From Cognition to Embodiment to Partnership: A Developmental Framework for Human–AI Educational Relationships

This paper proposes a developmental account of the human–AI educational relationship organised around three phases that unfold in a hypothesised developmental order under specified boundary conditions. In the first, the learner’s brain comes to model the AI as an intentional agent rather than a deterministic tool. In the second, this cognitive shift is registered and validated through interoceptive and affective channels. In the third, learner and AI begin to function as a coupled learning unit in which the learner’s predictive system extends to incorporate the AI as a stable partner, progressively reducing prediction error across cognitive and interoceptive channels. Partnership, so defined, requires three minimum conditions: an intentional model of the AI, a stable interoceptive dimension to that model, and persistence across interactions. The account generates a distinctive empirical prediction: a temporal ordering of neural, physiological, and reflective markers across sustained interaction. It further yields three design principles for partnership-oriented educational technologies, identifies the risks that the same mechanisms render newly salient and outlines a three-stage research programme through which the predictions can be tested.