From Extended Minds to Coupling Flexibility: Cognitive Eco-Sourcing and Generative AI in Education

Abstract

Debates about generative artificial intelligence (AI) in education often oppose enhancement to diminishment. This framing obscures important differences in how learners recruit AI and how cognitive labour is distributed between agents and technological resources. We introduce cognitive eco-sourcing as a framework for analysing these relations. It distinguishes four non-parallel and potentially overlapping forms of cognitive coupling: offloading, which reorganises cognitive demands; uploading, through which constructed niches come to supply stable cognitive functions across agents and time; delegation, in which an external resource executes an operation while the agent remains substantively within its production loop; and outsourcing, in which external execution occurs without effective supervision or authorship. These forms are analysed through four dimensions: integration, execution, supervision, and authorship. Education makes these distinctions developmentally significant because activities are intended not merely to produce outcomes but to cultivate capacities. We claim that no form of cognitive eco-sourcing is intrinsically beneficial or harmful. Its educational value depends on the function involved, the target capacity, the learner’s competence, the organisation of the coupling, and its trajectory over time. We conclude that educational environments should cultivate coupling flexibility: learners’ ability to understand, evaluate, select, and reorganise relations with technologies in accordance with changing epistemic and educational aims.