Global workspace theory proposes consciousness arises when information enters a global workspace—a network of prefrontal and parietal regions—broadcasting it widely for reasoning, planning, and verbal report. The workspace has limited capacity (bottleneck), enabling only one conscious thought at a time. Information in the workspace becomes consciously accessible and influences multiple cognitive systems. This explains attentional blink, consciousness of goals we can't execute, and the dissociation between conscious access and unconscious processing.
Your prerequisite on neural correlates of consciousness introduced the question of what distinguishes brain activity that produces conscious experience from activity that doesn't. Global Workspace Theory (GWT) — developed by Bernard Baars and extended in neural terms by Stanislas Dehaene and Jean-Pierre Changeux — offers one of the most influential answers: consciousness is not a local property of any single brain region but a broadcast event that occurs when information is amplified and transmitted across a large-scale network.
The core metaphor is a theater: most cognitive processing happens backstage (unconsciously), in specialized, modular systems — visual feature extraction, auditory processing, motor planning, emotional appraisal. These systems operate in parallel and in isolation. The global workspace is the spotlight: a network centered on prefrontal and parietal cortex (the "ignition" network) that, when information reaches it, broadcasts that information widely to many distant brain regions simultaneously. This broadcast is what we experience as conscious awareness. Information that stays backstage — processed in sensory cortex but never broadcast — remains unconscious. The key neural signature is a dramatic late "ignition" of frontal-parietal activity, visible in EEG around 300–400ms after stimulus onset, that distinguishes consciously perceived stimuli from equally processed but subliminal ones.
The bottleneck structure of the workspace explains a wide range of phenomena. The attentional blink — the failure to consciously detect a second target presented 200–500ms after the first in a rapid sequence — occurs because the global workspace is still occupied with the first target's broadcast when the second arrives. There simply isn't capacity to ignite twice in quick succession. Inattentional blindness (famously, missing the gorilla in a basketball-watching task) occurs because the workspace is allocated to other content, and the unattended stimulus never achieves the threshold needed for ignition, even though it was fully processed at earlier visual stages. This is why attention and consciousness are closely linked — attention gates what gets broadcast — but are not identical: you can attend to something without becoming conscious of it, and vice versa in some conditions.
GWT also clarifies the relationship between conscious and unconscious processing. The global workspace is not needed for skill execution, emotional reactions, or perceptual processing — all of this proceeds without it. The workspace is specifically needed for flexible, arbitrary combination of information across modules: connecting a visual percept to a verbal label, using a current intention to override a habitual response, or reporting an internal state. This explains why you can drive a familiar route unconsciously while holding a conscious conversation, and why some goals ("I intend to be patient with my colleagues today") can sit in conscious awareness without yet being translated into behavior — the workspace encodes the goal, but execution depends on circuits that don't always respond to workspace signals.