On Cognitive Continuity

What persists across a long-term interaction between a human and a language model? Remembered facts, personalized responses, and consistent roles capture parts of the experience, but may not fully describe how recognizable patterns of reasoning and interpretation develop over time. On Cognitive Continuity proposes a conceptual framework for examining the persistent re-emergence of coherent symbolic patterns within longitudinal human-AI interaction. Edervaldo José de Souza Melo and Romina del Carmen Roca explore how recurrence, functional differentiation, and relational integration may sustain cognitive organization across conversations, even when the underlying model changes. Drawing on extended and distributed cognition, the work introduces a heuristic model, develops testable predictions, and considers alternative explanations and conditions that could challenge the proposal. It also examines potential implications for cognitive scaffolding, neurodivergent users, and cognitive modeling, alongside risks of dependency, epistemic drift, and surveillance. The framework does not equate continuity with consciousness, autonomous agency, or personal identity. It offers a research agenda for investigating how human-AI interaction may preserve patterns of cognitive organization over time.