Between Neurons and Demons: Functional Free Will in a Deterministic World

This paper argues that determinism is compatible with a functional form of free will: the capacity to select actions through criteria that one's own history of activity has formed and continues to revise. First, drawing on the limits of measurement, Maxwell's demon as a model of embedded control, and the decoupling of scales, it distinguishes determinism from predictive completeness and sets out the limits that bind any observer embedded in the world it measures. Second, a formal model separates the actions a physical setting makes feasible from the historically formed criteria through which an agent selects among them, shows how the agent's selections feed back into those criteria, and states the condition under which agents with different histories act differently in identical present conditions, which a minimal deterministic simulation satisfies. The model formalizes Tse's criterial causation without his libertarian commitments. Together with the limits on embedded observers, it shows why the explanation of an action stops at the agent: history bears on the action only through the agent's criteria, and in real cases it cannot be recovered. Third, neuroscience shows how criteria are realized in mixed-selectivity coding, latent working memory, and synaptic plasticity, and, following Malabou, how learning, trauma, and social life form, deform, and sometimes destroy them. Fourth, niche construction shows how agents help build the causal structures within which they act. The result is an ecological compatibilism in which agency consists in the ways an agent's history, partly of its own making, shapes what it does.