What Embodiment Can and Cannot Explain: Validating the Emcorp Framework on Perceptions of Robots and Virtual Agents

Two decades after the first comparisons of virtual agents and robots, researchers still lack a clear understanding of when and why one form of embodiment is preferred. The Embodiment and Corporeality (EmCorp) framework suggests that embodiment effects are mediated by users’ perceptions of an artificial entity’s bodily-related capabilities, with moderators such as morphology and interaction scenario shaping these perceptions. However, this framework has not yet been empirically tested in interaction studies. In this work, we examine the proposed mediation of perceived capabilities on outcome variables such as entity evaluation, and test two moderators proposed in the EmCorp framework by Hoffmann et al. (2018), conceptually replicating prior studies. First, we test the role of interaction scenario in a 2×2 between-subjects design (interaction scenario: physical task vs. conversational task; embodiment of artificial entity: virtual agent vs. robot; n = 93). Second, we explore the moderating effect of participants’ tendency to anthropomorphize. The results show the expected difference between embodiment conditions for EmCorp perceptions, but not on outcome variables. Participants’ anthropomorphic tendencies positively influenced both entity and interaction evaluation. This pattern underscores the importance of considering individual differences in anthropomorphism when interpreting embodiment effects and suggests that embodiment alone may be insufficient to explain evaluative responses to artificial entities. Our findings validate the EmCorp scale, demonstrating consistent effects in line with theoretical predictions, though additional replication across diverse contexts is needed to further strengthen its explanatory value.