Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Cognitive behavioral therapy (CBT) is a cornerstone of evidence-based psychological treatment, but its implementation continues to be limited by poor patient adherence and shortages of clinical resources. At the same time, most applications of artificial intelligence (AI) and games in mental health have focused on direct intervention, although the evidence base remains underdeveloped and AI has important limitations in diagnosis and treatment decisions. In response, this article proposes a Wizard-of-Oz (WoZ) framework for adjunctive treatment that positions AI and board games as tools that support conventional CBT rather than replace it. The framework draws on the mechanism of recontextualization in acceptance and commitment therapy (ACT). By introducing an alternative functional context, recontextualization reduces the control that verbal relations exert over behavior and helps patients shift from cognitive fusion (“I am my thoughts”) to an observer perspective (“I am noticing my thoughts”). The physical interaction and rule system of a board game create an embodied magic circle in which cognitive defusion and cognitive preparation can be practiced in a low-pressure, reversible, and symbolic environment. The AI module remains an information curator and prudent advisor: it organizes structured records and produces descriptive summaries of game behavior and experience logs to assist, rather than determine, the therapist’s clinical judgment. In the proposed implementation, physical board actions are entered through a constrained digital companion interface; the architecture does not assume RFID, sensor-augmented boards, computer vision, or other automatic sensing. The framework uses an asynchronous Wizard-in-the-Loop design. The therapist does not need to intervene during play but reviews structured, AI-assisted summaries in the background and remains outside the patient’s immediate view. The digital companion provides rule prompts and a controlled form for recording board choices, dice outcomes, piece movements, and voluntary reflections. This arrangement preserves human clinical judgment while reducing patient defensiveness. The game design minimizes external rewards such as points and leaderboards and instead emphasizes the deeper experiences created by feedback and challenge. It also uses three types of metaphor—maps, role-play, and monsters—to integrate embodied cognition with the learning of CBT skills. At the ethical level, the framework addresses concerns about emotional dependence, cultural bias, privacy, and data security through prior informed consent, data minimization, de-identified presentation, and patient control over data. Its main contribution is a cautious and practical design strategy that does not depend on the still-debated effectiveness of direct AI intervention. Instead, it uses AI and games to reduce peripheral barriers and prepare patients for therapist-led CBT, with the aim of improving adherence, making better use of limited resources, and reducing treatment resistance.