Free Will
In 1983, Benjamin Libet wired volunteers' scalps and asked them to flick a wrist whenever they wished. A negative voltage shift in motor cortex — the readiness potential — appeared 300–500 ms before subjects reported the conscious urge to move. The brain had started without asking.
Forty years and several replications later, the head start has grown. Using fMRI, Soon and Haynes (2008) decoded left-vs-right button press from supplementary motor and dorsolateral prefrontal cortex up to 10 seconds before subjects said they decided. In 2013, Haynes's group pushed the result to abstract cognition: a choice to add vs subtract was decodable from medial prefrontal and parietal cortex 4 seconds before reported awareness.
If decisions are detectable in neural tissue before the conscious mind notices, who exactly is choosing?
The case, condensed
| Study | Year | Lead time | Decision type |
|---|---|---|---|
| Libet | 1983 | 300–500 ms | Motor (wrist flick) |
| Soon & Haynes | 2008 | up to 10 s | Motor (left/right press) |
| Haynes et al. | 2013 | ~4 s | Abstract (add vs subtract) |
Libet's own gloss on his result was narrower than the popular reading. The conscious will cannot initiate, he argued, but in the final 200 ms before movement it can veto — "free won't" rather than free will. Haynes was less consoling: "How can I call a will mine if I don't even know when it occurred and what it has decided to do?"
What's contested
Three rebuttals matter.
The W problem. Subjects estimate the moment of conscious decision by watching a rotating clock and reporting where the dot was when they "decided." Banks and Isham (2009) showed those reports can be shifted by giving fake post-movement feedback. The W judgment is partly reconstructed memory, not a clean timestamp.
The RP isn't a decision signal. The readiness potential rises before voluntary movements, but also before trials in which subjects choose not to move. Schurger et al. (2012) modelled it as ongoing neural noise that crosses a threshold — a stochastic accumulator, not a decision. A 2024 methodological review by Dominik and colleagues argues the original evidence for temporal precedence of unconscious over conscious decision was weaker than the textbooks claim.
Prediction is not the same as causation. A 60–70% above-chance fMRI decode of a binary choice tells you the brain was preparing a bias, not that the conscious self was bypassed. The accuracy ceiling in these experiments is closer to "the brain has leanings" than "the brain has already chosen."
Honest summary: Libet shook the naive picture, but the strong claim "neuroscience disproves free will" is doing more work than the data supports.
The three philosophical camps
Compatibilism (Dennett, Frankfurt, Churchland — the dominant academic position). An act is free if it flows from your beliefs, desires and character, even though those were shaped by genes, gut bacteria, and upbringing. Indeterminism was never the point.
Hard incompatibilism (Sapolsky, Pereboom). If every action traces back to prior causes you didn't author, no one deserves praise or blame. Sapolsky's Determined (2023) argues for replacing retribution with incapacitation, rehabilitation and prevention.
Libertarian free will. A minority position that pins agency to genuine indeterminism, sometimes quantum collapse in neurons. The standing objection: a choice made by a random quantum event is not more free, only less determined. Randomness isn't authorship.
Predictive coding: the brain that hallucinates its own choosing
Friston, Clark and Anil Seth describe the brain as a prediction machine that generates models of its own next state and only updates on prediction error. In that frame, "deciding" is the brain forecasting its own motor output; the conscious experience of willing arrives after the model has fired. Seth calls the sense of agency a controlled hallucination — useful, even necessary for social life, but not a faithful report of causal order.
This matters practically: priming people to believe they have no free will makes them cheat more in laboratory tasks (Vohs & Schooler, 2008). The belief in agency is itself part of the causal chain, whether or not the metaphysics behind it holds up.
Why this has to do with other realms
The free-will problem and the concept arrow of time are the same kind of illusion in two disguises. Fundamental physics has no built-in time direction; thermodynamics gives it one through entropy, but the brain perceives causality as a sharp forward arrow. Neuroscience says consciousness has no clean causal priority over the neural events that produce action, yet we experience ourselves as forward-acting choosers. Both the felt arrow and the felt agency may be narratives the brain constructs from data that doesn't contain them — adaptive fictions running on the same prediction hardware.
A stranger bridge runs through the gut. The concept gut brain axis estimates that around 90% of serotonin and a majority of tryptophan precursors are processed in the gut, and gut microbial composition correlates with fairness judgments in ultimatum-game studies. If a fraction of "your" decisions are biased by microorganisms you never chose to host, the boundary of the deciding self gets fuzzy before the philosophy even starts.
An open question
If a perfect brain scanner could decode your next choice five seconds early with 99% accuracy and announce it on a screen — would you still feel free reading the screen, and would you be wrong to?
Key sources
- Libet, B. (1983). "Time of conscious intention to act in relation to onset of cerebral activity." Brain 106: 623–642 — the original readiness-potential experiment.
- 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 — the 10-second fMRI result.
- Schurger, A., Sitt, J.D., Dehaene, S. (2012). "An accumulator model for spontaneous neural activity prior to self-initiated movement." PNAS 109: E2904–E2913 — the noise-accumulator reinterpretation of the RP.
- Sapolsky, R. (2023). Determined: A Science of Life Without Free Will — the strongest recent hard-incompatibilist case.
- Seth, A. (2021). Being You: A New Science of Consciousness — predictive coding and agency as controlled hallucination.
- Dennett, D. (2003). Freedom Evolves — the canonical compatibilist counter.
Further reading
- Determined by Robert Sapolsky — the book to read if you want the case against retribution stated without hedging.
- Being You by Anil Seth — best lay treatment of predictive coding and why agency may be a model.
- Schurger 2012 PNAS paper — read this if you only know the textbook Libet story; the accumulator model changes the inference.
- Free Will by Sam Harris (2012) — short, sharper than its critics admit, useful as a foil even if you disagree.
- Freedom Evolves by Daniel Dennett — required if you want to take compatibilism seriously rather than dismissing it.
See Also
- concept gut brain axis — bacteria help author judgments your conscious mind takes credit for
- concept arrow of time — the sense of temporal direction and the sense of agency may be the same illusion
- concept emergence — if the mind is causally more powerful than its neurons, does top-down agency get a second life?
- concept distributed cognition — octopus arms with semi-autonomous control; what is the minimum unit of "deciding"?
- concept simulation hypothesis — deterministic computation and hard incompatibilism end up in the same place
- concept overview effect — astronauts report a dissolved self; does that dissolve the premise of individual free will too?
- concept spontaneous stochasticity free will — the neural extension of weather-forecasting physics: if brain dynamics exhibit spontaneous stochasticity at sufficient Reynolds number, the Libet "free won't" window is a classical thermodynamic horizon, not a quantum mystery or a philosophical fiction