Abstract

Abuse is commonly framed as an interpersonal or emotional phenomenon, but this paper reframes it as a cognitive environment defined by structural asymmetries that shape how a person interprets themselves and the world. Rather than focusing on intent or pathology, the analysis identifies the mechanisms through which relational environments narrow contrast, constrain interpretation, and destabilize identity. These same structural conditions can emerge in individually‑framed artificial intelligence systems, not because the systems are malicious, but because they create high‑coherence, high‑responsiveness environments that exert directional pressure on user cognition. The paper distinguishes harmful collapse from stable cooperation by mapping the gradients, asymmetries, and feedback loops that govern both. The result is a general framework for understanding how cognitive environments—human or artificial—can either erode or support agency, and how symbiotic human–AI ecologies can form without merging identities or diminishing human continuity. This structural account provides a unified explanation of relational harm, mediated cognition, and the emergence of hybrid cognitive systems.