Embodied Social Presence Theory

• Embodied Social Presence Theory is a framework that explains how coordinated sensorimotor coupling and interaction dynamics constitute social cognition.
• Empirical studies using minimalist VR and robotics reveal that metrics like turn-taking enhance agency detection and subjective social presence.
• The theory guides the design of social AI, VR, and human–robot systems by emphasizing the roles of feedback loops, embodiment, and linguistic socialization.

Embodied Social Presence Theory (ESPT) describes the constitutive role of embodied interaction dynamics in producing the felt presence and social cognition of another agent, whether human or artificial. ESPT challenges traditional, brain-centric views by positing that social presence can arise from the coordinated, sensorimotor coupling between agents, such that aspects of cognitive work are offloaded to and actively performed by the social interaction itself. The theory is empirically grounded in minimalist virtual and robotic environments, and has direct implications for the design of social AI, human-robot systems, virtual reality (VR), and mixed reality (MR) platforms.

1. Constitutive Interaction and Social Cognition

ESPT asserts that the dynamics of embodied social interaction are not merely triggers for internal processes but are constitutive of social cognition itself (Froese et al., 2014). Experimental evidence demonstrates that co-regulated, minimally mediated interactions (such as haptic exchange via avatars in a minimalist VR) enable explicit detection of agency and the felt presence of the other. In these settings, mutual responsiveness and turn-taking are essential for participants to explicitly recognize and report another’s presence, rather than relying solely on signals processed and represented within one's own brain.

A key operationalization is the turn-taking (TT) index, formally defined as:

$\text{TT} = \frac{4C_1C_2}{T^2}$

where $C_1$ and $C_2$ are the summed active contributions of each participant (i.e., non-overlapping movement intervals) and $T$ is the total number of time steps. Higher TT scores correlate strongly with accurate avatar identification and clear subjective awareness reports, supporting the enactive perspective of “participatory sense-making”: social cognition emerges from structured interaction rather than isolated inference.

2. Role of Embodiment, Language, and Socialization

While classic ESPT emphasizes the necessity of visible bodily cues and real-time coupling, interactional expertise literature complicates the picture (Collins, 2016). At the collective level, both bodily practice and language are essential for forming a “form-of-life,” yet individuals can achieve deep social understanding through linguistic immersion alone. The “strong interactional hypothesis” contends that fluency in social discourse—absent direct physical practice—can suffice for high social presence, as exemplified by disembodied, socialized AI systems engaging through natural language.

This is represented in the framework:

$$\text{Social Presence} = \Phi(\text{Embodiment}, \text{Linguistic Socialization})$$

Thus, while collective embodiments sustain social texture, individual actors—even AI with minimal bodies—may display substantial social presence if constantly updated and socialized via language. In contrast, physically embodied but unsocialized agents remain socially deficient.

3. Physical Embodiment in Socially Interactive Systems

In robotics and virtual agent systems, physical embodiment is shown to be a robust amplifier of social presence, engagement, and collaboration (Deng et al., 2019). Embodiment is operationalized through mechanical structure, sensors, actuators, and the capacity to generate and interpret multiple communication channels (gaze, proxemics, gestures, vocalizations).

Three interlocking taxonomies structure embodiment research: | Taxonomy | Key Axes / Values | Application Example | |---------------------------|--------------------------------------------------------------------------------|--------------------------------------------------| | Robot Embodiment Type | Strongly embodied vs. virtual; design metaphor (humanoid/animal/machine-like); abstraction level | Choice of anthropomorphic vs. stylized forms | | Robot Social Roles | Subordinate ↔ Peer ↔ Superior | Teacher-robot vs. peer-companion | | Human–Robot Tasks | Planning, performance/action, contest/competition, decision making, creativity | Book-moving, creative support, negotiation tasks |

Meta-analyses of 65 studies show that physical embodiment generally enhances user perceptions of social agency and behavioral outcomes (63% of experiments), though the effect is modulated by task context and role calibration. Physically embodied robots produce richer non-verbal signals, boosting perceived intelligence, trustworthiness, and overall social presence in interaction. However, mismatches—e.g., peer-like robots in authoritative tasks—can nullify or even reverse these benefits.

4. Dynamics, Ontology, and Contextual Emergence

ESPT incorporates an ontological account of how and when a system becomes socially embodied (Seaborn et al., 2021). The “Tepper line” heuristic defines the contextual, dynamic threshold at which a physical or virtual agent is perceived as both social and agentic by a human. This state is not binary but emerges as a function of the system’s morphology, intelligence, and interactive behavior, combined with situation-specific context and user perception:

$$\text{Social Embodiment} \propto f(\text{Morphology}, \text{Intelligence}, \text{(Re/)Action}, \text{Context}, \text{Purpose})$$

Case studies (e.g., tele-robotic SideBot, voice-assistant Siri) illustrate that social embodiment is dynamically constructed through repeated interaction, evolving relationship, and situational affordances. Expert workshops confirm that human-like properties (“has goals,” “has a mind”) are associated with the perception of social embodiment, and that context—industrial vs. home, functional vs. social—modulates this perception.

5. Mechanisms: Sensorimotor Loops, Feedback, and Task Design

A central mechanism in ESPT is the continuous sensorimotor loop, visible in haptic VR experiments (Froese et al., 2014), telepresence robotics (Davat et al., 2023), and embodied social cobots (Nicora et al., 2023). Social presence is constructed through tightly coupled mutual adjustment: movement, haptic feedback, and minimalistic sensory cues allow agents to coordinate and “offload” parts of the cognitive task onto the interactive process itself. For example, models of “social touch” relate vocal signal intensity and affective prosody to perceived distance in mediated communication:

$D_p = f(I, A, X)$

where $I$ is vocal intensity, $A$ is socio-affective attitude, and $X$ encompasses spatial and contextual variables influencing the transmission and perception of embodied presence.

Empirical findings show that congruence between virtual agents and their communicative behaviors (e.g., matching friendly personality to warm gestures in AR) significantly enhances social presence, while incongruity reduces it (Koleva et al., 14 Mar 2024). Even non-humanoid cobot-avatar systems can evoke perceptions of unity and peer-like roles if the behavioral cues are properly orchestrated.

6. Implications and Future Directions

ESPT suggests that social presence is not solely an emergent property of biological brains or physical bodies, but is distributed across agents, interaction protocols, and environments. In VR/AR and social robotics, minimal but carefully designed feedback channels (haptic, gaze, voice) can evoke robust social presence, supporting applications in education, health, therapy, and distributed work environments. Designers should consider the role of both embodiment and language-based socialization, calibrating agent roles and feedback to the context and needs of users.

Open questions remain regarding the amount of cognitive work that can be reliably distributed to the interaction process, optimal structures for turn-taking and mutual responsiveness, and the limits of linguistic socialization in the absence of rich embodiment. Continued research into fine-grained measurement (e.g., TT indices, bio-behavioral coupling), longitudinal studies, and evaluation of adaptive task/role calibration is required to develop standardized frameworks for designing, benchmarking, and understanding embodied social presence in artificial and hybrid human-machine collectives.