The Half-Second Before the Choice: Libet, the Readiness Potential, and the Neuroscience of Free Will

The Hook: A Signal That Arrives Before You Do

Picture yourself in a laboratory. In front of you a clock ticks with a fast-moving spot of light that completes a full revolution in under three seconds. Your only task: at some point, entirely as you please, to flex your wrist. No cue, no compulsion, no correct moment—you simply wait until you feel the urge, and then you act. And you are to remember where the spot of light stood the instant you first noticed that urge.

It feels completely free. You wait, feel nothing, wait some more, and then—now—you move. The impulse came out of nowhere, out of yourself, from that inner place we call "I." That is how you experience it.

But electrodes are glued to your scalp, and they tell a different story. They show that roughly half a second before you believe you formed the decision, your brain had already begun preparing the movement. A slowly swelling electrical signal built up while you still thought you were waiting undecided. By the time your consciousness announced the decision, the train had long since left the station.

This result, published in 1983 by the neurophysiologist Benjamin Libet and his team, has become one of the most-discussed findings in all of cognitive science. To some it is the experimental death blow to free will: if the brain has already decided before "I" notice it, then my conscious will is merely a spectator that mistakes itself for the director. To others it is an ingeniously constructed misunderstanding about what such a brain signal even means. This article follows the trail of that half-second—through the intensifications delivered by modern imaging, through an elegant statistical counter-reading, all the way to the ancient philosophical question of whether we are the authors of our deeds or merely their narrators.

Part 1: The Question Behind the Experiment

Three Camps, One Problem

Before we enter the laboratory, a map of positions is worthwhile, because without it one cannot grasp what Libet's numbers actually threaten to shake.

The classic dispute runs: is free will compatible with determinism—the view that every state of the world (including every brain state) is fully fixed by the preceding states plus the laws of nature? From that question spring three great camps.

The libertarians (in the metaphysical, not the political sense) say: genuine freedom requires that under identical conditions I could also have acted otherwise. Therefore determinism must be false for the relevant decisions—somewhere in the act of will there is a real openness that is not exhaustively fixed by past and laws.

The hard determinists (and their modern relatives, the hard incompatibilists around Derk Pereboom) say: determinism is true in the relevant sense, and therefore the freedom we imagine simply does not exist. Our sense of being an author is a useful illusion.

The compatibilists, finally—from David Hume through Harry Frankfurt to Daniel Dennett—say: the whole question is wrongly posed. Freedom does not mean escaping the laws of nature but acting as one wants to act, without coercion, compulsion, or manipulation. A person who acts out of their own deliberations and desires is free—regardless of whether those desires themselves have causes. Everything has causes; that is not a threat but the precondition for action to lead to results at all.

At first glance Libet's experiment seems to pose a very concrete, empirical question that cuts across this philosophical map: does the conscious will arrive in time to cause the movement—or does it lag behind? Precisely here lies the explosive force and, as we will see, the danger of confusion.

Why, of All Things, a Twitch of the Finger?

A word about the choice of task, for it will later become the pivot of the criticism. Libet needed an action triggered purely endogenously—not as a reaction to a stimulus, but spontaneously from within. Only that way could the temporal origin of the "decision" in the brain be cleanly measured. A spontaneous, meaningless flexion of the wrist is ideal for this: there is no external trigger, no weighing, no pro and con. But this very meaninglessness will plague the later debate. For an arbitrary twitch, in which it is utterly indifferent when one performs it, may be the worst conceivable model for those deliberate, consequential decisions that "free will" is really about.

Part 2: The Libet Experiment in Detail

The Three Time Markers

Libet's ingenious achievement was the simultaneous measurement of three events that had never before been related to one another.

First the movement itself (M), precisely captured via the electromyogram of the muscle—the objective zero point.

Second the readiness potential (in German Bereitschaftspotential, RP), a slowly swelling negative electrical potential over the motor areas of the scalp. This signal was not Libet's discovery: Hans Helmut Kornhuber and Lüder Deecke had already described it in 1965 in Freiburg and given it the German name that remains in use even in the English-language literature to this day. It was regarded as the neural signature of movement preparation.

Third—and this was Libet's actual methodological innovation—the moment of conscious will (W). To measure it, Libet had participants watch that fast clock and remember the position of the spot of light at the moment they first felt the urge or intention to act. After the movement they reported this position, from which W could be reconstructed. Libet carefully corrected this figure for systematic perceptual delays, which he determined in his own control conditions using skin stimuli.

The Disturbing Order

The result, published in the journal Brain, can be summed up in a few numbers. For spontaneous movements without advance planning, the readiness potential began on average about 550 milliseconds before the movement. The conscious will W, by contrast, appeared only about 200 milliseconds before the movement.

Between the onset of neural preparation and the conscious decision, then, there yawned a gap of roughly 350 milliseconds. The brain, on the obvious reading, had already initiated the movement before the person even knew they were about to move. The conscious will was not the trigger but came after the trigger—it merely reported, as it were, what was already underway.

For many this was a scientific sensation with worldview-shaking force. If the neural cause of the action precedes the conscious willing, then that willing cannot have caused the action. The will would be dethroned—a commentator on events, not an author.

Libet's Own Way Out: "Free Won't"

Remarkably, Libet himself did not draw this radical conclusion. He salvaged a narrow but decisive remnant of agency. Between the conscious will at about −200 ms and the final point after which the movement can no longer be motorically stopped (about −50 ms), a window of roughly 100 to 150 milliseconds remains. In this window, Libet argued, consciousness could still abort the already-initiated action.

From this he coined his famous quip: we may have no free will in the sense of consciously triggering, but we do have a free not-willing—a "free won't," a conscious veto power. The role of consciousness would accordingly not be to generate impulses (the brain handles that unconsciously) but to filter them: to approve or block the rising drafts of action. I am of the opinion that this reinterpretation is psychologically apt—it describes well how self-control feels—yet it incurs the same objection: is the veto not also a brain process, which in turn could be preceded by an unconscious "veto readiness potential"? Libet saw this danger of regress and left it open.

Part 3: The Intensification Through Modern Imaging

Libet's EEG could measure only a coarse signal averaged across the scalp. Two later studies pushed the finding, with finer tools, into seemingly even more dramatic numbers.

Soon and Haynes 2008: Seconds Instead of Milliseconds

In 2008 Chun Siong Soon, Marcel Brass, Hans-Jochen Heinze, and John-Dylan Haynes published a functional magnetic resonance (fMRI) study in Nature Neuroscience that caused a stir. The participants were to decide freely whether to press a button with the left or the right hand—so here it was about the content of the decision, not merely its timing.

The result: using pattern recognition, the outcome of the decision could be predicted from brain activity—specifically from the frontopolar cortex (Brodmann area 10) as well as from the precuneus/posterior cingulate—as early as up to seven seconds before the person became aware of their decision. Adding the sluggishness of the BOLD signal, one arrives at a lead time of up to roughly ten seconds.

That sounds devastating. But here scientific honesty is called for, honesty almost always missing in popular accounts: the prediction accuracy lay at only about 60 percent—barely above the chance level of 50 percent. The brain thus showed a weak tendency, not a settled decision. One should picture it less as "the die is already cast" and more as "the board is slightly tilted in one direction." A slight bias is something other than a determination.

Fried 2011: Looking at Single Neurons

Itzhak Fried, with Roy Mukamel and Gabriel Kreiman, went closer still, publishing in Neuron in 2011. In epilepsy patients who, for therapeutic reasons, already had electrodes implanted in the brain, they could record the activity of individual nerve cells in the supplementary motor area (SMA)—an opportunity extremely rare at this resolution in humans.

They found a gradual recruitment of neurons over about 1500 milliseconds before the reported decision. From the activity of roughly 256 neurons the impending movement could be predicted in single trials with over 80 percent accuracy, and that already 700 milliseconds before the person became aware of the decision. Some neurons increased their firing rate, others lowered it—together they formed a rising level of excitation that, upon crossing a threshold, issued in the movement. This observation—a swelling level up to a threshold—is precisely the thread the great counter-reading pulls on.

Part 4: The Counter-Movement—Rethinking the Readiness Potential

Schurger's Stochastic Accumulator (2012)

In 2012 Aaron Schurger, Jacobo Sitt, and Stanislas Dehaene published in the Proceedings of the National Academy of Sciences (PNAS) a paper that elegantly undermined the standard reading of the readiness potential. Their question was not philosophical but soberly methodological: what is the readiness potential actually—and is it really the beginning of a decision?

Their model is a leaky stochastic accumulator. Picture a level that constantly fluctuates up and down due to random neural noise while slowly "leaking," that is, sinking back down of its own accord. In the Libet task there is an implicit imperative to produce a movement "sometime soon." The brain therefore holds the level near the trigger threshold. When exactly the movement occurs is then decided by a random upward fluctuation that crosses the threshold.

The decisive trick concerns the data analysis. The readiness potential is never measured in a single trial but through the averaging of many trials, all aligned backward onto the movement onset. But if you superimpose precisely those moments in which a random upward wave crossed the threshold, then on average a slowly rising curve necessarily emerges before the movement—not because the brain "planned" anything, but because by construction one selects and sums only the ascending flanks. The readiness potential would accordingly be to a large extent an artifact of averaging autocorrelated noise, not a goal-directed preparation.

With this model Schurger's team could reproduce both the characteristic shape of the readiness potential and the distribution of the participants' waiting times. And it made a testable prediction: if you interrupt the waiting subjects at random moments with a tone and ask them to react immediately, they should be faster when their inner level happened to stand high. Precisely this was found.

Why That Rehabilitates Free Will (or at Least Relieves It)

The consequence is profound. If the readiness potential is not a "starting shot" the brain fires 550 ms in advance, but merely the averaged picture of random fluctuations, then the disturbing time delay vanishes. The actual decision then falls only late—at the moment of threshold crossing—and thus lies close to the reported time of conscious will. The conscious will no longer arrives embarrassingly late; it roughly coincides with the threshold crossing. The apparent proof against free will rested on a confusion: one had mistaken the statistical shadow of random fluctuations for the cause of the action.

In 2021 Schurger, Pengbo Hu, Joanna Pak, and Adina Roskies summed up the state of the art in a widely noted review in Trends in Cognitive Sciences: the readiness potential is probably not a marker of an already-fallen, unconscious decision, but reflects general expectation, time perception, and autocorrelated baseline noise. I am of the opinion that this work does not "prove" that we are free—no experiment can—but it withdraws from the most famous neuroscientific refutation of freedom a large part of its foundation.

Part 5: The Methodological Objections at a Glance

Besides the accumulator reading, several independent objections have accumulated over the years that together yield a sobering picture of the original interpretation.

The unreliability of W. The moment of conscious will is elicited via an introspection that is itself error-prone: one must relate two events (the inner urge and the clock position) to one another in time, and the perception of simultaneity is notoriously distorted (think of the "flash-lag" effect). Small systematic errors in W shift the entire argument. The W value reported by Libet is more a rough estimate than a hard measurement.

RP is perhaps merely general preparation. Early on, critics such as Trevena and Miller objected that the readiness potential may indicate only an unspecific activation—an "I'm getting ready to do something in a moment"—not a specific decision to act "now and with this hand." A signal that also appears when one ultimately decides against the movement can hardly be the causal trigger of precisely that movement.

Arbitrary is not the same as deliberate. Perhaps the most important objection concerns transferability. In 2019 Uri Maoz, Gideon Yaffe, Christof Koch, and Liad Mudrik showed, in a careful study (published in eLife), a remarkable difference: for arbitrary decisions (it does not matter which option one chooses) the readiness potential appeared as usual. For deliberate, consequential decisions (for instance, which of two charities receives real money) it was practically undetectable. This suggests that the readiness potential is a phenomenon of the arbitrary, meaningless twitch—and precisely not of the meaningful choice that the freedom debate is about. The Libet task would then have chosen the wrong model from the outset.

The meta-perspective. A meta-analysis of Libet-style experiments (2021 in Neuroscience & Biobehavioral Reviews) confirms the basic existence of a lead in neural activity, but at the same time stresses the considerable scatter of the results and the dependence on task details and analysis methods. The finding is real, but far less unambiguously interpretable than the popular narrative suggests.

Part 6: Back to Philosophy

Even if one sets aside all neuroscientific objections and assumes the brain does initiate actions unconsciously—would free will thereby be refuted? Here it becomes clear that much of the excitement rests on a tacit philosophical assumption.

Compatibilism Defuses the Finding

For the compatibilist, the discovery that conscious decisions are carried by unconscious brain processes is no surprise but a matter of course. What else should decisions consist of? A will that did not arise from neurons would be no natural will but magic. Daniel Dennett put this sharply in Elbow Room (1984) and Freedom Evolves (2003): Libet enthusiasts confuse "my brain did it" with "I did not do it"—as if I were something other than my brain at work. The relevant freedom does not consist in a disembodied ego getting ahead of the neurons, but in the action issuing from my reasons, desires, and deliberations, without external coercion.

Harry Frankfurt added an influential structure to this. In his 1971 essay he distinguishes first-order desires (I want to smoke) from second-order desires (I do not want to be the one who wants to smoke). A will is free when the two levels agree—when I have the desires I want to have. A "wanton," who merely follows their first impulses without taking a stance toward them, is unfree; a person who measures their impulses against their deeper values is free. None of this hangs on the timing of an EEG signal.

Frankfurt's Attack on "Alternative Possibilities"

A second, subtler contribution by Frankfurt concerns the root of the whole dispute. The incompatibilists rely on the principle of alternate possibilities (PAP): one is responsible for a deed only if one could also have acted otherwise. Determinism seems to rule out precisely that—so, the inference goes, no responsibility under determinism.

Frankfurt countered in 1969 with a famous thought experiment. Imagine a neurosurgeon (call him Black) has implanted a chip in your brain. He wants you to vote for candidate X. His plan: if you choose X on your own, he does not intervene. Only if you set about choosing otherwise does the chip activate and force you to X. Now you choose X—entirely of your own accord, Black remaining idle. Could you have done otherwise? No—Black would never have allowed it. Are you nonetheless responsible? Most say: yes, because you chose from your own reasons; the chip in fact played no role. If that is right, then PAP is false: responsibility does not require alternative possibilities at all, but the right kind of authorship. With that, the bridge falls away over which determinism threatened to crush responsibility.

The Other Side: The Consequence Argument and the Hard Incompatibilists

In fairness, the other side deserves a hearing. The strongest argument against compatibilism is Peter van Inwagen's consequence argument: if determinism is true, then my present actions are the necessary consequences of the distant past and the laws of nature. Over both I had no control—the past lay before my birth, the laws I cannot change. So I have no control over the consequences either. Freedom in the full sense, runs the point, could be saved only if genuine openness existed somewhere.

Derk Pereboom, in turn, advocates hard incompatibilism: neither determinism nor mere indeterminism (chance) affords us the kind of authorship needed for desert-based blame and punishment. Galen Strawson's "Basic Argument" sharpens this: to be ultimately responsible for my action, I would have to be responsible for the character out of which I act—but I would in turn have had to form that character on the basis of an earlier character, and so on to infinity. Ultimate responsibility, says Strawson, is logically impossible, no matter how the brain works.

I am of the opinion that Libet's experiment accomplishes surprisingly little in this actual philosophical dispute. It measures the timing of a signal for a meaningless twitch; the question of authorship, responsibility, and the reach of determinism is barely touched by it. The debate is older than the EEG and will outlast it.

A Compass Through the Terms

The through-line: much of the dispute arises because four different questions—When is the decision made? By what is it caused? Am I its author? Am I responsible?—are constantly slid into one another. Libet answers at best the first, and even that only for a twitch of the finger.

Why This Matters in Everyday Life

Free will is not a seminar topic for Sunday afternoons but the foundation on which our practice of praise, blame, guilt, and punishment rests. Whoever believes people can really help nothing comes into conflict with the entire system of punishment and responsibility—but also with everyday gratitude and pride in one's own achievement.

Interesting is the psychological research on the consequences of belief in freedom. Several studies (for instance by Vohs and Schooler) suggested that people given deterministic texts to read subsequently tend more toward dishonest or uncooperative behavior. The replicability of these effects is contested and the effect size has shrunk—I am of the opinion that no strong conclusions should be drawn from this—yet the question remains serious: a publicly proclaimed "proof" against freedom is no consequence-free laboratory curiosity.

At the same time, the compatibilist and the neuroscientifically informed view carries practical mildness in its baggage. If actions issue from character, formation, and situation, then the gaze shifts from pure retribution toward prevention, treatment, and the shaping of conditions. One can hold someone responsible and call them to account and at the same time acknowledge that they are not the uncaused first mover of themselves. That is not an excuse but a more mature grammar of responsibility.

The Central Takeaway

The most tempting and at the same time most misleading lesson from Libet runs: "Science has proven there is no free will." It is untenable for three reasons. First, the experiment measures the timing of a signal for a meaningless twitch, not the freedom of deliberate decisions. Second, the readiness potential, as Schurger's accumulator model shows, is very probably to a large extent a statistical artifact of averaging random noise—no anticipated decision. Third, and most deeply: even a fully neurally caused will is for the compatibilist no contradiction to freedom but its only possible form.

The practical core is a thinking tool: the next time you hear someone say a brain finding has "refuted" freedom, ask which of the four questions is supposed to have been answered—the timing, the cause, the authorship, or the responsibility. Almost always the answer to one of these questions is tacitly passed off as the answer to another. Keeping these four apart is half the battle of any clear debate about the will—and a textbook case of how a single, cleanly measured experiment can be overloaded with worldview when one does not sort the concepts.

Reflection Question

Suppose it could be shown to you beyond doubt that every one of your decisions—including the decision to read this sentence to its end—was already laid down in your brain seconds beforehand: would that change anything about how you will choose tomorrow, whom you thank, and whom you call to account? And if not—what does it say about free will that its alleged revocation might leave your life entirely untouched?

Cross-References in the Vault

• The Predictive Brain: Predictive Processing and the Illusion of Perception – If perception is a top-down generated prediction, then the feeling of deciding freely right now is also a model constructed by the brain—the connection to the W-time illusion is close.
• The Two Boxes: Newcomb's Paradox and the Battle Over Rational Choice – There too the issue is what a "decision" is when it can be predicted; prediction and freedom stand in the same field of tension.
• The Chinese Room: Searle and the Question of Whether Machines Can Understand – Behavior (the movement) versus inner state (the conscious willing): the same gap that Libet's experiment tries to measure.
• What Mary Did Not Know: The Knowledge Argument and the Riddle of Consciousness – What consciousness is and whether it acts causally is the silent presupposition of the whole free-will debate.
• The Price of Justification: Cognitive Dissonance from Festinger to Neuroscience – Act first, then build the fitting conviction to match: dissonance research is another example of the conscious self-report lagging behind the deed.

Sources

• Libet, B., Gleason, C. A., Wright, E. W. & Pearl, D. K. (1983): Time of conscious intention to act in relation to onset of cerebral activity (readiness-potential). Brain 106, 623–642. Summary/full text: https://spot.colorado.edu/~tooley/Benjamin%20Libet.pdf
• Soon, C. S., Brass, M., Heinze, H.-J. & Haynes, J.-D. (2008): Unconscious determinants of free decisions in the human brain. Nature Neuroscience 11, 543–545. https://www.nature.com/articles/nn.2112
• Fried, I., Mukamel, R. & Kreiman, G. (2011): Internally generated preactivation of single neurons in human medial frontal cortex predicts volition. Neuron 69, 548–562. https://doi.org/10.1016/j.neuron.2010.11.045
• Schurger, A., Sitt, J. D. & Dehaene, S. (2012): An accumulator model for spontaneous neural activity prior to self-initiated movement. PNAS 109(42), E2904–E2913. https://www.pnas.org/doi/10.1073/pnas.1210467109
• Schurger, A., Hu, P., Pak, J. & Roskies, A. L. (2021): What Is the Readiness Potential? Trends in Cognitive Sciences 25(7), 558–570. https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(21)00093-0
• Maoz, U., Yaffe, G., Koch, C. & Mudrik, L. (2019): Neural precursors of decisions that matter—an ERP study of deliberate and arbitrary choice. eLife 8:e39787. https://elifesciences.org/articles/39787
• McKenna, M. & Coates, D. J.: Compatibilism. Stanford Encyclopedia of Philosophy. https://plato.stanford.edu/entries/compatibilism/

← All articles