Materials and procedure

Participants sat next to each other facing a screen (Fig. 2A). Once the NIRS cap was placed and recording locations localised, the experiment started. Participants first watched one episode of the BBC Dipdap animated series (Phase 1), then chatted about unrelated topics for about 20 minutes (Phase 2), before watching another (new) Dipdap episode (Phase 3). Each episode lasts 2 minutes and shows the adventures of Dipdap, an animated puppet who has to face a series of challenges drawn on the screen by a line. The episodes are all non-verbal and can be watched in any order as they are all self-contained. They are particularly useful to engage the watcher’s imagination, as one follows the drawing line creating new and surprising scenarios for the puppet. The two episodes, ‘Balloon’ (no.7) and ‘Headphones’ (no.38), were selected randomly from the full list of episodes (available at https://www.bbc.co.uk/iplayer/episodes/b00xgpj9/dipdap?page=1), and were presented in a counter-balanced order over phase 1 (pre-conversation) and phase 3 (post-conversation). During the co-watching parts (phase 1 and 3), a separator was placed in between the two participants, ensuring they did not engage in any social communication during the presentation of the episode. In fact, it has been shown that even eye-contact without verbal exchange modulates brain synchrony in dyads [17]. From an ecological point of view, this would resemble being at the cinema, where the dark environment prevents one from seeing their friends, although they may be aware of their presence next to them. Similarly, in this experiment participants could not see their partner’s face but could perceive their presence next to them as they were sitting next to each other. During the social interaction part (Phase 2, conversation phase), participants undertook another experiment where they engaged in a face-to-face semi-structured conversation for about 20 minutes, during which they shared facts about novel items (e.g. exotic animals, musical instruments etc) to each other. Specifically, each participant alternatively played the role of the teacher (sharing facts previously learned) and the role of the learner (listening to the teacher’s description and memorising the facts). During the conversation phase, participants were instructed to memorise facts from each other (as part of another experiment). None of the participants ever mentioned the cartoon episode during the conversation phase.

Example of data processing streamline for one dyad (and one channel/ROI) – A. participants seat next to each other and watch an episode of the BBC series ‘Dipdap’ (Phase 1). After watching one episode, the two participants engage in a social interaction task, when they chat about unrelated topics (Phase 2). They than watch another – novel - episode of Dipdap (Phase 3). During co-watching (phase 1 and 3), a separator ensures that participants do not engage in any form of communication. The two Dipdap episodes were randomly allocated to phase 1 or 3 (counterbalanced across dyads). They are all non-verbal, self-contained, identical for duration and comparable in terms of audio/visual features. B. Full session Nirs Signal (HbCBSI) plotted for participant A (red) and participant B (blue). Nirs signal during each video cowatching is highlighted. C. Wavelet coherence spectrogram for video 1 and video 2. Bars show the frequency of interest used in analysis. D. Bars plot of the mean for the three frequencies of interest (High: 0.1-0.2 Hz, Medium: 0.03-0.1 Hz, Low: 0.02-0.03 Hz) for video 1 and video 2. Data plotted in B., C. and D. belongs to the same dyad.