchronous partners behavior would be congruent with

1092the abstract trait inference that this partner is dissimilar

1093(e.g., outgroup homogeneity) and may therefore elicit little

1094additional attention or mentalizing, whereas the temporal

1095aspects of the synchronous partners behavior would be

1096more reminiscent of the self and therefore may require

1097additional processing. Although speculative, the

1098correlational analyses revealed a positive correlation in

1099the right cerebellar tonsil, a region involved in trait

1100abstraction particularly based on others nonverbal

1101behavior (van Overwalle et al., 2014). The median split

1102analyses of activation in the cerebellar tonsil region were

1103entirely consistent with high-level abstractions being

1104formed (and attention, cognitive control, and mentalizing

1105being truncated) for synchronous partners in the former

1106group of participants and for asynchronous partners in the

1107latter group of participants.

1108Finally, whole-brain correlational analyses based on

1109differences in reported feelings of affiliation for

1110synchronous versus asynchronous partners, two regions



5 August 2014




15/17

111 emerged: the right lingual gyrus (BA 19) and in the inferior

112 parietal/supramarginal gyrus (BA 40; seeFig. 5). As noted

113 above, the right lingual gyrus is involved in third-person

114 perspective taking, and the inferior parietal/supramarginal

115 gyrus is involved in sensorimotor mirroring. These results

116 suggest that for participants who perceive the

117 synchronous partner as relatively more likable than the118 asynchronous partner, regions associated with third-

119 person perspective-taking and mirroring are more active

120 when the partners behavior is asynchronous than

121 synchronous. In contrast, for participants who perceive

122 the synchronous and asynchronous partners as being

123 more equivalent in likability, these regions are more active

124 when the partners behavior is synchronous rather than

125 asynchronous.

126 Limitations of the current study include the exploratory

127 nature of the correlational analyses and the relatively

128 small sample size of the fMRI study in contrast to the

129 behavioral studies. Among the strengths of the current

130

paradigm is the experimental control that it affords. For131 instance, rather than relying on natural variations in

132 synchrony between two participants, the current

133 paradigm permits the temporal parameters used to

134 experimentally manipulate interpersonal synchrony to be

135 standardized and precisely controlled using computer

136 programs. Second, the task does not require face-to-

137 face interactions, so characteristics of the ostensible

138 partner (e.g., age, gender, attractiveness, group identity)

139 that may prove to be moderator variables can be

140 experimentally controlled. Third, participants can be an

141 actor (e.g., trials on which participants bext with a

142 partner) or an observer (e.g., trials on which they watch

143 two partners bext), making it possible to examine the

144 observational effects of interpersonal synchrony. Finally,145 the task involves minimal movement (finger tapping) so

146 that the bexting paradigm can be used in neuroimaging

147 studies.

148 CONFLICT OF INTEREST

149 The authors declare no competing financial interests.

150 UNCITED REFERENCES

151 Brown and Bru ne (2012), van Baaren et al. (2003b) and

152 Yun et al. (2012).Q9

153 AcknowledgmentsGM was supported by a grant from the

Q10154 Swiss National Science Foundation (FNS_PP00_1_128599/1 to

155 SC) andQ11 by the Center for Cognitive and Social Neuroscience

156 (CCSN). The authors thank John S. Irick for this technical

157 assistance.

158 REFERENCES

159 Amodio DM, Frith CD (2006) Meeting of minds: the medial frontal

160 cortex and social cognition. Nat Rev Neurosci 7:268277.

161 Bernieri FJ (1988) Coordinated movement and rapport in teacher

162 student interactions. J Nonverbal Behav 12(2):120138.

163 Bernieri FJ, Davis JM, Rosenthal R, Raymond Knee C (1994)

164 Interactional synchrony and rapport: measuring synchrony in

1165displays devoid of sound and facial affect. Pers Soc Psychol Bull

116620(3):303311. http://dx.doi.org/10.1177/0146167294203008.

1167Bijsterbosch JD, Lee KH, Hunter MD, Tsoi DT, Lankappa S,

1168Wilkinson ID, Barker AT, Woodruff PW (2011) The role of the

1169cerebellum in sub-and supraliminal error correction during

1170sensorimotor synchronization:

cacioppo et al ns of sync

Documents