Virtual social interactions during the COVID-19 pandemic: the effect of interpersonal motor synchrony on social interactions in the virtual space

The study investigates whether social phenomena known from face-to-face interactions—specifically interpersonal motor synchrony and emotional alignment (contagion)—also emerge during virtual video-based interactions and whether such interactions induce pro-social effects. Prior work shows motor synchrony can occur intentionally or spontaneously and is linked to increased liking, trust, cohesion, and perceived similarity. Emotional alignment is thought to share mechanisms with synchrony via a feedback loop integrating error monitoring, alignment, and reward systems. With the COVID-19 pandemic pushing social exchanges online, the authors ask if video-based virtual interactions can elicit these core processes and thereby foster connectedness, reduce negative affect, and promote pro-social outcomes similar to in-person settings.

The paper reviews evidence that motor synchrony is fundamental to human social behavior, arising early in development and facilitating prosocial outcomes (liking, trust, cohesion). It occurs both intentionally (e.g., dance) and spontaneously (e.g., posture, walking cadence). Emotional alignment (contagion) often co-occurs with synchrony and may share neural mechanisms within the mirror neuron/action-observation network, engaging embodied simulations and reward processes. Group membership moderates both synchrony and emotional contagion, and both are associated with perceived similarity and closeness. While many face-to-face studies reveal pro-social effects of synchrony, some virtual agent studies do not replicate these effects, suggesting context and modality matter. The expansion of virtual communication during COVID-19 raises questions about whether audio-video interactions can elicit synchrony and alignment and support connectedness, given mixed findings on engagement and creativity in virtual contexts.

Design: Experimental task inducing a brief, emotionally salient virtual social interaction via Zoom between two previously unacquainted, same-sex partners. Participants were instructed to share personal difficulties experienced during COVID-19. Participants: 196 Hebrew-speaking students (ages 19–30) from Ariel University, randomly assigned to 98 same-sex stranger dyads. After exclusions for technical issues (unstable internet, audio/video lag preventing Motion Energy Analysis), 69 dyads remained for MEA analyses. IRB approval obtained; informed consent provided. Procedure: Prior to each session, participants completed demographic questions and baseline PANAS online. In Zoom, a research assistant (RA) instructed participants to pin the partner’s video and exchange brief introductions (2 minutes). Participants then turned off cameras/microphones to complete baseline measures of trust, liking, perceived similarity, self-other overlap (IOS), and cohesion. Next, for 5 minutes with cameras and microphones on, dyad members discussed personal COVID-19-related difficulties while the RA’s camera/mic were off. Finally, cameras/microphones were turned off, and participants completed post-interaction measures (trust, liking, perceived similarity, IOS, cohesion, PANAS). Sessions were recorded (1920×1080, 30 fps), with standardized conditions (static background, stable lighting). Some noisy frames were manually removed. Measures: - Motor synchrony: Motion Energy Analysis (MEA) using rMEA in R. Regions of interest (ROI) encompassed head and shoulders of each participant. Motion energy time series (greyscale frame-to-frame change) were extracted. Synchrony quantified via time-lagged cross-correlation (MEAccf), using 5-minute series segmented into 60 s windows (30 s overlap), lag window ±5 s. Correlations Fisher z-transformed; absolute values aggregated to yield a global simultaneous synchrony value per dyad. Pseudo-synchrony: Created 9,660 pseudo-dyads by shuffling and pairing participants from different dyads; analyzed identically to estimate chance-level synchrony. - Emotional alignment: PANAS Positive Affect (PA) and Negative Affect (NA) administered pre- and post-interaction. Emotional alignment indices computed as absolute differences between partners’ PA and NA scores within dyads; lower values indicate greater alignment. Change scores computed as post minus pre. - Pro-social outcomes (self-report): • Trust (5 items; 1–5 Likert; α pre=0.72, post=0.76) • Liking (4 items; 1–5 Likert; α pre=0.80, post=0.86) • Cohesion (3 adapted items on similarity, connectedness, trust; continuous scale; α pre=0.87, post=0.93) • Self-other overlap (IOS single-item pictorial scale) • Perceived similarity (25 items; background and attitudes subscales; 1–5 Likert; α pre=0.89, post=0.89) Analytic strategy: - Task validation: Independent-samples t-test comparing global synchrony in real versus pseudo-dyads. - Emotional alignment: Paired t-tests comparing pre vs post absolute difference indices for NA and PA. - Pro-social effects: Repeated-measures MANOVA on dyad-average scores (PA, NA, trust, liking, cohesion, IOS, perceived similarity) pre vs post. - Associations with synchrony: Multiple linear regression predicting dyadic synchrony from changes in emotional alignment (PA and NA indices) and changes in each pro-social measure (trust, liking, cohesion, IOS, perceived similarity).

- Motor synchrony occurred during virtual interactions: Real dyads showed significantly higher synchrony (M=0.14, SD=0.04) than pseudo-dyads (M=0.04, SD=0.02); t(9727)=39.92, p<0.001. - Emotional alignment: Negative affect alignment increased (absolute difference decreased) from pre (M=6.50, SD=4.74) to post (M=4.01, SD=4.39); t(68)=4.12, p<0.001. Positive affect alignment did not significantly change (pre M=8.67, SD=5.80; post M=7.83, SD=6.35); t(68)=1.12, p=0.268. - Pro-social outcomes (dyad averages) improved from pre to post (repeated-measures MANOVA): • Positive affect: F(1,68)=32.82, p<0.001 (means 32.39→35.80) • Negative affect: F(1,68)=27.03, p<0.001 (means 11.88→9.80) • Trust: F(1,68)=45.25, p<0.001 (means 14.57→15.67) • Liking: F(1,68)=40.02, p<0.001 (means 16.92→17.86) • Cohesion: F(1,68)=107.40, p<0.001 (means 130.20→160.80) • Self-other overlap: F(1,68)=222.72, p<0.001 (means 41.43→57.43) • Perceived similarity: F(1,68)=10.99, p=0.001 (means 55.87→58.06) - Regression linking synchrony with alignment and pro-social changes: Overall model R²=0.293, F(7,61)=3.62, p=0.003. Significant predictors: • Positive affect alignment change (β=−0.394, p<0.001): Greater increase in alignment (larger reduction in PA difference) associated with higher synchrony. • Change in liking (β=0.496, p<0.001): Greater increase in liking associated with higher synchrony. Other predictors (negative affect alignment, trust, cohesion, IOS, perceived similarity) were not significant in the final model at the reported thresholds.

The study demonstrates that core social processes—motor synchrony and emotional alignment—can emerge spontaneously during brief, audio-video virtual interactions between strangers. These interactions also yielded pro-social outcomes: reduced negative affect, increased positive affect, and heightened trust, liking, cohesion, perceived similarity, and self–other overlap. Importantly, higher synchrony was specifically associated with stronger positive affect alignment and greater increases in liking, suggesting synchrony may be a mechanism supporting connection and positive evaluation in virtual contexts. In light of pandemic-induced reliance on virtual platforms, the findings indicate that video-based interactions can approximate key social dynamics of in-person encounters. Differences across the literature may be attributable to modality (text vs audio-video), context, and partner type (human vs agent), pointing to the importance of audio-visual cues for eliciting embodied alignment processes. The results support the hypothesis that virtual interactions, when rich in sensory and temporal information, can foster connectedness via mechanisms similar to face-to-face interactions.

This work provides evidence that both motor synchrony and emotional alignment can arise spontaneously in virtual, audio-video interactions and that such interactions enhance multiple pro-social outcomes. Synchrony was particularly linked to positive affect alignment and increased liking, highlighting a potential mechanism for virtual connectedness. These findings suggest that thoughtfully designed virtual social protocols could mitigate social distancing consequences and inform interventions targeting loneliness and social well-being. Future research should include direct comparisons with face-to-face interactions, manipulate interaction modalities (text vs audio vs video), explore causal pathways between synchrony, emotional alignment, and pro-social outcomes, and test diverse populations and contexts to assess generalizability and optimize virtual intervention designs.

- No direct face-to-face comparison group, limiting conclusions about equivalence to in-person interactions. - Correlational/associational design precludes causal inference regarding relations among synchrony, emotional alignment, and pro-social effects. - Homogeneous sample: university students (ages 19–30) from a single Israeli institution; uneven gender ratio; limits generalizability. - Motor synchrony was measured from head-and-shoulders only; full-body dynamics were not captured. - Presence of a research assistant during sessions (camera/mic off) may have influenced behavior. - Technical exclusions (unstable internet, audio/video lag) reduced the analyzed sample to 69 dyads, potentially introducing selection effects. - Task and context specificity: a 5-minute emotionally focused conversation; findings may not generalize to other virtual interaction types or durations.