Within-group synchronization in the prefrontal cortex associates with intergroup conflict

Abstract

Individuals immersed in groups sometimes lose their individuality, take risks they would normally avoid and approach outsiders with unprovoked hostility. In this study, we identified within-group neural synchronization in the right dorsolateral prefrontal cortex (rDLPFC) and the right temporoparietal junction (rTPJ) as a candidate mechanism underlying intergroup hostility. We organized 546 individuals into 91 three-versus-three-person intergroup competitions, induced in-group bonding or no-bonding control manipulation and measured neural activity and within-group synchronization using functional near-infrared spectroscopy. After in-group bonding (versus control), individuals gave more money to in-group members than to out-group members and contributed more money to outcompete their rivals. In-group bonding decreased rDLPFC activity and increased functional connectivity between the rDLPFC and the rTPJ. Especially during the out-group attack, in-group bonding also increased within-group synchronization in both the rDLPFC and the rTPJ, and within-group rDLPFC synchronization positively correlated with intergroup hostility. Within-group synchronized reduction in prefrontal activity might explain how in-group bonding leads to impulsive and collective hostility toward outsiders.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: Experimental procedures.
Fig. 2: Intergroup hostility as a function of in-group bonding and group role (attack versus defense).
Fig. 3: GNS as a function of in-group bonding and group role (attack versus defense).
Fig. 4: GNS in the rDLPFC (channel 8) correlates with intergroup hostility.
Fig. 5: Influence of in-group bonding on within-group averaged rDLPFC activity and group-averaged functional connectivity of rDLPFC–rTPJ.

Data availability

All behavioral data and materials have been made publicly available via the Open Science Framework and can be accessed at https://osf.io/uh3sx/. The neural data that support the findings of this study are available from the corresponding author upon reasonable request.

Code availability

The custom routines for data analysis written in MATLAB are available from the corresponding author upon reasonable request.

References

  1. 1.

    Hamilton, W. D. Geometry for the selfish herd. J. Theor. Biol. 31, 295–311 (1971).

  2. 2.

    Raafat, R. M., Chater, N. & Frith, C. Herding in humans. Trends Cogn. Sci. 13, 420–428 (2009).

  3. 3.

    Flack, A., Nagy, M., Fiedler, W., Couzin, I. D. & Wikelski, M. From local collective behavior to global migratory patterns in white storks. Science 360, 911–914 (2018).

  4. 4.

    Laughlin, P. R., Hatch, E. C., Silver, J. S. & Boh, L. Groups perform better than the best individuals on letters-to-numbers problems: effects of group size. J. Pers. Soc. Psych. 90, 644–651 (2006).

  5. 5.

    Woolley, A. W., Chabris, C. F., Pentland, A., Hashmi, N. & Malone, T. W. Evidence for a collective intelligence factor in the performance of human groups. Science 330, 686–688 (2010).

  6. 6.

    Le Bon, G. The Crowd: A Study of the Popular Mind (Macmillan, 1968).

  7. 7.

    Choi, J. K. & Bowles, S. The coevolution of parochial altruism and war. Science 318, 636–640 (2007).

  8. 8.

    De Dreu, C. K. et al. In-group defense, out-group aggression, and coordination failure in intergroup conflict. Proc. Natl Acad. Sci. USA 113, 10524–10529 (2016).

  9. 9.

    Postmes, T. & Spears, R. Deindividuation and antinormative behavior: a meta-analysis. Psych. Bull 123, 238–259 (1998).

  10. 10.

    Mead, G. H. Mind, Self, and Society (University of Chicago Press, 1934).

  11. 11.

    Selous, E. Bird Life Glimpses. (G. Allen, 1905).

  12. 12.

    Couzin, I. Collective minds. Nature 445, 715 (2009).

  13. 13.

    Dikker, S. et al. Brain-to-brain synchrony tracks real-world dynamic group interactions in the classroom. Curr. Biol. 7, 1375–1380 (2017).

  14. 14.

    Hasson, U. & Frith, C. D. Mirroring and beyond: coupled dynamics as a generalized framework for modelling social interactions. Phil. Trans. R. Soc. B 371, 1693–1702 (2016).

  15. 15.

    Echterhoff, G., Higgins, E. T. & Levine, J. M. Shared reality: experiencing commonality with others’ inner states about the world. Perspect. Psychol. Sci. 4, 496–521 (2009).

  16. 16.

    Shamay-Tsoory, S. G., Saporta, N., Marton-Alper, I. Z. & Gvirts, H. Z. Herding brains: a core neural mechanism for social alignment. Trends Cogn., Sci. 23, 174–186 (2019).

  17. 17.

    Stephens, G. J., Silbert, L. J. & Hasson, U. Speaker-listener neural coupling underlies successful communication. Proc. Natl Acad. Sci. USA. 107, 14425–14430 (2010).

  18. 18.

    Jiang, J. et al. Leader emergence through interpersonal neural synchronization. Proc. Natl Acad. Sci. USA. 112, 4274–4279 (2015).

  19. 19.

    Gould, R. V. Collective violence and group solidarity: evidence from a feuding society. Am. Soc. Rev 22, 356–380 (1999).

  20. 20.

    Glowacki, L. et al. Formation of raiding parties for intergroup violence is mediated by social network structure. Proc. Natl Acad. Sci. USA. 113, 12114–12119 (2016).

  21. 21.

    Efferson, C., Lalive, R. & Fehr, E. The coevolution of cultural groups and in-group favoritism. Science 321, 1844–1849 (2008).

  22. 22.

    De Dreu, C. K. et al. The neuropeptide oxytocin regulates parochial altruism in intergroup conflict among humans. Science 328, 1408–1411 (2010).

  23. 23.

    Zhang, H., Gross, J., De Dreu, C. K. & Ma, Y. Oxytocin promotes coordinated out-group attack during intergroup conflict in humans. Elife 8, e40698 (2019).

  24. 24.

    Ma, Y., Liu, Y., Rand, D. G., Heatherton, T. F. & Han, S. Opposing oxytocin effects on intergroup cooperative behavior in intuitive and reflective minds. Neuropsychopharmacology 40, 2379–2387 (2015).

  25. 25.

    Gläscher, J. et al. Lesion mapping of cognitive control and value-based decision making in the prefrontal cortex. Proc. Natl Acad. Sci. USA. 109, 14681–14686 (2012).

  26. 26.

    Wang, X. J. Decision making in recurrent neuronal circuits. Neuron 60, 215–234 (2008).

  27. 27.

    Knoch, D. et al. Disruption of the right prefrontal cortex by low-frequency repetitive transcranial magnetic stimulation induces risk-taking behavior. J. Neurosci. 26, 6469–6472 (2006).

  28. 28.

    Knoch, D., Pascual-Leone, A., Meyer, K., Treyer, V. & Fehr, E. Diminishing reciprocal fairness by disrupting the right prefrontal cortex. Science 314, 829–832 (2006).

  29. 29.

    Cikara, M., Jenkins, A. C., Dufour, N. & Saxe, R. Reduced self-referential neural response during intergroup competition predicts competitor harm. Neuroimage 96, 36–43 (2014).

  30. 30.

    Carter, R. M. & Huettel, S. A. A nexus model of the temporal-parietal junction. Trends Cogn. Sci. 17, 328–336 (2013).

  31. 31.

    Suzuki, S., Adachi, R., Dunne, S., Bossaerts, P. & O’Doherty, J. P. Neural mechanisms underlying human consensus decision-making. Neuron 86, 591–602 (2015).

  32. 32.

    Lin, L., Qu, Y. & Telzer, E. H. Intergroup social influence on emotion processing in the brain. Proc. Natl Acad. Sci. USA 115, 10630–10635 (2018).

  33. 33.

    Prochazkova, E. et al. Pupil mimicry promotes trust through the theory of mind network. Proc. Natl Acad. Sci. USA 115, E7265–E7274 (2018).

  34. 34.

    Baumgartner, T., Schiller, B., Rieskamp, J., Gianotti, L. R. & Knoch, D. Diminishing parochialism in intergroup conflict by disrupting the right temporo-parietal junction. Soc. Cogn. Affect. Neurosci. 9, 653–660 (2014).

  35. 35.

    Holmes, A. P., Blair, C. R., Watson, J. D. & Ford, I. Nonparametric analysis of statistic images from functional mapping experiments. J. Cereb. Blood Flow Metab. 16, 7–22 (1996).

  36. 36.

    Nichols, T. E. & Holmes, A. P. Nonparametric permutation tests for functional neuroimaging: a primer with examples. Hum. Brain Map. 15, 1–25 (2002).

  37. 37.

    Zhang, L., Sun, J., Sun, B., Luo, Q. & Gong, H. Studying hemispheric lateralization during a Stroop task through near-infrared spectroscopy-based connectivity. J. Biomed. Opt. 19, 057012 (2014).

  38. 38.

    Brewer, M. B., Manzi, J. M. & Shaw, J. S. In-group identification as a function of depersonalization, distinctiveness, and status. Psychol. Sci. 4, 88–92 (1993).

  39. 39.

    Whitehouse, H. Dying for the group: towards a general theory of extreme self-sacrifice. Behav. Brain Sci. https://doi.org/10.1017/S0140525X18000249 (2018).

  40. 40.

    Bernhard, H., Fischbacher, U. & Fehr, E. Parochial altruism in humans. Nature 442, 912–915 (2016).

  41. 41.

    Chowdhury, S. M. The attack and defense mechanisms–perspectives from behavioral economics and game theory. Behav. Brain Sci. 42, e121 (2019).

  42. 42.

    De Dreu, C. K. & Gross, J. Revisiting the form and function of conflict: neurobiological, psychological, and cultural mechanisms for attack and defense within and between groups. Behav. Brain Sci. 42, e116 (2019).

  43. 43.

    Wrangham, R. W. Two types of aggression in human evolution. Proc. Natl Acad. Sci. USA 115, 245–253 (2018).

  44. 44.

    Yamagishi, T., Takagishi, H., Fermin, A. D., Kanai, R., Li, Y. & Matsumoto, Y. Cortical thickness of the dorsolateral prefrontal cortex predicts strategic choices in economic games. Proc. Natl Acad. Sci. USA 113, 5582–5587 (2016).

  45. 45.

    Steinbeis, N., Bernhardt, B. C. & Singer, T. Impulse control and underlying functions of the left DLPFC mediate age-related and age-independent individual differences in strategic social behavior. Neuron 73, 1040–1051 (2012).

  46. 46.

    Piva, M., Zhang, X., Noah, J. A., Chang, S. W. & Hirsch, J. Distributed neural activity patterns during human-to-human competition. Front. Hum. Neurosci. 11, 571 (2017).

  47. 47.

    Yu, H., Li, J. & Zhou, X. Neural substrates of intention–consequence integration and its impact on reactive punishment in interpersonal transgression. J. Neurosci. 35, 4917–4925 (2015).

  48. 48.

    Schilbach, L. et al. Toward a second-person neuroscience. Behav. Brain Sci. 36, 393–414 (2013).

  49. 49.

    Redcay, E. & Schilbach, L. Using second-person neuroscience to elucidate the mechanisms of social interaction. Nat. Rev. Neurosci. 20, 495–505 (2019).

  50. 50.

    Faul, F., Erdfelder, E., Buchner, A. & Lang, A. G. Statistical power analyses using G* Power 3.1: tests for correlation and regression analyses. Behav. Res. Methods 41, 1149–1160 (2009).

  51. 51.

    Duan, L. et al. Cluster imaging of multi-brain networks (CIMBN): a general framework for hyperscanning and modeling a group of interacting brains. Front. Neurosci. 9, 267 (2015).

  52. 52.

    Nozawa, T., Sasaki, Y., Sakaki, K., Yokoyama, R. & Kawashima, R. Interpersonal frontopolar neural synchronization in group communication: an exploration toward fNIRS hyperscanning of natural interactions. Neuroimage 133, 484–497 (2016).

  53. 53.

    Balliet, D., Wu, J. & De Dreu, C. K. Ingroup favoritism in cooperation: a meta-analysis. Psychol. Bull. 140, 1556–1581 (2014).

  54. 54.

    Liebe, U. & Tutic, A. Status groups and altruistic behaviour in dictator games. Ration. Soc. 22, 353–380 (2010).

  55. 55.

    Gaertner, L. & Schopler, J. Perceived ingroup entitativity and intergroup bias: an interconnection of self and others. Eur. J. Soc. Psychol. 28, 963–980 (1998).

  56. 56.

    Ensari, N. & Miller, N. Decategorization and the reduction of bias in the crossed categorization paradigm. Eur. J. Soc. Psychol. 31, 193–216 (2001).

  57. 57.

    Osaka, N., Minamoto, T., Yaoi, K., Azuma, M., Shimada, Y. M. & Osaka, M. How two brains make one synchronized mind in the inferior frontal cortex: fNIRS-based hyperscanning during cooperative singing. Front. Psychol. 6, 1811–1821 (2015).

  58. 58.

    Koessler, L. et al. Automated cortical projection of EEG sensors: anatomical correlation via the international 10–10 system. Neuroimage 46, 64–72 (2009).

  59. 59.

    Lu, K., Xue, H., Nozawa, T. & Hao, N. Cooperation makes a group be more creative. Cereb. Cortex 22, 1–14 (2018).

  60. 60.

    Obrig, H. & Villringer, A. Beyond the visible—imaging the human brain with light. J. Cereb. Blood Flow Metab. 23, 1–18 (2003).

  61. 61.

    Hoshi, Y. Functional near-infrared spectroscopy: current status and future prospects. J. Biomed. Opt. 12, 062106 (2007).

  62. 62.

    Huppert, T. J., Hoge, R. D., Diamond, S. G., Franceschini, M. A. & Boas, D. A. A temporal comparison of BOLD, ASL, and NIRS hemodynamic responses to motor stimuli in adult humans. Neuroimage 29, 368–382 (2006).

  63. 63.

    Cui, X., Bray, S., Bryant, D. M., Glover, G. H. & Reiss, A. L. A quantitative comparison of NIRS and fMRI across multiple cognitive tasks. Neuroimage 54, 2808–2821 (2011).

  64. 64.

    Strangman, G., Culver, J. P., Thompson, J. H. & Boas, D.A. A quantitative comparison of simultaneous BOLD fMRI and NIRS recordings during functional brain activation. Neuroimage 17, 719–731 (2002).

  65. 65.

    Duan, L. et al. Wavelet-based method for removing global physiological noise in functional near-infrared spectroscopy. Biomed. Opt. Express 9, 3805–3820 (2018).

  66. 66.

    Jang, K. E. et al. Wavelet minimum description length detrending for near-infrared spectroscopy. J. Biomed. Opt. 14, 034004 (2009).

  67. 67.

    Molavi, B. & Dumont, G. A. Wavelet-based motion artifact removal for functional near-infrared spectroscopy. Physiol. Meas. 33, 259–270 (2012).

  68. 68.

    Liu, W. et al. Shared neural representations of syntax during online dyadic communication. Neuroimage 198, 63–72 (2019).

  69. 69.

    Cui, X., Bryant, D. M. & Reiss, A. L. NIRS-based hyperscanning reveals increased interpersonal coherence in superior frontal cortex during cooperation. Neuroimage 59, 2430–2437 (2012).

  70. 70.

    Torrence, C. & Compo, G. P. A practical guide to wavelet analysis. Bull. Am. Meteorol. Soc .79, 61–78 (1998).

  71. 71.

    Hu, Y., Hu, Y., Li, X., Pan, Y. & Cheng, X. Brain-to-brain synchronization across two persons predicts mutual prosociality. Soc. Cogn. Affect. Neurosci. 12, 1835–1844 (2017).

  72. 72.

    Xue, H., Lu, K. & Hao, N. Cooperation makes two less-creative individuals turn into a highly-creative pair. Neuroimage 172, 527–537 (2018).

  73. 73.

    Baker, J. M. et al. Sex differences in neural and behavioral signatures of cooperation revealed by fNIRS hyperscanning. Sci. Rep 6, 26492 (2016).

  74. 74.

    Liu, T., Saito, G., Lin, C. & Saito, H. Inter-brain network underlying turn-based cooperation and competition: a hyperscanning study using near-infrared spectroscopy. Sci. Rep. 7, 8684–8695 (2017).

  75. 75.

    Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B Methodol. 57, 289–300 (1995).

  76. 76.

    Singh, A. K. & Dan, I. Exploring the false discovery rate in multichannel NIRS. Neuroimage 33, 542–549 (2006).

  77. 77.

    Tak, S. & Ye, J. C. Statistical analysis of fNIRS data: a comprehensive review. Neuroimage 85, 72–91 (2014).

  78. 78.

    Hampshire, A., Chamberlain, S. R., Monti, M. M., Duncan, J. & Owen, A. M. The role of the right inferior frontal gyrus: inhibition and attentional control. Neuroimage 50, 1313–1319 (2010).

  79. 79.

    Poldrack, R. A. Region of interest analysis for fMRI. Soc. Cogn. Affect. Neurosci. 2, 67–70 (2007).

  80. 80.

    Bilek, E. et al. Information flow between interacting human brains: identification, validation, and relationship to social expertise. Proc. Natl Acad. Sci. U.S.A 112, 5207–5212 (2015).

  81. 81.

    Ding, X. P., Sai, L., Fu, G., Liu, J. & Lee, K. Neural correlates of second-order verbal deception: a functional near-infrared spectroscopy (fNIRS) study. Neuroimage 87, 505–514 (2014).

  82. 82.

    Taga, G., Asakawa, K., Maki, A., Konishi, Y. & Koizumi, H. Brain imaging in awake infants by near-infrared optical topography. Proc. Natl Acad. Sci. USA 100, 10722–10727 (2003).

  83. 83.

    Tachtsidis, I. & Scholkmann, F. False positives and false negatives in functional near-infrared spectroscopy: issues, challenges, and the way forward. Neurophotonics 3, 031405 (2016).

  84. 84.

    Ou, W. et al. Study of neurovascular coupling in humans via simultaneous magnetoencephalography and diffuse optical imaging acquisition. Neuroimage 46, 624–632 (2009).

Download references

Acknowledgements

We thank C. Hao, C. Yang and X. Zou for their assistance in data collection. This work was supported by the National Natural Science Foundation of China (Projects 31722026, 31771204 and 91632118 to Y.M.); the Fundamental Research Funds for the Central Universities (2017XTCX04 and 2018EYT04 to Y.M.); and the Spinoza Award from the Netherlands Science Foundation (NWO SPI-57-242 to C.K.W.D.D.).

Author information

Affiliations

Authors

Contributions

Y.M. and C.K.W.D.D. conceived of the project. J.Y., H.Z., C.K.W.D.D and Y.M. designed the experiments. J.Y., H.Z. and J.N. implemented the experiment and collected data. J.Y. and H.Z. pre-processed the data and performed analyses. All authors discussed results. J.Y., H.Z., C.K.W.D.D. and Y.M. wrote the paper.

Corresponding author

Correspondence to Yina Ma.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 The Effects of Winning or Losing a Contest Round on Behavioral Measures.

We examined how the contest outcome (win or lose the contest) influenced group members’ behaviors, respectively for attacker and defender groups. a, We found that, after losing the last round, attackers decreased their contribution to group fighting (F1,87 = 22.336, p= 8.74 × 10−6, η2 = 0.204) whereas defenders increased their contribution (F1,87 = 77.894, p= 1.02 × 10−13, η2 = 0.472). The effect of Outcome was not affected by the bonding conditions (utcome-by-Bonding interaction, attacker: F1,87 = 0.013, p= 0.911, η2 = 1.44 × 10−4; defender: F1,87 = 0.007, p= 0.936, η2 = 7.52 × 10−5). b, Outcome did not influence within-group decision similarity (main effect of Outcome, attacker: F1,87 = 0.074, p= 0.786, η2 = 0.001; defender: F1,87 = 2.860, p= 0.094, η2 = 0.032; Outcome-by-Bonding interaction, attacker: F1,87 = 2.055, p= 0.155, η2 = 0.023; defender: F1,87 = 1.265, p= 0.264, η2 = 0.014). Data from two sessions were excluded as attackers in these two sessions did not win any of the 24 rounds. Outcome-by-Bonding mixed-model ANOVAs (n = 89, 43 three-person groups under no-bonding control and 46 three-person groups under in-group bonding separately for attacker groups and defender groups). Data showed as Mean ± S.E. with overlaid dot plots. ***p< 0.001, n.s. not significant. Note: Given the limited and unequal number of after-win and after-lose rounds, we calculated another index to indicate group decision coordination, i.e., within-group decision similarity. For each 3-person group, we calculated the within-group decision similarity as the investment difference of each pair of the 3-person group for each round [i.e., (|x1x2| + |x2x3| + |x1x3|) for attacker group; (|y1y2| + |y2y3| + |y1y3|) for defender group].

Extended Data Fig. 2 Increased Within-group Neural Synchronization during Decision-making Phase relative to Waiting Phase.

We compared within-group neural synchronization (GNS) during the group decision-making phase with that during the waiting screen of the intergroup contest game. a,b, We averaged GNS across the channels located in the rDLPFC and rTPJ respectively, and showed higher GNS during decision-making than waiting phases (a, rDLPFC: t85 = 5.823, p= 1.00 × 10−7, Cohen’s d = 0.628, 95% CI = 0.003, 0.006, b, rTPJ: t85 = 6.578, p= 3.70 × 10−9, Cohen’s d = 0.709, 95% CI = 0.004, 0.007), indicating increased GNS during group decision-making. Two-tailed paired samples t-tests, 86 six-person groups during decision-making and during waiting phases. c, To validate the Phase effect on the GNS, we generated within-condition pseudo groups for comparison purpose by randomly grouping 3 individuals from different original real groups under the same condition as a pseudo group, and treated Group (real vs. pseudo groups) as a between-subjects factor. d,e, We conducted ANOVAs on GNS with factors of Phase (decision-making vs. waiting) and Group (real vs. pseudo groups) in 86 six-person real groups and 86 six-person pseudo groups. We observed significant Phase-by-Group interactions on GNS in rDLPFC and rTPJ (d, rDLPFC: F1,170 = 10.161, p = 0.002, η2 = 0.056; e, rTPJ: F1,170 = 13.920, p = 2.60 × 10−4, η2= 0.076). In addition, two-tailed paired t-tests on 86 six-person pseudo groups showed no significant difference between GNS during decision-making and waiting phases (d, rDLPFC: t85 = 1.400, p= 0.165, Cohen’s d = 0.151, 95% CI = −4.16 × 10−4, 0.002; e, rTPJ: t85 = 1.209, p= 0.230, Cohen’s d = 0.130, 95% CI = −0.001, 0.003). Data are plotted as boxplots for each condition in which horizontal lines indicate median values, boxes indicate 25/75% quartiles, and whiskers indicate the 2.5-97.5% percentile range. Data points outside the range are shown separately as circles. Solid lines start from the mean and reflect the intervals for the Mean ± S.E. **p< 0.01, ***p< 0.001, n.s. not significant. f, g, One-sided permutation test was used to verify the stronger decision-making-increased GNS in real than pseudo groups. Specifically, we calculated the mean difference between the two phases (GNSdecision-making – GNSwaiting) to indicate decision-making-increased GNS respectively for each real or pseudo group. We then compared the real-group sample with 1000 pseudo-group samples18,68,80. We tested the decision-making-increased GNS of the real sample against permutation samples based on decision-making-increased GNS (n = 1000, each permutation sample contains 172 within-condition 3-person pseudo groups). To test whether the effects observed in real groups was larger than that in pseudo groups, we reported the 1-sided p-values. The empirical p value was calculated as80: p=j/1000, j is the number of samples out of the 1000 permutation samples, of which the decision-making-increased GNS was larger than the observed value of real groups. We showed that, for both rDLPFC and rTPJ, the observed difference in decision-making-increased GNS in real groups were outside the upper limit of 95% CI of the permutation distribution. The one-sided p-values indicated specific decision-making-increased GNS in real groups rather than pseudo groups (f, rDLPFC: p = 0.004; g, rTPJ: p = 0.012).

Extended Data Fig. 3 Validation of Bonding-by-Role interaction on Within-group Neural Synchronization.

a, To confirm stronger Bonding-by-Role interaction in real than pseudo groups, we conducted 3-way ANOVAs on GNS with factors of Bonding, Role, and Group, at channels that showed significant Bonding-by-Role interaction in real groups (86 six-person real groups and 86 six-person pseudo groups). We observed significant 3-way interaction of Bonding-by-Role-by-Group on GNS in rDLPFC and TPJ (channel 8: F1,168 = 7.578, p = 0.007, η2 = 0.043; channel 11: F1,168 = 8.318, p = 0.004, η2 = 0.047; channel 4: F1,168 = 7.085, p = 0.009, η2 = 0.040; channel 13: F1,168 = 5.111, p = 0.025, η2 = 0.030, survived FDR correction for the testing channels). Bonding-by-Role interaction in pseudo groups were not significant (channel 8: F1,84 = 0.292, p = 0.590, η2 = 0.003; channel 11: F1,84 = 1.096, p = 0.298, η2 = 0.013; channel 4: F1,84 = 1.424, p = 0.236, η2 = 0.017; channel 13: F1,84 = 0.016, p = 0.900, η2 = -1.88 × 10−4). Data are plotted as boxplots for each condition in which horizontal lines indicate median values, boxes indicate 25/75% quartiles, and whiskers indicate the 2.5-97.5% percentile range. Data points outside the range are shown separately as circles. Solid lines start from the mean and reflect the intervals for the Mean ± S.E. *p< 0.05, **p< 0.01, n.s. not significant. b, One-sided permutation test was conducted to verify the stronger Bonding-by-Role on GNS in real than pseudo groups. Specifically, we compared the real-group sample with 1000 pseudo-group samples18,68,80. We calculated the Bonding-by-Role interaction on GNS: No-bonding (GNSdefender − GNSattacker) − Ingroup-bonding (GNSdefender − GNSattacker), for each real and pseudo group. We then tested the observed Bonding-by-Role interactive effects on GNS of the real groups against the permutation samples based on the Bonding-by-Role interactive effects on GNS (n = 1000, each permutation sample contains 172 within-condition three-person pseudo groups). We showed that the observed differences in the interactive effects on GNS in real groups were outside the upper limit of 95% CI of the permutation distribution. The empirical p value was calculated as80: p=j/1000, j is the number of samples out of the 1000 permutation samples, of which the Bonding-by-Role interaction on GNS was larger than the observed value of real groups. The one-sided p-values indicated stronger interaction on GNS in real groups (channel 8: j = 0; channel 11: j = 0; channel 4: p = 0.015; channel 13: p = 0.001, survived FDR-correction for the testing channels).

Extended Data Fig. 4 The Effect of Decision Similarity on Within-group Neural Synchronization (GNS).

To test whether GNS just reflected individual making similar decisions, we included a within-subject factor (within-group decision similarity) in the analyses, resulting in Role (attacker vs. defender) × Bonding (in-group bonding vs. control) × Similarity (similar vs. dissimilar) mixed-model ANOVAs on GNS. For each 3-person group, each contest round, we calculated the within-group decision similarity as the investment difference of each pair of the 3-person group for each round [i.e., (|x1x2| + |x2x3| + |x1x3|) for attacker group; (|y1y2| + |y2y3| + |y1y3|) for defender group]. Using median split on the mean value of decision similarity across all rounds, we categorized all the rounds into “similar” and “dissimilar” decisions. First, the Bonding × Role interaction remained significant when including within-group decision similarity in the analysis in rDLPFC. However, the main effect of Similarity (a, channel 8: F1,84 = 2.334, p= 0.130, η2 = 0.027; channel 11: F1,84 = 3.553, p= 0.063, η2 = 0.041; channel 4: F1,84 = 1.568, p= 0.214, η2 = 0.018; channel 13: F1,84 = 1.203, p= 0.276, η 2 = 0.014) or its interaction with Role and/or Bonding (b, Similarity-by-Role-by-Bonding: channel 8: F1,84 = 2.019, p= 0.159, η2 = 0.023; channel 11: F1,84 = 1.191, p= 0.278, η 2 = 0.014, channel 4: F1,84 = 3.627, p= 0.060, η2 = 0.041; channel 13: F1,84 = 0.579, p= 0.449, η2 = 0.007; Similarity-by-Role: channel 8: F1,84 = 1.445, p= 0.233, η2 = 0.017; channel 11: F1,84 = 0.016, p= 0.899, η2 = 1.94 × 10−4, channel 4: F1,84 = 0.126, p= 0.724, η2 = 0.001; channel 13: F1,84 = 0.026, p= 0.873, η2 = 3.07 × 10−4; Similarity-by-Bonding: channel 8: F1,84 = 0.157, p= 0.693, η2 = 0.002; channel 11: F1,84 = 0.473, p= 0.493, η2 = 0.006, channel 4: F1,84 = 0.172, p= 0.679, η2 = 0.002; channel 13: F1,84 = 0.160, p= 0.690, η2 = 0.002) did not reach significance. Mixed-model ANOVAs, n= 86 three-versus-three-person intergroup competitions. Data are plotted as boxplots for each condition in which horizontal lines indicate median values, boxes indicate 25% and 75% quartiles, and whiskers indicate the 2.5-97.5% percentile range. Data points outside the range are shown separately as circles. Solid lines start from the mean and reflect the intervals for the Mean ± S.E. n.s. not significant.

Extended Data Fig. 5 Illustration of Group-averaged Neural Activity and rDLPFC-rTPJ Connectivity Analyses.

a, Group-averaged Neural Activity. We first assessed the neural responses of a single brain (i.e., individual neural activity) by performing pre-processing on the fNIRS denoised Oxy-Hb data, including discrete cosine transforms with cut-off period of 128 s and pre-coloring based on hemodynamic response function (HRF). The preprocessed Oxy-Hb time series of each channel were segmented into 3 conditions, i.e., the decision-making phase (illustrated in the figure), the waiting screen, and the outcome screen. The pre-processed Oxy-Hb during decision-making (outcome) phase was z-score transformed using the mean value and standard deviation of the waiting period (as baseline) and indicated decision-making (outcome) related activity74. For each intergroup contest, we averaged across 3 participants sharing the same role to indicate the round-level neural responses of each 3-person group. b, Group-averaged rDLPFC-rTPJ connectivity. Similar to previous studies37, we performed coherence analysis between each of the 7 channels in the rDLPFC with each of the 7 channels in the rTPJ (i.e., 49 channel pairs) for each participant. The calculation of functional connectivity between channel 8 in the rDLPFC and channel 9 in the rTPJ was illustrated in the figure. We then averaged the coherence values of the 3 participants within the same group to indicate the group-averaged functional connectivity (GFC) of rDLPFC–rTPJ.

Extended Data Fig. 6 In-group Bonding Increased Group-averaged Functional Connectivity (GFC) of rDLPFC-rTPJ.

We conducted Bonding (in-group bonding vs. no-bonding control) × Role (attacker vs. defender) mixed-model ANOVAs on the GFC of rDLPFC-rTPJ. We showed that in-group bonding (relative to no-bonding control) increased rDLPFC-rTPJ connectivity in 18 rDLPFC-rTPJ channel pairs (FDR corrected for 49 rDLPFC-rTPJ channel pairs, ar, Supplementary Table 8 gives the full statistical report) and grand mean rDLPFC-rTPJ connectivity (s, F1,84 = 9.033, p= 0.003, η2 = 0.097). Mixed-model ANOVAs, n= 86 three-versus-three-person intergroup competitions. Data are plotted as boxplots for each condition in which horizontal lines indicate median values, boxes indicate 25/75% quartiles, and whiskers indicate the 2.5-97.5% percentile range. Data points outside the range are shown separately as circles. Solid lines start from the mean and reflect the intervals for the Mean ± S.E. *p<0.05, **p<0.01, ***p<0.001.

Extended Data Fig. 7 Within-group Neural Synchronization after Winning or Losing.

We compared the GNS after the group won or lost the last round. We conducted Bonding (in-group bonding vs. no-bonding control) × Outcome of last-round T-1 (win(T-1) vs. lose(T-1)) mixed-model ANOVAs on GNS of round T. For both attacker (a) and defender (b) groups, the main effect of Outcome was not significant (a, Attacker: channel 8: F1,82 = 3.357, p = 0.071, η2 = 0.039; channel 11: F1,82 = 1.164, p= 0.284, η2 = 0.014; channel 4: F1,82 = 2.893, p= 0.093, η2 = 0.034; channel 13: F1,82 = 0.824, p = 0.367, η2 = 0.010; b, Defender: channel 8: F1,82 = 1.321, p= 0.254, η2 = 0.016; channel 11: F1,82 = 2.100, p = 0.151, η2 = 0.025; channel 4: F1,82 = 0.008, p= 0.930, η2 = 9.39 × 10−5; channel 13: F1,82 = 0.183, p= 0.670, η2 = 0.002). The Outcome effect was not modulated by in-group bonding (a, Attacker: channel 8: F1,82 = 0.019, p= 0.890, η2 = 2.35 × 10−4; channel 11: F1,82 = 1.122, p= 0.293, η2 = 0.014; channel 4: F1,82 = 0.244, p= 0.622, η2 = 0.003; channel 13: F1,82 = 0.888, p = 0.349, η2 = 0.011; b, Defender: channel 8: F1,82 = 0.189, p= 0.665, η2 = 0.002; channel 11: F1,82 = 1.988, p = 0.162, η2 = 0.024; channel 4: F1,82 = 0.124, p= 0.725, η2 = 0.002; channel 13: F1,82 = 3.39 × 10−4, p= 0.985, η2 = 4.14 × 10−6). Data from two sessions were excluded as attackers in these two sessions did not win any of the 24 rounds. Outcome-by-Bonding mixed-model ANOVAs (n = 84, 42 three-person groups under no-bonding control and 42 three-person groups under in-group bonding) separately for attacker groups and defender groups. Data are plotted as boxplots for each condition in which horizontal lines indicate median values, boxes indicate 25/75% quartiles, and whiskers indicate the 2.5-97.5% percentile range. Data points outside the range are shown separately as circles. Solid lines start from the mean and reflect the intervals for the Mean ± S.E. n.s. not significant.

Extended Data Fig. 8 Within-group Averaged Neural Activity at Channel 8 in the rDLPFC after Winning or Losing.

We compared the within-group averaged neural activity in rDLPFC after the group won or lost the last round. We conducted Bonding (in-group bonding vs. control) × Outcome of last-round T-1 (win(T-1) vs. lose(T-1)) mixed-model ANOVAs on the within-group averaged neural activity of round T. First, we found a significant main effect of Outcome, i.e., stronger neural activity in the rDLPFC during the next-round decision-making phase after the group lost (relative to won) the game, for both the attacker (a, channel 8: F1,82 = 29.791, p= 5.00 × 10−7, η2 = 0.266) and defender (b, channel 8: F1,82 = 26.595, p= 1.71 × 10−6, η2 = 0.245). Moreover, for the attacker group, we found a significant interaction of Bonding and Outcome on within-group averaged rDLPFC activity (channel 8: F1,82 = 13.207, p = 4.85 × 10−4, η2 = 0.139). Outcome-by-Bonding mixed-model ANOVAs (n = 84, 42 three-person groups under no-bonding control and 42 three-person groups under in-group bonding separately for attacker groups and defender groups). Data from two sessions were excluded as attackers in these two sessions did not win any of the 24 rounds. Data are plotted as boxplots for each condition in which horizontal lines indicate median values, boxes indicate 25/75% quartiles, and whiskers indicate the 2.5-97.5% percentile range. Data points outside the range are shown separately as circles. Solid lines start from the mean and reflect the intervals for the Mean ± S.E. ***p< 0.001.

Extended Data Fig. 9 Effect of In-group Bonding Manipulation on Intergroup Discrimination.

We compared in-group bonding index indicated by greater intergroup discrimination (a measure independent of the intergroup contest) before and after the bonding manipulation to quantify the effect of bonding on increasing intergroup discrimination. a, Before the bonding manipulation, participants in different conditions showed the same level of intergroup discrimination in the intergroup Dictator Game (iDG) (Bonding: F1,89 = 2.249, p= 0.137, η2 = 0.025; Role: F1,89 = 0.181, p= 0.672, η2 = 0.002; Bonding × Role: F1,89 = 0.243, p = 0.623, η2= 0.003). Mixed-model ANOVA, n= 91 three-versus-three-person intergroup competitions. b, In-group bonding (relative to no-bonding control) increased intergroup discrimination in both attacker (t89 = 5.453, p= 4.39 × 10−7, Cohen’s d = 1.142, 95% CI = 3.173, 6.812) and defender groups (t89 = 2.788, p= 0.006, Cohen’s d = 0.584, 95% CI = 0.810, 4.833). Two-tailed independent-samples t-test, 47 three-person attacker (defender) groups under in-group bonding and 44 three-person attacker (defender) groups under no-bonding control. c, We directly examined the in-group bonding induced intergroup discrimination by calculating the change from before to after bonding manipulation (i.e. iDG-iDG0). We showed that, in-group bonding manipulation indeed induced a reliable increase in intergroup discrimination (t46 = 7.220, p = 4.27 × 10−9, Cohen’s d = 1.053, 95% CI = 2.354, 4.173, two-sided one-sample t-test, 47 three-versus-three-person intergroup competitions under in-group bonding condition), which was observed in both attacker (t46 = 3.171, p = 0.003, Cohen’s d = 0.463, 95% CI = 0.925, 4.140, two-sided one-sample t-test, 47 three-person attacker groups) and defender groups (t46 = 5.222, p = 4.16 × 10−6, Cohen’s d = 0.762, 95% CI = 2.455, 5.534, two-sided one-sample t-test, 47 three-person defender groups). In addition, under no-bonding control condition, there was no change of intergroup discrimination for attacker (t43 = −1.464, p = 0.150, Cohen’s d = −0.221, 95% CI = −2.867, 0.455, two-sided one-sample t-test, 44 three-person attacker groups) or defender groups (t43 = 1.936, p = 0.059, Cohen’s d = 0.292, 95% CI = −0.075, 3.692, two-sided one-sample t-test, 44 three-person defender groups). Data showed as Mean ± S.E. with overlaid dot plots. **p < 0.01, ***p< 0.001, n.s. not significant.

Extended Data Fig. 10 Illustration of group-level neural activity and rDLPFC-rTPJ connectivity calculation.

a, Different from group-averaged neural activity in Extended Data Fig. 5, we calculated the group-level neural activity by first averaging the denoised Oxy-Hb neural activity across 3 participants of each pseudo groups. The group-level, preprocessed Oxy-Hb time series of each channel were then segmented into 3 conditions, i.e., the decision-making phase, the waiting screen, and the outcome screen. The pre-processed, group-level Oxy-Hb during decision-making (outcome) phase was z-score transformed using the mean value and standard deviation of the group-level waiting period (as baseline) and indicated group-level decision-making (outcome) related activity. b, The group-level function connectivity between rDLPFC and rTPJ was calculated by first averaging the denoised Oxy-Hb neural activity in each channel across 3 participants of each group. We then performed coherence analyses between each of the 7 channels in the rDLPFC with each of the 7 channels in the rTPJ (i.e., 49 channel pairs) to index channel-pairwise group-level functional connectivity (GFC) of rDLPFC–rTPJ. We also calculated at the grand mean level (i.e., the averaged coherence value of 49 channel pairs) to index grand mean group-level functional connectivity.

Supplementary information

Supplemental Information

Supplementary Figures 1–5 and Supplementary Tables 1–11.

Reporting Summary

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Yang, J., Zhang, H., Ni, J. et al. Within-group synchronization in the prefrontal cortex associates with intergroup conflict. Nat Neurosci 23, 754–760 (2020). https://doi.org/10.1038/s41593-020-0630-x

Download citation