Source
Luppi, Uhrig, Tasserie et al. (2026), Nature Human Behaviour 10(4), LIT-1357, read in full in NOTE-1064: Figs. 2–4 and their text (pp. 778–784), the robustness section (pp. 790–791), the Discussion’s limitations (pp. 792–793), and Methods on ΦR (pp. 795–796). The Supplementary Information was not read. Second source: Luppi et al. (2024), LIT-949, the human propofol and disorders-of-consciousness result that this one extends.
What was actually shown
- The measure. For each pair of regions, ΦR = Φ2008 + Red→Red: the information the pair’s past carries about its joint future, minus each region’s information about its own future, plus back the persistent redundancy that the naive sum counts twice. It is computed with the Gaussian minimum-mutual-information redundancy and averaged over all pairs (pp. 779, 795–796).
- The fall. Mean ΦR is lower under anaesthesia than awake in humans (sevoflurane, n = 15), macaques (sevoflurane, propofol, ketamine; five animals), marmosets (sevoflurane, propofol, isoflurane; four animals) and mice (halothane), “except for medetomidine–isoflurane in the mouse” (p. 781). A six-region bipartition version also finds the medetomidine–isoflurane fall (p. 791).
- The return. Human ΦR rises again on spontaneous recovery. In two macaques under continuous propofol, central thalamic stimulation at 3 V and 5 V restored arousal and raised ΦR. High-intensity stimulation of the ventrolateral thalamus, one animal, restored neither (p. 781).
- Against rival measures. In a dominance analysis of the macaque arousal score, ΦR carries 51% of the explained variance (R² = 0.57), and 46% with functional connectivity added (15%). Φ2008 does not track anaesthesia consistently and goes negative; causal density misses recovery (pp. 781–782, 790–791).
What could have come out otherwise: ΦR could have stayed up under some drug, or in some species; it could have stayed down when stimulation restored arousal under continued propofol, which would have tied it to the drug, not to the state; and one of the rival measures could have tracked arousal as well. Apart from the one mouse regimen in the main analysis, none of these happened.
What this does not say
- Not that ΦR tracks experience. The target is responsiveness. The authors: “Behavioural unresponsiveness is an imperfect marker of unconsciousness since it can also occur as a result of sensory disconnection or motor impairment, neither of which is the same as unconsciousness” (p. 792). No condition here separates the two.
- Not that ΦR is integrated information in IIT’s sense. The authors’ intention is “‘not’ to test integrated information theory” (p. 796). ΦR is a pairwise, linear-Gaussian whole-minus-sum quantity, a “lower bound” in their words (p. 793).
- Not a regional claim. The averaged ΦR says nothing about which regions carry the change. The regional maps put the largest losses in unimodal visual and somatomotor cortex, with default-mode losses too (p. 784). So this THEORY does not carry THEORY-240’s gateway claim.
- Not that low ventrolateral stimulation is null. It raised ΦR, less, without restoring arousal (p. 781). The account tolerates this only if a partial rise below the behavioural threshold is allowed, which the authors propose and do not test.
- Not the mechanism. The PVALB correlations and the models are left out. They show sufficiency within models, not how anaesthetics act.
- Not about contents. Resting state, no stimulus, no report. Nothing here concerns what is integrated or for whom.
Connections
No answers: relation is declared. The question this bears on most,
QUESTION-048, asks what such evidence can establish about experience as
opposed to access, and this result is about responsiveness, so it adds a
case to that question without answering it. It is THEORY-091’s limit
again: a capacity lost with consciousness is lost with access too.
- THEORY-240 is the regional, gateway-side claim from LIT-949. This THEORY is the whole-brain claim it rests beside, now cross-species and with a causal arm. It half meets THEORY-240’s promotion test (restoration by thalamic stimulation) without the regional half.
- THEORY-063’s promotion test asks for a computed integration measure across anaesthesia. ΦR is one, but it measures the amount of pairwise integration, not the presence of one dominant cluster.
- THEORY-497. ΦR is mostly synergy-containing atoms, so this is a fall in the synergy THEORY-497 describes, under the same redundancy choice.
- CLAIM-412 and CLAIM-411. Integration is one of the candidate conditions the exchange named. This gives it a measured form that falls with responsiveness. It is not a measure of interpretation, and it gives no case that separates conscious from unconscious processing within a waking brain.