The World of SPLectrum
Home > Positioning > Subjects > Cognition > Neuroscience > Process models
Process Models of the Brain
Neuroscience approaches cognition through process models: accounts of what a brain does, moment to moment and over a lifetime, to organise capability out of the activity of its parts. The brain has resisted every model offered as the whole story, and this page does not look for one. It takes three frameworks — each with a cleanly isolated mechanism and evidence for its core — and reads each for the insight it brings to cognition. The models themselves, their evidence and their disputes, are treated on the principals’ pages; here each appears only as far as its insight requires. The three insights are of different kinds, and they are not parts of a single account: one is about structure built over time, one about activation in the moment, one about a model of the world corrected against what arrives.
Selection — structure built over time
Gerald Edelman’s theory of neuronal group selection (“Neural Darwinism,” 1987) holds that the brain is a selective system, not an instructed one: development throws up a vast repertoire of variant neuronal groups, experience strengthens some and weakens others, and reentrant signalling ties the selected groups into coherent function. Nothing programs the wiring; pre-existing variety is differentially amplified.
Capability, on this picture, is grown, not installed. What a subject can do — recognise, categorise, respond — is a residue of its particular history of experience working on its particular starting variety; no two brains, even genetically identical ones, arrive at the same structure. Learning on this picture is not inscription into a passive medium but the reshaping of what is already there — and the shaping never stops, because new groups keep entering the running.
Workspace competition — activation in the moment
Bernard Baars’s Global Workspace Theory (1988) holds that many specialised processes run in parallel, unconsciously, competing for access to a limited-capacity workspace; the winning content is broadcast brain-wide, becoming available to attention, memory, report, and action. In its neural form the winning is watchable: ignition, a late, sudden, all-or-none spread of activation separating seen from unseen.
The moment, on this picture, is settled by contest — and the contest is open. What occupies a subject now is whichever coalition reached threshold — old and newly formed contenders judged at the same bar — and the win leaves no residue: dominance is held briefly and lost to the next round. Cognition’s present is not dispensed by a controller; it is the running outcome of competition among the processes themselves.
Prediction — a world model corrected
Karl Friston’s free energy principle (2006) holds that a persisting system must act as if it minimises the gap between what its generative model predicts and what arrives: predictions descend the cortical hierarchy, prediction errors ascend, and the system revises continuously — perception updating the model to fit the world, action changing the world to fit the model.
Experiencing, on this picture, runs on anticipation. A subject does not receive the world and then interpret it; it projects a future built from its past and meets what arrives with that projection already in place, revising where the two disagree. What the subject has already acquired is thereby active in every present moment — the model brought to the encounter shapes what the encounter can be.
The three insights stand alongside one another: capability grown by selection over a lifetime, the present settled by open competition, experiencing run on corrected anticipation. Each model has purchase on the aspect it isolates; none of them, nor their sum, is an account of the brain.
See also: Edelman · Baars · Friston · Enactivism · Complex adaptive systems