Source
- Schurger, Sitt & Dehaene (2012), LIT-605, read in full in NOTE-462, except the supporting information.
- Dennett & Kinsbourne (1992), LIT-431, read in NOTE-367: §3 on Libet, target article only.
The inference under attack
The readiness potential is a slow negativity in the average of EEG epochs time-locked to a self-initiated movement. In Libet’s task it starts a second or more before the movement, and people report the urge to move only about 200 ms before it. Read as the signature of planning, that gives the inference this account denies: the brain decides to act, unconsciously, and awareness is told later.
What was actually shown
The potential is what averaging noise at a threshold produces. Schurger et al. (LIT-605) model the decision of when to move as a single leaky accumulator, δx = (I − kx)Δt + cξ√Δt, with a weak constant urgency I, leak k, Gaussian noise and a threshold. They fitted three parameters to the waiting times of 14 participants in Libet’s task, and only to them. With those values, the sign-reversed average of simulated traces aligned to threshold crossings fitted the measured readiness potential from −3 to −0.15 s with r² = 0.96. The reason is selection. Only epochs that end in a crossing are averaged, so the average recovers the autocorrelated noise that happened to carry the trace to threshold. No single trace was building toward anything.
A prediction that the planning reading does not make held. If the buildup is ongoing fluctuation, it should be present whenever the system happens to be near threshold, whether or not the person was about to move. Interrupting 13 participants with clicks at random times, Schurger et al. found that their fastest responses were preceded by a slow negativity that began over 0.5 s before the click (P < 0.005), while the times of fast and slow clicks were spread equally through the trial (P = 0.64). That excludes a slowly building general readiness as the cause.
Where the decision goes instead. On this model the “neural decision to move now” is the threshold crossing, about 150 ms before movement. It coincides with the lateralised readiness potential and with the reported urge, which was 152 ms before movement in their data. So the urge is not late; the decision has not yet happened when the potential begins. The authors: Libet’s conclusion that the neural decision precedes awareness “by 1/2 s or more” is “unfounded”.
The clock report cannot time awareness anyway. Dennett and Kinsbourne (LIT-431) argue that the time a content is represented is not the time it represents, as a letter dated before a battle can arrive after news of it. The position of a clock hand reported as simultaneous with the urge is a judgment the brain concludes about content, “an artifact of the experimental situation” (p. 198), and not a reading of when an event in the brain occurred. If that is right, the comparison Libet made has no well-defined second term.
Two independent arguments. Schurger et al. attack the neural premise and accept the clock report as a rough estimate. Dennett and Kinsbourne attack the clock report and do not re-analyse the potential. They disagree on how far the report can be trusted (NOTE-462). Either one, if sound, removes the inference, and that is why both are sources.
What this does not say
- Not that the potential is noise. “Although our study demonstrates that the readiness potential could reflect non–goal-directed (spontaneous) neural activity, it does not prove that this possibility is in fact the case.” The claim is that the inference fails, not that its negation is established. That is why this is Proposed.
- Not that the conscious urge causes the movement. The model is silent on the urge. It places the urge and the threshold crossing together in time and says nothing about which, if either, does causal work.
- Not a claim about free will. It removes an empirical premise sometimes used against conscious will. It does not support or test the record’s free-will accounts, THEORY-029 and THEORY-040, which are about ownership and manipulation and do not rest on Libet.
- Narrow task. One movement per trial, timed for no reason. Deliberate choices with something at stake were not tested, and the account says nothing about them.
Connections
- The unity it is about.
behavioral-integration: when a single action is triggered, and by what. Awareness enters only as a reported time, so this is not a claim about phenomenal unity (ADR-024). - THEORY-068. Schurger et al.’s accumulator is the leak-dominant, single-unit case of the Usher–McClelland process that account treats. In Bogacz et al.’s terms it is a stable O-U process with an attractor at I/k below threshold, so crossings are driven by noise (NOTE-461; the mapping is the record’s).
- THEORY-061 generalises Schurger et al.’s picture of how the task goal acts: by raising the baseline toward threshold, with “the precise moment … not directly decided by a goal-directed operation”.