Understanding and Supporting Co-viewing Comedy in VR with Embodied Expressive Avatars

TL;DR

The paper investigates how embodied expressive avatars with visible laughter cues influence co-viewing of comedy in VR. Using a within-subjects mixed-methods design (VO vs VLC) with N=24 in eight triads, participants watched four 7-minute episodes of a Japanese animation while avatars displayed laughter cues and a mirror allowed viewing of expressions. Findings show that embodied laughter cues shift engagement from individual immersion to socially coordinated participation, amplify emotional contagion, and foster emergent co-viewing norms, though some participants prefer private, avatar-free viewing for certain content. Design implications emphasize flexible emotional visibility, self-avatar feedback loops, and accommodation strategies to balance individual and group-oriented co-viewing.

Abstract

Co-viewing videos with family and friends remotely has become prevalent with the support of communication channels such as text messaging or real-time voice chat. However, current co-viewing platforms often lack visible embodied cues, such as body movements and facial expressions. This absence can reduce emotional engagement and the sense of co-presence when people are watching together remotely. Although virtual reality (VR) is an emerging technology that allows individuals to participate in various social activities while embodied as avatars, we still do not fully understand how this embodiment in VR affects co-viewing experiences, particularly in terms of engagement, emotional contagion, and expressive norms. In a controlled experiment involving eight triads of three participants each (N=24), we compared the participants' perceptions and reactions while watching comedy in VR using embodied expressive avatars that displayed visible laughter cues. This was contrasted with a control condition where no such embodied expressions were presented. With a mixed-method analysis, we found that embodied laughter cues shifted participants' engagement from individual immersion to socially coordinated participation. Participants reported heightened self-awareness of emotional expression, greater emotional contagion, and the development of expressive norms surrounding co-viewers' laughter. The result highlighted the tension between individual engagement and interpersonal emotional accommodation when co-viewing with embodied expressive avatars.

Figures (2)

• Figure 1: Implementation overview. (A) We provided six default avatars from ReadyPlayerMe for participants to choose from. (B) Expressive cues of the avatars include FACS-based facial expressions (e.g., cheek raising, mouth opening) and whole-body movements for seated laughter animation. (C) Users' first-person view in VR. The virtual co-viewing environment consists of a TV screen, a three-person sofa, and a mirror above the TV, allowing users to observe their own and co-viewers' expressions. Users can trigger the laughter expressions of their embodied avatars with the laughter cue trigger on the panel in front of them
• Figure 2: Themes mapped to our three research questions. First, participants’ self‑awareness and engagement shifted once co‑viewers’ expressions became visible (Section \ref{['sec:Content to Social Context']}). Second, visibility of laughter cues promoted emotional contagion and mutual responsiveness (Section \ref{['sec:Emotional Amplification']}). Third, groups developed normative coordination in how and when they displayed laughter (Section \ref{['sec:Reaction Accommodation']}).