Synchronized affect in shared experiences strengthens social connection

Abstract

People structure their days to experience events with others. We gather to eat meals, watch TV, and attend concerts together. What constitutes a shared experience and how does it manifest in dyadic behavior? The present study investigates how shared experiences-measured through emotional, motoric, physiological, and cognitive alignment-promote social bonding. We recorded the facial expressions and electrodermal activity (EDA) of participants as they watched four episodes of a TV show for a total of 4 h with another participant. Participants displayed temporally synchronized and spatially aligned emotional facial expressions and the degree of synchronization predicted the self-reported social connection ratings between viewing partners. We observed a similar pattern of results for dyadic physiological synchrony measured via EDA and their cognitive impressions of the characters. All four of these factors, temporal synchrony of positive facial expressions, spatial alignment of expressions, EDA synchrony, and character impression similarity, contributed to a latent factor of a shared experience that predicted social connection. Our findings suggest that the development of interpersonal affiliations in shared experiences emerges from shared affective experiences comprising synchronous processes and demonstrate that these complex interpersonal processes can be studied in a holistic and multi-modal framework leveraging naturalistic experimental designs.

Fig. 1. Components of shared experiences.

Simultaneously experiencing a stimulus between individuals can lead to affective experiences synchronized in time (temporal affective synchrony), similar patterns of facial muscle activations in how emotion is displayed (spatial motor alignment), physiological synchrony, and similar cognitive appraisals of the experience (appraisal alignment) that help facilitate the development of interpersonal affiliation.

Fig. 2. Temporal facial expression synchrony.

a Examples of participant facial expressions. Synchrony in the alone group was computed between all pairwise participants who watched individually whereas synchrony in the dyad group was computed only between the pairs of participants who watched physically together sitting next to each other. Images are displayed with written consent from the participants. b Synchrony of positive facial expressions while watching episodes between conditions in each episode. Significance of group differences was tested with a two-sample permutation test (error bars indicate 95% confidence intervals *p < 0.05, **p < 0.01, ***p < 0.001). c Dynamic facial expression synchrony for the dyad and alone group in episode 1. Dotted boxes indicate clusters at which dyad participants synchronized more than alone participants. Select depictions for the significant clusters are shown above. These moments correspond to scenes with high joy ratings rated by N = 188 independent observers and are indicated by orange color bars along the x-axis (shaded areas indicate standard error of the mean). Colored ticks along the horizontal axis indicate the most representative emotion at that time from online crowdsourced self-reported emotion ratings (see Supplementary Methods). Original illustrations of highlighted scenes depicted from the Friday Night Lights television show were created by Dr. Stephanie Lee and shared with permission.

Fig. 3. Correlation between connectedness and facial expression synchrony in each episode.

a Synchrony of positive facial expressions and average connectedness ratings after each episode. b Synchrony of negative facial expressions and average connectedness ratings after each episode. CI indicates subject-wise bootstrapped 95% confidence interval. *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 4. Spatial configuration synchrony results.

a Estimated average shared response trajectories. The trajectory of the first shared response (Shared Response 1; solid line) included peaks at scenes when crowdsourced joy ratings (orange) were high, such as when the characters were joking around or when a football team won the match. The trajectory of the second shared response (Shared Response 2; dotted line) included peaks at scenes when crowdsourced fear ratings (purple) were high, such as when the coach was warned about the social consequences of losing a football match or when the main character gets injured during a football match. Shared response trajectories were linearly detrended and standardized for visualization. Colored ticks along the horizontal axis indicate the most representative emotion at that time from online crowdsourced self-reported emotion ratings. b Examples of participant-specific spatial patterns of action units activated for each shared response. Note the variability in expressions despite their occurrences are aligned to the shared response trajectories shown in Fig. 4a. These values have been exponentiated for display purposes. Faces are generated using open-source py-feat software which allows facial expressions to be displayed without revealing participants identity. c Shared Response 1 spatial configuration pattern similarity is associated with connectedness ratings in dyads. CI indicates subject-wise bootstrapped 95% confidence interval. ^十p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 5. EDA synchrony.

a Average global EDA synchrony in the dyad and alone groups across each 45-min episode. b Dynamic EDA synchrony of dyad and alone groups. Dashed boxes show moments when the EDA synchrony of dyads was significantly greater than the EDA synchrony of alone participants. The dominant crowdsourced emotions at these scenes included fear, anger, and surprise. A sample image from select scenes is depicted in shaded boxes showing football players smashing into each other (7th minute), couples having a verbal argument (20th minute), and a football player getting seriously injured (34th minute). Shaded areas indicate standard error of the mean. Colored ticks along the horizontal axis indicate the most representative emotion at that time from online crowdsourced self-reported emotion ratings. c Effect of EDA synchrony on average self-reported connectedness ratings for dyads. Brackets show subject-wise bootstrapped 95% confidence intervals on the correlation estimates. *p < 0.05, **p < 0.01, ***p < 0.001. Original illustrations of highlighted scenes depicted from the Friday Night Lights television show were created by Dr. Stephanie Lee and shared with permission.

Fig. 6. Similarity of character impressions.

a Cumulative explained variance of principal components across impression ratings. The dotted line indicates 90% cumulative explained variance. b Component loadings on each impression dimension. c Correlation between impression similarity and dyad average connection ratings for each episode. CI indicates subject-wise bootstrapped 95% confidence interval. ^十p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 7. Path diagram of relationships between synchrony measures, shared experience, and connection ratings estimated using structural equation modeling.

Here we plot the path diagram of our structural equation model. Measured variables are depicted with a square and include temporal affective synchrony, spatial motor similarity of facial expressions, EDA synchrony, similarities in character impression ratings, and social connection ratings. Shared experience is modeled as a latent variable and depicted with a circle. Direct relationships are visualized with straight lines with a single arrow. Covariance between variables is visualized using curved lines with two arrows. The disturbance or unexplained residual error for each measured variable is depicted using circular arrows. Numerical values along the paths indicate standardized contributions. Overall, we find that temporal affective synchrony most strongly loads on the latent shared experience experience variable, followed by spatial motor synchrony and character impression similarity. Shared experiences, in turn, predict social connection controlling for the amount of time spent by participants together during the experiment (episode number). *p < 0.05, **p < 0.01, ***p < 0.001.