COMPUTATIONAL IRREDUCIBILITY AND THE PHENOMENOLOGY OF FREE WILL

The debate over free will has traditionally been framed as a choice between libertarian indeterminism and hard determinism, with compatibilism attempting to reconcile freedom with deterministic causation. However, this framing overlooks a crucial insight from computational theory: that determinism does not entail predictability. We propose that the phenomenological experience of free will — the felt sense of open alternatives and genuine deliberation — corresponds to a real feature of complex systems: computational irreducibility. Drawing on Wolfram's taxonomy of computational complexity, we argue that human decision-making exhibits Class 4 computational behavior, meaning that outcomes cannot be determined without step-by-step simulation. This creates an epistemic gap between determinism (metaphysical closure) and predictability (epistemic openness) that grounds the phenomenology of agency. We defend this position against four major objections, including concerns about Libet's neurophysiological experiments, and conclude by proposing testable hypotheses for neuroscience and implications for artificial intelligence, moral responsibility, and the philosophy of mind.