Hyperscanning: Beyond the Hype

Hyperscanning—the recording of brain activity from multiple individuals—can be hard to interpret. This paper shows how integrating behavioral data and mutual prediction models into hyperscanning studies can lead advances in embodied social neuroscience. Social interaction is central our cognition health. Our actions decision-making everyday life are heavily influenced by others, atypical interactions a feature the majority psychiatric mental health conditions. Understanding mechanisms therefore an important goal for human Hyperscanning, measurement more than one individual at same time (Figure 1A), has recently been hailed as game-changer study (Gvirts Perlmutter, 2019Gvirts H.Z. Perlmutter R. What Guides Us Neurally Behaviorally Align With Anyone Specific? A Neurobiological Model Based on fNIRS Hyperscanning Studies.Neuroscientist. 2019; 26: 108-116Crossref PubMed Scopus (29) Google Scholar). takes critical look some claims that have made limitations many analyses. We further suggest that, behavior, it will possible move beyond hype realize potential this new domain. focus primarily functional near-infrared spectroscopy (fNIRS; Pinti et al., 2020aPinti P. Tachtsidis I. Hamilton A. Hirsch J. Aichelburg C. Gilbert S. Burgess P.W. The present future use (fNIRS) cognitive neuroscience.Ann. N Y Acad. Sci. 2020; 1464: 5-29Crossref (168) Scholar) because most widely used modalities, but we also consider EEG electrophysiology. good example genre influential Cui colleagues (Cui 2012Cui X. Bryant D.M. Reiss A.L. NIRS-based reveals increased interpersonal coherence superior frontal cortex during cooperation.Neuroimage. 2012; 59: 2430-2437Crossref (339) captured prefrontal pairs participants performing two distinct tasks: cooperation task where both must try press button time, competition fast possible. cannot communicate directly receive feedback after each trial, which enables them improve their performance. signals were analyzed with wavelet 1B), showed greater right blocks compared or solo number replicated extended finding when engaged interaction, including situations conversation, eye contact, decision-making, motor coordination tasks (Fishburn 2018Fishburn F.A. Murty V.P. Hlutkowsky C.O. MacGillivray C.E. Bemis L.M. Murphy M.E. Huppert T.J. Perlman S.B. Putting heads together: neural synchronization biological mechanism shared intentionality.Soc. Cogn. Affect. Neurosci. 2018; 13: 841-849Crossref (25) Furthermore, differences strength cross-brain seen ingroup outgroup (Yang 2020Yang Zhang H. Ni De Dreu C.K.W. Ma Y. Within-group associates intergroup conflict.Nat. 23: 754-760Crossref Scholar), mixed gender within pairs, relation autistic traits children (reviewed these results, strong importance neuroscience psychiatry. For example, brain-to-brain coupling described “mechanism transmitting information… regarding remote events” (Hasson 2012Hasson U. Ghazanfar A.A. Galantucci B. Garrod Keysers Brain-to-brain coupling: creating sharing world.Trends 16: 114-121Abstract Full Text PDF (469) intentionality” One suggests “might trigger guiding alignment” It suggested measures might allow us diagnose developmental disorders (Leong Schilbach, 2019Leong V. Schilbach L. promise two-person psychiatry: using interaction-based sociometrics identify interaction.Br. Psychiatry. 215: 1-3Crossref (18) stimulation impose even treat such These ambitious emerging technology deserve careful detailed examination. clearly show there robust patterns across brains, best way interpret less clear. Claims extending feel like telepathy, disconnected standard plausible framework. To obtain better understanding what actually means challenges faced domain, helps first understand basic measures. First, describe relationship between series data, do not take account any other factors: changes participant behavior testing environment Second, only symmetrical effects, brains pattern change. Many (e.g., giving/taking object, turn-taking conversation) asymmetric, having different roles, clear if analysis capture this. Third, results traditional clear—because specific situation, sometimes examine one-person contrasts neuroscientist expect see. An challenge interpretation 1A) inter-subject correlations recorded sequential 1C). know people viewing movie alone MRI scanner similar effect person tells story listens reflect fact common processing external stimulus give rise brains. As pointed out (Burgess, 2013Burgess A.P. On studies: cautionary note.Front. Hum. 2013; 7: 881Crossref (81) could arise respond environmental stimuli. Thus, argued reported reveal set experienced additional factors. There feasible responses processing. design complex methods factor effects. Several increasing tested. records three (or more) who room those while others not, argue matched sensory driving Scholar; Yang However, remains interacting drives second option accept argument measure relevant outcome. Studies levels groups typically overall level index well coordinating. Recent correlate performance Such require very large sample sizes order relate dyad-level behavior. may impossible pin down effects pair. limits usefulness analyses tracking diagnosis clinical populations, single subject define pair group showing activity. Finally, context really measuring exactly thing intersubject fMRI context, then asking watch repeatable use. pioneering contribution done above mean claim revolutionize seem just hype. If unable distinguish processing, value limited. hype, need bigger, experiments stronger theoretical starting point recognition exist bodies. Visual, auditory, processes mediate so bodily together. labs analyze participants, considering uses signal receives information others. now flexible technologies available participants’ hand body movements, facial expressions, physiological changes, systems should lab 1D). tools possible, desirable, away experimental designs discrete trials repeated times. tight control experimenter create natural meaningful engage variety behaviors. Moving toward scenarios, work together over longer period observing responding other, suppressed controlled situation. By capturing analyzing interactions, neurocognitive theory kind interaction. candidate emerges doing They own moving hands, face, gaze interact partner, they predict partner’s act appropriately. Both acting predicting general requirements—whether cooking meal together, playing piano duet, taking turns requirement one’s partner perform action essential fluent idea formalized (Kingsbury 2019Kingsbury Huang Wang Gu K. Golshani Wu Y.E. Hong W. Correlated Neural Activity Encoding Behavior Brains Socially Interacting Animals.Cell. 178: 429-446.e16Abstract (75) here (A-self B-self Figure 1E) (A-Other B-Other 1E). co-localized all summed crude measure, similarity operate mutually coherent Evidence support was found mice. Kingsbury microendoscopic calcium imaging record hundreds neurons dorsomedial (dmPFC) mice tracked interactions. identified encoded mouse’s mouse, consistent model. More importantly, PFC animals, demonstrating fine-grained whole-brain level. quantify asymmetry mice, subordinate dominant Data adding linear model (GLM) individual’s terms (cross-brain GLM xGLM; 1F). outputs researcher (self-behavior, other-behavior, task, contributions) interpreted xGLM allows researchers integrate modalities freely animals test hypothesis tractable fashion. applies humans, useful approaches modeling studies. Initial hints recent (Pinti 2020bPinti Devoto Greenhalgh role anterior (area 10) face-to-face deception measured fNIRS.Soc. 2020: 1-14Google play card game poker—one informer choose lie tell truth about her card, guess high low win points correct answer. linked deceptive intent. critically, range lags channels informer’s reliably preceded guesser’s 2 s lag. provides clue unfortunately challenging accurately (thus making see signals) did factors sufficient detail implement full approach. Future contrast cooperative competitive without ability recordings valuable. provide initial evidence humans help older datasets original Scholar above. That “cooperation” “competition” cooperation. close examination demands requires degree prediction—participants other’s movements time. In contrast, would distraction. entirely hypothesis. make does demand prediction, winning faster limit. Overall, demonstrate examining Within idea, framework animal several advantages. proposes predictive causes based line cognition. motivates integration unified working enact approach builds long tradition research neuroscience, allowing extend existing include symmetric asymmetric wider contexts fundamental theories examples bodies applied settings unstructured ecologically valid real world. short review interpreting ways overcome challenges. designing research, collect (hands, faces, eyes, speech) cause xGLMs accomplish integration, Granger causality mechanisms, gives tasks, predictions combining go