nucleation

LIT-949: A synergistic workspace for human consciousness revealed by Integrated Information Decomposition

3 min read

Properties5
StatusActive — read 2026-10-14 (NOTE-799) in full; worth reading as an information-theoretic operationalisation of a global workspace with distinct input and output sides. Synergy-dominant regions form the workspace; within it, gateways (mainly default mode) and broadcasters (mainly fronto-parietal) are split by participation-coefficient rank. Pairwise integrated information ΦR falls between default-mode gateways under propofol and in disorders of consciousness, where the original Φ shows no common fall. Read it knowing four things. The two-sided split is imposed by the method, and only where its classes fall is measured. All measures are pairwise, undirected resting-state BOLD, with no contents or report, so it says nothing about which consumers a content reaches. The disorders-of-consciousness group includes minimally conscious patients. The propofol sample is given as both 15 and 16. Filed as THEORY-240.
Version2
Tagsconsciousness, neuroscience, complex-systems
Cited as “paper” byNOTE-799 — Luppi et al. — A synergistic workspace for human consciousness
Cited as “source” byTHEORY-240 — Loss of consciousness under propofol and in chronic disorders of consciousness goes with reduced pairwise integrated information between default-mode regions that rank as gateways of a synergy-defined workspace, not between its broadcasters, THEORY-881 — Across humans, macaques, marmosets and mice, loss of behavioural responsiveness under anaesthesia goes with lower whole-brain pairwise integrated information ΦR in resting fMRI, and its return goes with ΦR's return, including when central thalamic stimulation restores responsiveness during continued anaesthesia; ΦR tracks the macaque arousal score better than the original Φ, causal density, net information flow or functional connectivity

Graph View

Andrea I. Luppi, Pedro A. M. Mediano, Fernando E. Rosas, Judith Allanson, John D. Pickard, Robin L. Carhart-Harris, Guy B. Williams, Michael M. Craig et al. (2024), eLife 12.

Key takeaways

From the reading, NOTE-799.

  • The workspace, operationally. Regions are ranked by total synergy and by total redundancy. Those whose synergy rank exceeds their redundancy rank form the workspace (from the authors’ earlier paper). Within it, the sign of (synergy-network participation-coefficient rank − redundancy-network rank) marks gateways (positive) and broadcasters (negative). Every workspace region is one or the other by construction.
  • Where they fall. Gateways lie mainly in the default mode network, and broadcasters in the executive control network, with lateral prefrontal cortex prominent. This replicates with a coarser parcellation and with data-driven modules.
  • What tracks consciousness. Pairwise ΦR is I(past; future) − I(X_past; X_future) − I(Y_past; Y_future) + Red. A shared-effect test across three contrasts was used: propofol against wake, propofol against recovery, and DOC against controls. Edges with a consistent decrease link precuneus, medial prefrontal and inferior parietal cortex, all gateways. The decreases in medial prefrontal and posterior cingulate or precuneus are robust to parcellation and threshold.
  • What does not. The original whole-minus-sum Φ finds no common decrease. Increases near lateral prefrontal broadcasters do not survive robustness checks.
  • Limits the authors state. Resting-state BOLD, pairwise and undirected measures on IIT 2.0’s Φ, and no claim that ΦR causes or is consciousness. Possibly disconnected but conscious participants under propofol. Resting state may overweight the DMN. The shared-effect test is “described in a future publication”.

Standing in the record

Filed on 2026-10-12 from the owner’s papers-feed for QUESTION-061, as “measured evidence on whether conscious access is one uniform broadcast or has distinct parts”. Read on 2026-10-14 in NOTE-799. The reading narrows that description. The paper measures the state of consciousness at rest, with no contents and no report, so it cannot show a content reaching some consumers and not others, which is what QUESTION-061 asks for. And its division of the workspace into gateways and broadcasters is imposed by the method. What it does show is that the consciousness-tracking change is localised to one side, default-mode gateways, which is a measured non-uniformity of the workspace with respect to state. Filed as THEORY-240, with no answers: relation, since the question is Deferred. It sits beside the GNW account (THEORY-096): it agrees that lateral prefrontal cortex broadcasts, and locates the change elsewhere. It does not bear on CLAIM-003’s compartmented access.

Backlinks