The Proteus Effect in Immersive VR: A Systematic Review and Meta-Analysis of Psychological and Motor Outcomes

The Proteus Effect — the phenomenon by which individuals adapt their behavior to match the visual characteristics and stereotypes associated with their virtual avatar — represents a compelling intersection between digital identity and embodied cognition. In immersive virtual reality (VR), where users inhabit avatars whose appearance can be precisely manipulated, this effect raises a question of both theoretical and clinical significance: can the mere visual properties of a virtual body produce measurable changes in how the human motor system plans and executes movement? Answering this question has practical stakes. If avatar-induced behavioral shifts extend to the motor domain, they could inform scalable, non-invasive interventions for motor rehabilitation, sports training, and skill acquisition — applications that require a rigorous empirical foundation before clinical translation. To address this gap, the present thesis conducted a PRISMA-compliant systematic review and meta-analysis. Using the PICO framework, the study targeted healthy adults (Population), exposed to immersive virtual environments with manipulated avatar characteristics (Intervention), compared against baseline, neutral, or non-matched avatar conditions (Comparison), to assess immediate motor execution and long-term skill acquisition (Outcomes). From a synthesized dataset of 33 studies (n = 33), a random-effects meta-analysis using Restricted Maximum Likelihood (REML) estimation yielded large and statistically significant pooled effect sizes across both motor domains: Gross Motor: d = 0.8133, p < 0.0001 and Fine Motor: d = 0.8437, p < 0.0001. These findings indicate that modifying an avatar’s visual traits — its physique, age, or limb configuration — translates into measurable changes in real motor behavior, across tasks as diverse as whole-body locomotion, balance, and manual dexterity. Crucially, the effect is not confined to the moment of VR exposure: the qualitative synthesis shows that approximately 30% of the literature tracks Long-Term Outcomes (retention and transfer), suggesting that avatar-induced behavioral shifts may contribute to durable motor learning rather than transient online adjustments. A complementary meta-analysis on virtual Embodiment — operationalized as ownership, agency, and self-location — yielded an even larger pooled effect (d = 1.33, p < 0.0001), confirming that avatar manipulations strongly engage the sense of embodiment. Qualitatively, 60.61% of the reviewed literature reported embodiment-related outcomes alongside motor measures, underscoring its theoretical centrality. Whether embodiment acts as a psychological precondition for the Proteus Effect to produce motor changes — as the literature broadly assumes — remains an open empirical question, as formal mediation was not tested in the present study. Substantial statistical heterogeneity across all three meta-analytic axes reflects the methodological diversity of an emerging field — variation in hardware, avatar manipulation type, task design, and outcome measurement — and calls for caution in generalizing the pooled estimates. These findings nonetheless establish a quantitative baseline from which future research can build standardized protocols and, ultimately, evidence-based clinical applications in neuromotor rehabilitation, chronic pain management, and sports performance training.