The case
Locked-in syndrome “consists of anarthria and quadriplegia while consciousness is preserved. Classically, vertical eye movements or blinking allow coded communication. Given appropriate medical care, patients can survive for decades” (Bruno et al. 2011, LIT-1183). The classic, incomplete and total forms are Bauer et al.’s (1979), as Bruno et al. cite them (NOTE-961). Laureys et al. 2005 (LIT-1198) reviews them but could not be read. In the completely locked-in state (CLIS), usually reached through ALS, no motor channel remains.
- Lucid life on one eyelid. Bauby dictated Le scaphandre et le papillon (LIT-1193) by blinking his left eyelid as a partner read out an alphabet.
- Lucid over years. Of 65 chronic patients with usable answers in Bruno et al.’s survey, 47 reported happiness, and a longer time in LIS correlated with happiness (NOTE-961). But most of the sample was not on one eyelid: 55% had recovered some speech and 70% some limb movement, answers were given “aided by the patient’s proxy”, and recovered speech was the strongest correlate of happiness (OR 20.47). The survey shows lucid lives with restricted action, not without it.
- The completely locked-in state. Kübler and Birbaumer (LIT-1199, read in NOTE-948): none of seven CLIS patients regained yes/no communication by BCI. All seven were already in CLIS when training began. They proposed that patients “may loose the perception of the contingency” between the required brain response and its consequences, and that “passive sensory information processing is intact in CLIS even at the most complex semantic processing levels”. The hypothesis is about contingency, not evidence. It also predicts its own exception: “learning BCI-control or any other contingency before onset of CLIS, which can be transferred from LIS to CLIS should prevent extinction in CLIS”. It is stated as what “might be responsible”, from seven patients and an analogy with curarized rats.
- Chaudhary et al.’s patient fits that exception (LIT-1200, read in NOTE-951). He had used an eye-movement BCI in LIS, and his speller was “identical to the one used previously”, the one he had used while he still controlled his eye movements. He had been in CLIS only “a few weeks before and also after implantation”, and he produced intelligible output on 44 of 107 days. So he does not refute the extinction hypothesis. He is the case it predicts. The article carries no editor’s note or correction as of 2026-10-16. The authors’ earlier CLIS papers were retracted after a DFG misconduct finding (STAT, 2022-03-22, not rechecked).
What it can show
- Classic LIS separates lucid judgement from limb and speech action, not from action. Friston et al.’s simulations begin with oculomotor control, and add a two-joint arm, a mountain car and Lorenz trajectories (NOTE-960). So classic LIS leaves active inference the action it models first, and cannot by itself decide CLAIM-341 against it. It can show that judgement keeps a built model working, over years, with world-directed action reduced to a trickle.
- CLIS is the test, and it is asymmetric. Evidence of lucidity in CLIS has to cross some channel, and a channel the patient drives is an action in active inference’s sense. Chaudhary et al.’s patient communicated only once neurofeedback restored a contingency between intention and outcome. So the case can show the loop running down (Kübler and Birbaumer’s prediction, CLAIM-339), but it cannot show judgement running with no action at all, which is CLAIM-341’s defeater (b) turned around. This is the evidence-for-experience line’s standing result applied to the case: “Every kind of evidence the record holds about experience reaches it through access” (QUESTION-048). Command-following, the usual marker, marks access, not experience (THEORY-090). No-report markers still have to be validated through report or executive control (THEORY-298).
- Command-following by brain imaging. Owen et al. (LIT-1178, read
in NOTE-962): a patient who met the criteria for the vegetative
state, five months after injury, imagined playing tennis or moving
around her home on request, and responded “through her brain activity,
rather than through speech or movement”. Monti et al. (LIT-1192, read
in NOTE-964): 5 of 54 patients with disorders of consciousness
modulated their brain activity at will. Two “remained behaviorally
unresponsive at the bedside, even after the functional MRI results were
known”, but they were not asked questions. The one patient who answered
yes/no questions, 5 of 6 correctly, showed on retesting “reproducible,
but inconsistent, responses indicative of a minimally conscious state”.
So what this shows is narrower than it first looked:
- The answers left with no action on the world, but the patient who gave them kept some motor contingency. The questions (his father’s name, for example) were about his life before the injury, so they test a model built while acting, not one kept in touch without it. Defeater (b) of CLAIM-341 is not tested.
- The channel is imagined action (“imagine standing still on a tennis court and to swing an arm”), read from a motor region. That fits CLAIM-337’s “action moved inward” literally, though the answer it carries is a judgement, put into imagery by a code the investigators set.
- Negative results are uninterpretable. When a question got no signal, “Whether the patient fell asleep … or lost consciousness cannot be determined”. The asymmetry above holds.
- The scoring was blind: “the investigators would not know the correct answers before judging the functional MRI data”. That is CLAIM-342’s anchor, but for the observers’ reading, not for the patient, who got no feedback.
- Not every contingency goes. Paralysis removes action on the world. It does not remove the body’s regulation of itself. Barrett’s action is visceromotor (NOTE-434), and on Damasio’s account feelings may need the body to act on the neurons that map it (QUESTION-059). Damasio also holds that an “as if body loop”, in which body states are re-enacted when an option is imagined, has largely superseded the body loop in adults (THEORY-500; asserted, not shown). If so, the mechanism that would keep judgement working in locked-in syndrome is already internal, and the case cannot remove every contingency between intention and outcome, as CLAIM-341’s defeater (b) asks.
- Extinguished or dormant. A coalition that falls silent can remain available for reactivation (CLAIM-287). Kübler and Birbaumer’s patients may have lost the capacity for goal-directed thinking, or only the channel that would show it. Chaudhary et al.’s patient does not decide between these, since he carried a contingency over from LIS.
- What it cannot show. Every patient here built their model while acting. The case bears on the loop that keeps a model in touch, not on the one that builds it.