Global Workspace Theory: Consciousness as Broadcasting, and What It Leaves Out
Abstract
Global Workspace Theory is the most engineering-minded of the major accounts of consciousness, which is why it keeps being borrowed by people building artificial systems. It describes consciousness as a function — making information available across a system — rather than as a substance or a place.
This article follows Bernard Baars's own 2013 exposition of the theory's dynamics, together with a 2021 discussion of its relationship to prefrontal cortex, to set out what the theory actually claims and what it deliberately does not.
The theory's strength is that it explains something real and puzzling. Its limitation is what it leaves untouched, and it is more candid about that than its popularisers tend to be.
1.The Observation the Theory Was Built to Explain
Start with a fact about your own mind that is strange once you notice it. The contents of consciousness at any moment are extremely limited — a sentence, an image, a sensation — while the brain is simultaneously running an enormous number of unconscious processes in parallel.
Baars frames this as the driving question of the theory: how can the limited capacity of momentary conscious contents be reconciled with the widespread access that conscious contents enable. Whatever is in consciousness is available to speech, memory, planning and action. Whatever is not is confined to its own specialised system.
The limited capacity of conscious perceptual contents has been discussed since the nineteenth century and rests on a large body of reliable evidence. What has been missing is a satisfactory functional explanation for why it should be limited at all — a gap Baars notes is still open.
2.Binding and Broadcasting
The proposal is that conscious experience reflects a flexible binding and broadcasting function which mobilises a large, distributed collection of specialised cortical networks that are not conscious by themselves.
Two steps, then. First, information from separate specialised systems is bound into a single coherent content. Second, that content is broadcast widely, making it available to any process that might need it.
Baars offers an analogy that is more precise than it first appears. The broadcast phase should evoke widespread adaptation for the same reason a fire alarm should evoke widespread response: because which responders will turn out to be relevant cannot be known in advance. Broadcasting is what you do when you do not yet know who needs the information.
The theory also proposes a directional signal flow, from binding coalitions to receiving coalitions — a testable claim, and one the author identifies as distinguishing the theory from neighbouring proposals.
3.What Supports It
The best-known prediction — widespread integration and broadcasting — has been supported by brain imaging. The general finding is robust: conscious stimuli typically evoke cortical activity that is more widespread, more intense and more correlated than matched unconscious stimuli.
That comparison is the methodological heart of the field. Using binocular rivalry, the attentional blink, backward masking and attentional manipulation, researchers can present physically identical stimuli that are sometimes consciously perceived and sometimes not, and ask what differs in the brain.
Baars is careful about what this does not show. Both conscious and unconscious stimuli trigger sensory volleys that can be traced well into the cortex. The difference is not that unconscious stimuli fail to reach the brain; it is what happens after they arrive.
4.The Prefrontal Question
Where the workspace is implemented has become the theory's most contested claim, and a 2021 discussion in Frontiers in Psychology addresses it directly, revisiting the role of prefrontal cortex in light of newer findings.
This is not a peripheral detail. The version of the theory tested in the adversarial collaboration described elsewhere in this library predicts brief content-specific prefrontal ignition at stimulus onset and offset — and that specific prediction was not confirmed.
A distinction is worth holding onto here. The functional claim — that consciousness involves making information globally available — is separable from the anatomical claim about where that happens. The first could survive the failure of the second. Whether it should is the argument currently under way.
5.What the Theory Does Not Attempt
Global Workspace Theory explains access: why some information becomes available across the system and most does not. It does not attempt to explain why any of that is accompanied by experience.
This is not an oversight but a choice of target, and it marks the sharpest contrast with Integrated Information Theory, which starts from the phenomenal properties of experience itself. One theory asks how information gets distributed; the other asks what it is about a physical system that makes it feel like something.
It is also why this theory travels so easily into artificial intelligence. A broadcasting architecture is something an engineer can build. Whether building it would produce anything experienced is a question the theory, on its own terms, does not answer.
6.How We Read It
Global Workspace Theory is the most tractable account in this area and the one with the clearest empirical programme. It explains a genuine puzzle — the mismatch between narrow conscious contents and their broad availability — and it makes predictions that can fail, one of which recently did.
Its scope is narrower than its popular presentation. When the theory appears in writing about artificial consciousness, the functional claim is usually doing the work while the unanswered question is quietly dropped. Keeping the two apart is most of what it takes to read this literature well.
Editorial Comment
MindHeaven® makes no claim relating to consciousness, and nothing in this article concerns any product. We publish it because our own AI Consciousness Lab uses deliberately cautious language — artificial cognition rather than consciousness — and this is one of the theories that makes clear why that distinction is worth maintaining.
Human studies exist, but are limited in size, population or consistency.
- 1.Baars BJ, Franklin S, Ramsoy TZ. Global workspace dynamics: cortical 'binding and propagation' enables conscious contents. Frontiers in Psychology. 2013;4:200. doi:10.3389/fpsyg.2013.00200.
- 2.Baars BJ, Geld N, Kozma R. Global Workspace Theory (GWT) and Prefrontal Cortex: Recent Developments. Frontiers in Psychology. 2021;12:749868. doi:10.3389/fpsyg.2021.749868.
- 3.Cogitate Consortium; Ferrante O, Gorska-Klimowska U, Henin S, et al. Adversarial testing of global neuronal workspace and integrated information theories of consciousness. Nature. 2025;642(8066):133–142. doi:10.1038/s41586-025-08888-1.
- 4.Melloni L, Mudrik L, Pitts M, et al. An adversarial collaboration protocol for testing contrasting predictions of global neuronal workspace and integrated information theory. PLOS ONE. 2023;18(2):e0268577. doi:10.1371/journal.pone.0268577.