Free will is the old question of whether we genuinely choose: whether we can select among alternatives, control our behavior, and originate actions rather than merely be pushed by prior causes. Sam Harris argued in Free Will that it is an illusion, since decisions are fully caused by neural processes and background factors outside conscious control. Daniel Dennett, the leading defender of compatibilism, answered in a review that free will means control, responsiveness to reasons, and self-regulation, not a contra-causal power, and Harris replied that this changes the subject and sidesteps ultimate authorship.
I found the tone of Dennett’s side of that exchange disappointing, and it is one reason the Discourse Protocol exists: disagreement is inevitable, and progress depends on how it is conducted.
On the substance, three layers help. The metaphysical question is whether choices are determined by prior states, including neural events. The psychological question is how deliberation actually happens. The normative question is whether determined agents should be held accountable. Much of the debate turns on Benjamin Libet’s experiments, where a readiness potential precedes a voluntary act. Harris reads that as decisions being made before we know it. Later modeling by Aaron Schurger and colleagues, summarized under the neuroscience of free will, shows the potential is not necessarily causal, and Dennett adds that the conscious author of an act needs no access to how the brain produces it, much as a sailor steers without controlling wind or current.
The real divergence is degrees. Dennett treats freedom as a matter of more or less control; Harris treats it as all or nothing and concludes that no one controls anything. When one side means the ability to do otherwise under identical conditions and the other means being guided by one’s own reasons, positions that are compatible look opposed. Defining terms and separating the empirical from the conceptual dissolves much of the fight, which matters because responsibility, law, and self-conception all rest on it. My open question: is the demand for ultimate authorship a legitimate requirement or a conceptual error?