PROJECT SUMMARY – BEHAVIORAL INTERVENTION DEVELOPMENT CORE The goal of the Behavioral Intervention Core for the Penn State Roybal Center is to develop principle-driven scalable and potent digital intervention components that promote adherence to physical activity and enhance cognitive function following the end of the active intervention period. Drawing on an experimental medicine approach and Stage 0 research, our team identified three validated target mechanisms of adherence to behavior change: automatic affective evaluations, experienced automaticity (habit strength) and identity. The strategies for engaging these reflexive targets differ from those used to engage self-regulatory targets (e.g., goals, plans) because the reflexive targets are grounded in associative learning processes. We will investigate two intervention strategies that can engage these targeted mechanisms of adherence to behavior change: precision messaging and evaluative conditioning. A Community Advisory Board will assist us in refining our prompt engineering architecture for using generative artificial intelligence (AI) to create intervention content that is unbiased, unlikely to hallucinate, and culturally sensitive. Trial 1 (NIH Stage 1) will evaluate the non- inferiority of AI-generated content relative to human-generated content. Trial 2 (NIH Stage 1) will use a within- person experimental design to determine the optimal daily message budget for promoting physical activity in middle-aged and older adults. Trial 3 (NIH Stage 3) will use a factorial experiment to evaluate the main and interactive effects of the two intervention components on targeted mechanisms of adherence to behavior change, physical activity adherence dynamics following the end of intervention, and cognitive function across multiple time scales. These trials will accomplish four scientific aims involving content development, target engagement, adherence dynamics, and cognitive function. We will develop a scalable prompt engineering architecture that enables a transparent and controllable generative AI system to produce personalized principle-based intervention content for middle-aged and older adults. We will develop principle-driven behavioral interventions that engage targeted mechanisms of adherence to behavior change following intervention. We will elucidate factors contributing to behavioral adherence and enhanced cognitive function in the six months following the end of the active intervention period. We will characterize the effects of adherence to physical activity following intervention on cognition across multiple time scales. Collectively, these trials will fill a need for interventions that support long-term adherence to physical activity and enhance changes in cognitive function following the end of active intervention support. In combination with evidence from prospective longitudinal studies linking physical activity with reduced risk for Alzheimer’s Disease and related dementias (AD/AD...