The setup was almost insultingly simple. A subject sits in front of a modified oscilloscope on which a bright dot sweeps around a circular face like the second hand of a clock, completing a revolution every 2.56 seconds. The instruction: whenever you feel like it, flick your wrist. There is no cue, no signal, no reason to move at one instant rather than the next. It should be the purest possible act of will — a decision arising from nothing but you. And when you flick, note the position of the dot at the exact moment you first became aware of the urge to move. Meanwhile, electrodes on your scalp record the electrical activity of your brain, and a sensor on your wrist marks the muscle's firing to the millisecond.
Benjamin Libet, a physiologist at the University of California, San Francisco, ran this experiment in the early 1980s, and the result he published in 1983 has haunted the philosophy of mind ever since. The brain's preparation to move — a slow negative shift in scalp voltage called the readiness potential, or Bereitschaftspotential, discovered by Kornhuber and Deecke in 1965 — began around 550 milliseconds before the muscle moved. That was expected. What was not expected was the timing of the conscious decision. Subjects reported becoming aware of their intention to move only about 200 milliseconds before the movement. The brain, in other words, had started building the action roughly a third of a second before the person consciously decided to do it. The felt moment of choosing was not the cause of the action. It arrived late, after the work had already begun.
Read at face value, Libet's data say something almost unbearable: the conscious will is not the author of our actions but a witness to them, informed after the fact and mistaking its report for a command. The decision to move is made by unconscious neural machinery; consciousness is handed a press release and told it wrote the policy. If that is true of flicking a wrist, the worry runs, it may be true of everything — the job you took, the person you married, the words in this sentence — each one initiated in the dark and narrated in the light.
Libet himself did not accept the bleak conclusion. He noticed that a gap of about 200 milliseconds still separated the moment of conscious awareness from the muscle's firing — and since the final ~50 milliseconds are given over to an irreversible motor command, that left a usable window of roughly 100 to 150 milliseconds in which consciousness could still cancel the action. We may not have free will, he argued, but we have "free won't": a conscious veto over impulses the brain generates on its own. The initiation is unconscious; the last-second refusal is where freedom lives. It is a small door, but Libet insisted it was a real one. Critics pointed out that no direct evidence for the veto was ever produced, and that a veto decision should itself be preceded by its own readiness potential — pushing the problem back a step rather than solving it.
If Libet cracked the door, the neuroscientist John-Dylan Haynes appeared to kick it off its hinges. In a 2008 study in Nature Neuroscience, Haynes and colleagues at the Max Planck Institute in Leipzig put subjects in an fMRI scanner and asked them to press one of two buttons — left or right — whenever they chose, and to note when they decided. Using pattern-recognition on the brain images, the researchers found that the outcome of the choice, which button, could be decoded from activity in the frontopolar cortex (Brodmann area 10) and the precuneus up to seven to ten seconds before the subject reported consciously deciding — with accuracy above chance, around 60 percent.
Seven seconds is an eternity in neural time. A decision the subject experienced as spontaneous and immediate was, on this reading, quietly forming in the cortex long before it surfaced into awareness — long enough that the conscious "choice" looks like the last domino in a chain that started before the chooser knew a choice was being made. Between Libet and Haynes, the empirical case seemed to close: Consciousness does not initiate; it reports.
The tidy story has a serious problem, and it came from inside neuroscience. In 2012 Aaron Schurger, working with Jacobo Sitt and Stanislas Dehaene, published a reinterpretation of the readiness potential in PNAS that quietly reopened the whole question. Their model treats the brain, in the boring interval before a spontaneous movement, as a leaky stochastic accumulator: neural activity fluctuates randomly around a baseline, and a movement is triggered whenever those fluctuations happen to cross a threshold. Crucially, the readiness potential as Libet measured it is an artifact of averaging. Because you can only line up the trials by working backward from the moment of movement, you are guaranteed to find a rising ramp beforehand — you have selected for exactly the upward noise excursions that caused the threshold crossing. The RP, on this account, is not a decision being made. It is the shape of random noise, viewed in a mirror that only reflects the moments the noise won.
If Schurger is right, Libet's ramp does not show the brain deciding before you do. It shows that the timing of a truly arbitrary movement is partly random — which is not the same as showing that your considered choices are unconsciously predetermined. And the deeper methodological critiques bite too: pressing a button "whenever you feel like it" is a deliberately meaningless act, nothing like deciding to quit a job or forgive a friend, and the introspective report of "W-time" — noting when you became aware of an urge, by reading a spinning dot — is exactly the kind of self-observation psychology has spent a century showing to be unreliable. The experiments may be measuring the noise of a bored brain, not the machinery of the will.
The neuroscience is only the latest skirmish in a war that is millennia old, and stripped of the electrodes the structure is stark. If the universe is governed by physical law, then the state of your brain at this moment is the strict consequence of its state a moment ago, and so on back before you were born. Pierre-Simon Laplace made the point unanswerable in 1814: an intellect that knew the position and momentum of every particle could calculate the entire future, including every "choice" you will ever make. Determinism does not need Libet. It follows from taking Materialism seriously — from the causal closure of the physical world, the principle that every physical event has a sufficient physical cause. Quantum indeterminacy does not obviously rescue freedom; a choice made by a random quantum event is no more yours than one made by clockwork. Randomness is not agency.
This is the vise that has held the problem shut for centuries. Either your decisions are determined by prior causes you did not choose — in which case in what sense are they free? — or they are undetermined, injected with randomness — in which case they are not decisions at all. Libertarian free will, the everyday belief that you could have done otherwise holding the entire past fixed, seems to fall through the gap between these two, with nowhere physical to stand. The philosopher Galen Strawson pressed this into his "basic argument": to be truly responsible for an action you would have to be responsible for the character it flows from, and for the earlier character that shaped that, in an infinite regress that no finite being can complete.
Almost no one lives as if this were settled, and the escape routes are old and ingenious. The largest is compatibilism — the claim, running from Hume to Daniel Daniel Dennett & Modern Materialism, that freedom and determinism were never really enemies. On this view "free" does not mean "uncaused"; it means acting from your own desires and reasons without external coercion. A determined choice that flows from your character, deliberation, and values is free in every sense that matters, and the libertarian's uncaused will is not only impossible but undesirable — a choice disconnected from who you are is not freedom but caprice.
Dennett argues that Libet's experiments attack a cartoon of free will nobody should have believed in, a little conscious homunculus that must fire the first shot; the freedom worth wanting — the capacity to be moved by reasons, to foresee and avoid, to be the kind of agent that responds to persuasion and blame — evolves in a determined world and is perfectly real.
The Stoics got there first. Facing a cosmos they believed to be entirely fated, they relocated freedom inward, to assent. Chrysippus offered the image of a cylinder: push it and it rolls, but how it rolls is a function of its own cylindrical nature. External causes start the motion; your nature shapes the response; and in that shaping — the disciplined agreement or refusal of the mind — the stoicism|Stoic found all the freedom a human being needs or can have.
Kant & Transcendental Idealism built the most ambitious escape of all, splitting reality in two: in the phenomenal world of appearances everything, including your body and brain, is rigidly determined; but the noumenal self, the thing-in-itself outside space and time, is the seat of a transcendental freedom that grounds the moral law. We are determined as objects of science and free as moral agents, and both are true because they describe different worlds.
Against all of them stand the hard determinists, more confident now than in a century. Sam Harris, in his 2012 book Free Will, argues that the whole notion is incoherent — that when you watch your own thoughts closely you cannot find the author, only thoughts arising on their own, exactly as the The Illusion of the Self thesis predicts.
Robert Sapolsky, in Determined (2023), marshals the biology — genes, hormones, childhood, culture, the state of your blood sugar this second — to argue that there is no crack anywhere in the causal wall through which a free agent could reach in and choose. And here the stakes stop being academic. If no one is the ultimate author of their actions, the entire architecture of desert — of blame, of punishment, of hatred toward the guilty — rests on a mistake, and a criminal-justice system built on retribution rather than protection and repair is punishing people for the crime of being exactly what physics made them. That is the reason the argument will not die. It is not really about a flicking wrist. It is about whether anyone has ever deserved anything.
This is no longer confined to seminar rooms. In American and European courts a field called neurolaw has grown up around exactly the question Libet poked at: if an action can be traced to a tumor, a lesion, a malformed circuit, where does responsibility go? In 1991 Herbert Weinstein strangled his wife and threw her from a Manhattan window; his defense produced a PET scan showing a large arachnoid cyst compressing his frontal lobe, and prosecutors, fearing the image would sway the jury, accepted a reduced plea rather than let it reach them. The picture of a damaged brain is persuasive in a way the argument alone is not — and the uncomfortable question is why. Every violent act traces to some brain state; the tumor is only a brain difference we can see. If a visible cyst mitigates guilt, consistency demands that the invisible genetics, trauma, and neurochemistry Sapolsky catalogues should mitigate it too — which is either a reductio against the whole framework of desert, or a reason to rebuild justice around prevention and repair rather than retribution. Neuroscience did not create this dilemma. Libet's clock only made it impossible to look away from: the machinery that produces a choice is always, in the end, machinery, and the law has never decided how much of a person that leaves standing.