Few scientific findings have been quoted in philosophical arguments as often as Benjamin Libet's experiments on the timing of voluntary action. They are worth understanding carefully, because they are frequently described as proving more than they do.

The readiness potential

In 1965 the German researchers Hans Kornhuber and Lüder Deecke discovered the readiness potential: a slow buildup of brain activity, measurable at the scalp, before a voluntary movement. The question was how this buildup relates to the moment a person consciously decides to move.

Libet's experiment

In a 1983 study, Libet and his colleagues asked participants to flex their wrist whenever they felt the urge, while watching a fast-moving spot on a clock face. Afterward, each participant reported where the spot had been when they first became aware of the urge to move. Researchers recorded the readiness potential and the movement itself.

The results, roughly: for spontaneous movements the readiness potential began about half a second before the movement, while reported awareness of the urge came only about a fifth of a second before it. In other words, the readiness potential began a few hundred milliseconds before awareness. Some concluded that the brain “decides” before we do, and that conscious will is an afterthought.

Libet himself did not draw that conclusion. He suggested that consciousness could still veto a movement before it happens, sometimes summarized as free won't.

The limits of the experiments

• Measuring awareness is hard. The timing of awareness depends on subjects' reports of a clock position, which are vulnerable to errors of attention and memory.
• The task is unusual. Flexing a wrist at a random moment, with no reason to choose one time over another, is very unlike deciding whether to take a job or keep a promise.
• The meaning of the signal is disputed. In 2012 Aaron Schurger and colleagues proposed that the readiness potential may reflect random fluctuations building up to a threshold. Averaged over many trials, such noise looks like a steady ramp, even if no decision is being made early.
• Prediction is weak. A 2008 brain-imaging study by Chun Siong Soon and colleagues could predict whether people would press a left or right button several seconds before they reported deciding, but only modestly better than chance, at around sixty percent where guessing gives fifty.

What follows philosophically

Even if brain activity does precede awareness, what would that show? It would challenge a picture in which a conscious, nonphysical self starts each action from outside the brain. But compatibilists are generally untroubled, because they already think decisions are brain processes; it is no surprise that some of the process is unconscious. Much skilled behavior, such as returning a fast tennis serve, happens before we are aware of it, and we still count it as ours.

The psychologist Daniel Wegner argued that our sense of willing an action is a feeling the brain constructs, and can be mistaken; he described experiments in which people felt they had caused events they had not. Such findings show that introspection can mislead us. They do not by themselves show that our reasons and deliberations make no difference to what we do.

The careful conclusion is modest. Neuroscience is teaching us a great deal about how decisions unfold in the brain, but the existing experiments do not settle the philosophical question of free will.

Worked example: reading a headline

Suppose you see a headline: “Brain scans prove your choices are made before you know it.” How can you assess it using this level?

Step one: what was the task? The story describes an experiment where people chose, at random moments, which of two buttons to press. That is a decision with no reasons on either side. Ask whether it tells us much about choices involving deliberation, such as whether to move to a new city.

Step two: what was measured? The researchers recorded brain activity and asked people when they became aware of deciding. Awareness timing was based on self-report, which can be imprecise.

Step three: how strong was the prediction? Read further and you learn the scans predicted the button about sixty percent of the time. Guessing gives fifty. The headline's word “prove” now looks far too strong.

Step four: what could the signal be? Recall the 2012 proposal that early activity may be background noise drifting toward a threshold rather than a hidden decision.

Step five: what follows philosophically? Even if unconscious processes start a choice, a compatibilist will say that those processes are part of you. The finding may threaten the idea of a nonphysical self that acts from outside the brain, but not every idea of free will.

A fairer headline might read: “Study finds early brain activity modestly predicts simple choices.” Less exciting, but true to the evidence. The same five steps can be applied to many science stories about the mind.