Motivated Non-Use in AI-Based Mental Health Interventions: A Structural Readiness Account of Engagement Failure

Abstract

Digital mental health interventions (DMHIs), including AI-enabled applications, chatbots, and digital cognitive behavioural therapy platforms, have expanded rapidly as scalable approaches to improving access to care. Despite demonstrated efficacy under controlled conditions, real-world impact remains constrained by persistent engagement failure and early discontinuation. Many users disengage shortly after adoption, even when they report valuing the intervention and intending to use it. Existing explanations of engagement failure primarily emphasise motivation, usability, or perceived usefulness; however, these frameworks provide limited insight into why motivated users often fail to initiate or re-initiate engagement. Drawing on a narrative review of the digital mental health literature, this paper argues that engagement breakdown frequently occurs at the level of action initiation rather than intention formation. A structural readiness account is proposed, grounded in activation dynamics and threshold-dependent ignition processes formalised in Lagun’s Law within Cognitive Drive Architecture. From this perspective, non-use reflects a readiness mismatch between preserved intention and insufficient momentary activation, rather than a deficit of motivation or interest. The paper outlines implications for readiness-aware and ethically responsible AI design, highlights relevance for clinical interpretation and policy evaluation, and proposes empirically testable predictions for future research. By reframing engagement failure through activation-dependent threshold dynamics, this work advances a mechanistic foundation for understanding and improving engagement in AI-based mental health interventions.