Proceedings of the Annual Meeting of the Cognitive Science Society

In contemporary consciousness studies, the sense of body ownership (SBO) stands as a key marker of subjective embodiment and self-awareness. Recent progress in multimodal and agent AI has prompted the question: Could an artificial system develop something analogous to SBO, and would this require consciousness? This paper refines two core distinctions: (1) functional versus phenomenal SBO, (2) world‑model versus self‑model. Building on a functional reconceptualization of "body" as an interaction boundary, we argue that AI systems equipped with semantic-centric multimodal world models and complementary self-models can, in principle, instantiate a form of SBO. By integrating diverse sensory inputs—visual, tactile, and linguistic—into a cohesive self-representation, this approach suggests the possibility of a virtual body that evokes the contours of the human embodied experience. Such an account questions the strict divide between physical and virtual embodiment, offering new insights into how embodied cognition underpins consciousness. Also, we locate SBO within the broader debate on AI consciousness. Concrete design proposals are linked to existing multimodal agents (e.g., DeepMind's MIA). This inquiry highlights how AI SBO may arise from the interplay of sensory and semantic frameworks, prompting ethical and theoretical reflection on AI consciousness.