A theoretical review of the Proteus effect: understanding the underlying processes

The paper introduces humans’ longstanding attraction to stories and how immersive virtual reality (IVR) enables new forms of narrative engagement by letting users embody avatars. Prior studies show users adapt behaviors to virtual bodies (e.g., fear responses to virtual danger; adaptation to extra limbs), prompting investigation into how avatar appearance shapes behavior—the Proteus effect. The authors set the goal to critically review and synthesize theoretical frameworks that might explain the Proteus effect’s underlying social, affective, and cognitive mechanisms, outline gaps in current literature, and propose directions for future research. The review is structured to summarize evidence for the effect, evaluate existing theories (self-perception, deindividuation, priming), introduce additional hypotheses (cognitive dissonance, embodiment, perspective-taking), and argue for a combined explanatory framework.

The review first recounts the seminal publication (Yee & Bailenson, 2007) showing that avatar attractiveness increased self-disclosure and proximity, while avatar height increased assertiveness in negotiations, with confederates blind to manipulations, indicating users’ behavior aligned with avatar-congruent stereotypes. Subsequent literature demonstrates the Proteus effect across diverse avatar features: bodily traits (gender, height, attractiveness, age, ethnicity, weight, structural changes such as extra fingers or tails) and character-type embodiments (famous figures like Einstein and Leonardo da Vinci, uniforms such as KKK vs nurse/doctor, and non-human or omnipotent entities like God or aliens). Findings span cognition (improved performance and creativity), affect (aggressive intentions), motor/physiological changes (walking speed, heart rate), and psychological states (confidence, creativity). Contextual and technical factors (avatar-environment congruence, perspective, display type, graphics/context cues) also modulate user responses. Two meta-analyses report small-to-medium average effect sizes around r = 0.24, supporting robustness. However, theoretical explanations lag: much work demonstrates effects without probing mechanisms. Prior reviews either focus on game contexts (avatar characteristics influencing in-game and post-game attitudes) without theoretical challenge, or classify studies via identification processes but still favor self-perception and priming. The present review aims to reassess these theories, highlight limitations, and consider alternative frameworks.

This is a narrative/theoretical review and critical analysis rather than a systematic review with predefined inclusion criteria. The authors synthesize evidence from foundational experiments, replications, and meta-analyses and then conceptually evaluate six theoretical hypotheses potentially explaining the Proteus effect: H1 self-perception theory (SPT), H2 deindividuation (SIDE model), H3 priming (automatic perception-behavior links), H4 cognitive dissonance (including predictive processing accounts), H5 embodiment (sense of agency, self-location, body ownership), and H6 perspective-taking (empathy induction via VR). They compare each hypothesis against empirical demonstrations (e.g., mirror vs playback embodiment manipulation; double-blind replications), note scope conditions and contradictions, and propose an integrated explanatory framework. No formal database search strategy or quantitative synthesis is reported.

• The Proteus effect is robust across contexts and modalities, affecting cognition, affect, behavior, physiology, and psychological constructs; meta-analyses report small-to-medium effect sizes (approximately r = 0.24).
• Foundational results: attractive avatars increase proximity and self-disclosure; taller avatars yield more assertive negotiation behavior; embodiment of eminent or creative figures (Einstein, da Vinci) improves cognitive/creative performance; elderly embodiment slows post-exposure walking; non-human or omnipotent avatars reduce threat responses and increase confidence.
• Embodiment matters: seeing oneself in a mirror (embodiment) versus watching a playback distinguishes being from seeing; Proteus effects emerge under embodiment but not mere observation.
• Self-perception theory (H1) is widely cited but has conceptual issues when treating avatar appearance as an “external cue” and is difficult to falsify; mediation by explicit avatar-trait perception is not consistently supported.
• Deindividuation (H2) plausibly explains increased conformity to avatar-evoked norms in VR due to anonymity, identity cue replacement, and group immersion, aligning with SIDE; may account for stronger effects in immersive VR than on PC displays.
• Priming (H3) alone is insufficient; behavioral priming’s replicability is weak. However, mechanisms like spreading activation likely contribute to concept activation without fully explaining behavior change; embodiment confers self-relevance lacking in simple primes.
• Cognitive dissonance (H4) offers an alternative to SPT: users may adjust attitudes/behavior to maintain coherence with avatar appearance, potentially framed via predictive processing/predictive dissonance.
• Embodiment (H5) as moderator is inconsistent across studies; some link stronger body ownership to stronger effects, others find no moderation. Identification processes may interact with embodiment, potentially compensating when embodiment is low.
• Perspective-taking (H6) often predicts anti-stereotypical outcomes via empathy (e.g., reduced bias when embodying marginalized avatars), sometimes opposing Proteus predictions; thus it cannot account for much of the Proteus effect literature.
• Proposed framework: include deindividuation and cognitive dissonance as core mechanisms, with embodiment influencing effect strength; reject SPT and priming as full explanations (though priming-related activation may play a role).

The review addresses the central question of what mechanisms underlie the Proteus effect by weighing six theoretical accounts against empirical findings. It argues that avatar embodiment shifts self-relevance and can deindividuate users, redirecting behavior toward situational norms cued by avatar identity. Cognitive dissonance provides a coherent motivation for aligning one’s attitudes and actions with avatar-consistent expectations to reduce dissonance. Priming’s contribution is reframed as concept activation through spreading activation, but embodiment uniquely binds these concepts to the self, enabling behavioral change. The mirror-versus-playback finding reinforces that embodiment, not mere exposure, is critical. Mixed results on embodiment’s moderating role suggest additional factors (e.g., identification) and methodological differences, motivating nuanced future tests. The perspective-taking literature, while valuable for empathy interventions, often predicts outcomes opposite to Proteus and appears to represent a distinct phenomenon. Overall, a combined framework—deindividuation increasing susceptibility to avatar-cued norms, dissonance reduction driving congruent adjustment, and embodiment strengthening linkage between self and avatar—better aligns with available evidence than single-theory accounts like SPT or priming alone.

This theoretical review consolidates evidence that avatar appearance can shape users’ behaviors, cognitions, and affect, and critically evaluates competing explanations. The authors propose an integrated account in which VR-induced deindividuation and a drive to reduce cognitive dissonance, together with the sense of embodiment, underpin the Proteus effect. They reject self-perception theory and priming as sole explanations, while acknowledging priming-related activation may contribute. The paper calls for targeted empirical tests of deindividuation and dissonance mechanisms in Proteus paradigms, systematic measurement of embodiment and identification, manipulation of identity salience, and clearer differentiation between Proteus and perspective-taking interventions. Applications include leveraging avatar design to foster desirable behaviors and skills (e.g., creativity) in professional contexts.

• The review is narrative and conceptual without a reported systematic search strategy or quantitative synthesis.
• Key proposed mechanisms (deindividuation and cognitive dissonance) have not yet been directly tested within Proteus effect experiments.
• Evidence on embodiment’s moderating role is mixed and may be confounded by unmeasured factors such as identification.
• Some demonstrations involve sensitive stereotypes (e.g., race, gender, body size), raising ethical and generalizability concerns; framing and context can influence whether effects align with stereotypes or empathy-based counter-stereotyping.
• Distinguishing Proteus effects from related phenomena (e.g., perspective-taking) requires more precise methodological separation and measurement.