Three Gaps Between Libet's Readiness Potential and the Death of Free Will — Epoche C1
Neuroscience has not shown that the brain decides before the person does. That claim rests almost entirely on one experiment — the study Benjamin Libet, Curtis Gleason, Elwood Wright and Dennis Pearl published in the journal Brain in 1983 — and the experiment is real, careful and much replicated. What is not supported is the conclusion drawn from it. Three distinct gaps lie between what was measured and what is asserted, and each can be stated exactly, with the evidence that opens it and the result that would close it. What was actually measured Libet's subjects were asked to flex a wrist whenever they felt like it, with nothing in the situation to recommend one moment over another, and to make no advance plan. Three signals were recorded on every trial. The first was electromyographic activity in the forearm, which fixes the moment the muscle actually contracted; all times below are measured backwards from it. The second was the readiness potential : a slow negative drift in the voltage recorded by an electrode on the scalp above the midline motor areas, first described by Hans Kornhuber and Lüder Deecke in 1965, which builds up in the second or so before a voluntary movement. The third was a report, and it is the one that carries the argument. Subjects watched a bright spot revolving around a marked clock face, completing a circuit in 2.56 seconds, so that each of the sixty marks corresponded to about 43 milliseconds rather than a second. After each trial they said which mark the spot had been passing at the instant they first became aware of the urge to move. Libet called that reported instant W. Averaged over trials, the readiness potential for unplanned movements began roughly 550 milliseconds before the muscle acted, while W fell about 200 milliseconds before it. Preparatory brain activity therefore preceded the reported awareness by something like 350 milliseconds. Libet was aware that the clock method might itself be biased, and ran a control in which subjects reported, in the same way, the time of a weak electrical pulse delivered to the skin at a moment the experimenters knew. Reports in that condition were displaced by roughly 50 milliseconds, which he treated as the method's systematic error and corrected for. That control is important for what follows, because it concedes the form of the first objection while attempting to bound its size. Libet's conclusion, and the stronger one usually attributed to him Libet himself did not conclude that will is an illusion, and it is worth being clear about this because the popular version misreports him. He argued that consciousness retains a veto over the developing action during its final 100 to 150 milliseconds — the interval between W and the point, roughly 50 milliseconds before contraction, after which the motor command is already travelling down the nerve and cannot be recalled. A "free won't" rather than a free will, on his account. The stronger reading, that the sense of having decided is manufactured after the fact, belongs chiefly to Daniel Wegner (2002). His theory of apparent mental causation holds that we experience ourselves as causing an action when three conditions are met: the relevant thought precedes the action (priority), the thought is consistent with the action, and no rival cause of the action is salient (exclusivity). On this account the feeling of willing is not a perception of a causal process but an inference drawn from the same evidence available to an observer, and Libet's timings are offered as showing that the real cause was in place before the thought. That is a substantive extra claim, and the three sections below concern the extra premises it needs. Gap one: W is a judgement, not a reading The entire argument is a comparison of two times, so it requires both to be measurements of the same kind. The readiness potential is one: a voltage recorded continuously against a clock. W is not. It is a subject's retrospective perceptual judgement about where a fast-moving spot was at a remembered moment, and such judgements carry their own latencies and their own biases. If forming the judgement takes time, or if it is systematically pulled towards later events, then W is displaced relative to the inner event it is meant to date, and the 350-millisecond interval is partly an artefact of the ruler. This is not speculation, and the decisive test asks whether W can be moved by information that arrives after the action. William Banks and Eve Isham (2009) had subjects press a button and hear a click that they took to be the sound of their own press, and inserted delays of up to 60 milliseconds between press and click. The reported time of the decision shifted systematically later as the delay grew. If W were a read-out of an inner event already past, a click occurring afterwards could not alter it; the natural explanation is that subjects reconstruct when they decided partly from the outcome they observe, exactly as Wegner's own inferential account of self-attribution would predict. The irony is worth registering: the mechanism Wegner proposes to explain the feeling of willing also undermines the measurement he uses to indict it. What would close this gap is a demonstration that W is stable under manipulations of post-action feedback, or a measure of awareness that does not depend on a timed report at all. Libet's skin-pulse control was an attempt at the first and establishes only that the bias is small in a condition where nothing follows the event to be dated. Banks and Isham's result concerns the condition where something does. Gap two: the readiness potential is an average The second gap concerns the other side of the comparison. The classic readiness potential is not a waveform visible in a single trial; it is what remains after dozens of trials are aligned on the moment of movement and averaged. The rise looks like a build-up towards a decision, but averaging is capable of producing a smooth rising curve out of a