From sessions to ecosystems: real-world phenotyping of cognition, affect, and behavior

<p>Controlled laboratory studies in humans have been central to cognitive, affective, and behavioral science because they isolate mechanisms under well-defined conditions. Yet many of the processes we aim to explain are dynamically shaped by context, internal state, repeated experience, and voluntary engagement, features that brief laboratory sessions often cannot capture. Recent advances in smartphones, wearables, ambulatory physiology, emerging neural tools, and naturally occurring digital traces now make it possible to study these processes as they unfold in everyday life. We outline four directions for this shift: (1) redesigning experimental paradigms around naturalistic engagement rather than transplanting session-based tasks; (2) jointly modeling states, traits, and their dynamics; (3) linking real-world phenotypes to neural and psychophysiological mechanisms; and (4) integrating multiple active, passive, subjective, and objective data streams into a multi-level phenotyping ecosystem. We use habits as a guiding example because they are difficult to induce robustly in the laboratory, likely in part because they are shaped in daily life by repetition, context, motivation, and state. We also discuss implications for dynamic clinical phenotyping, real-time risk estimation, just-in-time intervention, and the role of AI-enabled and low-burden measurement.</p>