How a speaker herds the audience: Multi-brain neural convergence over time during naturalistic storytelling.

Sharing narratives is an ancient and efficient way for humans to transmit experiences to each other. The efficacy of a given narrative has been shown to be associated with the similarity between the brain activation patterns of the speaker and the listeners (SL), as well as the neural similarity among the listeners (LL). Operationalizing the pattern (dis)similarity as the distances between participants, this study proposes a "herding hypothesis". That is, like a group of sheep guided by a shepherd, the more closely the listeners follow the speaker, i.e. higher SL similarity, the more tightly the listeners will tend to cluster together, i.e. higher LL similarity. Using fMRI data collected during the verbal production of two spoken narratives as well as in an audience of listeners, we found that SL and LL similarities are correlated across time, as predicted by the herding hypothesis. In addition, the more "herded" brain regions also show a stronger LL similarity at the more engaging moments of the narrative, supporting an interpretation that the herding effect reflects effective storytelling. By taking both LL and SL neural coupling into consideration in a moment-by-moment manner, this study demonstrates that examining the dynamic multi-brain functional network can potentially reveal when and how the speaker loses the audience; for example, whether they go astray in all directions or they share the same misunderstanding.