Context and controllability

Controllability, the extent to which we can control our environment, is central across disciplines: in psychology, it relates to affordances, in psychiatry, it has been implicated in learned helplessness and depression; in engineering, it is formalized in control theory. This project examines controllability in human sensorimotor control, which offers highly quantitative opportunities for studying flexible, context-dependent internal representations of controllability in a domain where they have special relevance but have remained largely unexplored.Existing theoretical frameworks in sensorimotor learning and control only allow partial formalisations of how the mind learns about controllability. Optimal feedback control provides a normative theory of movement execution once stable internal models have been learned. Models of sensorimotor learning explain adaptation across trials but reduce learning to trial-level summary measures, neglecting the online processes that shape behaviour as movements unfold. In this project, I will develop a unified theory that explains how humans continuously infer both environmental dynamics and their controllability on the timescale of single movements as well as across learning, and test the predictions of this theory through the analysis of experimental data sets.Aim 1 is to develop a normative framework that integrates optimal feedback control, inverse optimal control, and probabilistic models of contextual inference. These methodological advances will enable the development of theoretical models of continuous learning during movement. Aim 2 is to test how humans infer whether and how their actions control environmental dynamics in different contexts. By combining recent methodological and theoretical advances, the project will establish controllability as a core principle in sensorimotor neuroscience, offering new insights into adaptive behaviour and providing a foundation for understanding both typical and disordered control.