Pietro Cipresso
Human behavior is typically measured by relating observed responses to latent traits inferred from statistical models. Although this approach has produced powerful tools for psychological assessment, it largely treats behavior as a static property of individuals rather than as the observable output of an evolving dynamical system. Yet most behavioral phenomena emerge from continuous interactions between internal states and environmental conditions, generating trajectories of actions, decisions, and physiological responses over time. Here we propose a framework in which human behavior is measured by reconstructing the dynamical systems that generate it through controlled perturbations. In this perspective, experimental environments act as measurement instruments that probe behavioral dynamics through structured perturbations. Behavioral observations are interpreted as the outputs of stochastic controlled dynamical systems, and behavioral measurement becomes a system identification problem: generative models are reconstructed so as to reproduce the trajectory distributions produced by human behavior under equivalent perturbation conditions. We formalize this framework using stochastic dynamical systems, introduce criteria for reconstruction and functional equivalence under unseen perturbations, and analyze the roles of perturbation design, identifiability, and experimental domain boundaries. The approach is illustrated across multiple behavioral domains, including affective dynamics, cognitive flexibility, stress reactivity, and social interaction. By integrating controlled perturbations, dynamical modeling, and trajectory-level validation, this framework reframes behavioral measurement as the reconstruction of generative behavioral dynamics. More broadly, it suggests a path toward a metrology of human behavior in which programmable experimental environments, such as immersive virtual reality platforms, serve as instruments capable of systematically probing and characterizing behavioral systems.