Christian List on Free Will, Levels of Reality, and Group Agency

Christian List, philosopher at the Munich Centre for Mathematical Philosophy, defends a non-reductive physicalist account of free will that dissolves the standard clash with determinism by insisting that determinism and indeterminism are level-relative properties. The episode moves from the free will debate through the metaphysics of emergence to group agency and the moral status of AI.

Key ideas

1. Determinism is not a property of the universe simpliciter, but of a level of description. A system can be deterministic at the microphysical level and genuinely indeterministic at the macroscopic level — and List argues this is not mere epistemic unpredictability but ontic, what he calls emergent indeterminism. This makes the internet debate over whether quantum mechanics rescues free will mostly beside the point: what matters is indeterminism at the agential level, which is compatible with determinism at the level of fundamental physics.

2. Free will requires three conditions, not one. List’s checklist: (i) intentional agency — the system must be interpretable as a goal-directed agent; (ii) alternative agential possibilities — genuine choice options must be open at the level of agency; (iii) causal control — the agent’s mental states must be the difference-making variables causally responsible for the action, not a mere reflex or underlying physical process. Indeterminism alone fails the test; consciousness is not required.

3. Supervenience does not imply reducibility — and combinatorics shows why. Even granting that all high-level facts supervene on microphysical facts, explanatory reduction fails for a principled reason: any countable description language can pick out only countably many properties, while the set of possible macro-level properties (each being a subset of microstates) is uncountably large. The probability that explanatorily salient high-level properties happen to be expressible in the microphysical vocabulary approaches zero.

4. The indispensability test replaces Dennett’s instrumentalism. Dennett holds that a system is an agent if it is usefully interpretable from the intentional stance. List goes further: a system is a genuine intentional agent if the intentional stance is explanatorily indispensable — if there is no adequate alternative description. This makes agenthood an ontic rather than purely epistemic matter, and provides a principled criterion for deciding when AI systems qualify.

5. Free will extends to group agents and, potentially, to AI. Corporations, states, and well-organised collectives can satisfy all three conditions; List and Philip Pettit have argued for a realist theory of group agency. Free will, severed from consciousness, is a functional notion portable across substrates — which is why it might apply to sophisticated AI systems, and why that finding matters: if group agents and AI have free will, they are candidates for moral responsibility, not just causal explanation.

Related

• Christian List — speaker
• Sean Carroll — host
• Free Will — existing concept page; this episode adds List’s three-condition framework and the levels-of-description reframe
• Levels of Explanation — the levels architecture undergirds List’s whole argument
• Materialism — List is a non-reductive physicalist; his view on consciousness diverges from his view on agency
• Consciousness — List distinguishes free will (third-personal, tractable) from consciousness (first-personal, hard problem)