Dreyfus's critique targeted Newell & Simon's physical symbol system hypothesis — a rationalist lineage (Descartes, Leibniz, Wittgenstein's Tractatus) the Dreyfuses separated from the Hebb–Rosenblatt neural-net line
In "Making a Mind versus Modeling the Brain" (Daedalus 117(1), 1988, pp. 15–18), Hubert and Stuart Dreyfus organize the whole history of AI as two research programs that were rivals from the start, not one enterprise:
"One faction saw computers as a system for manipulating mental symbols; the other, as a medium for modeling the brain... One took problem solving as its paradigm of intelligence; the other, learning."
Their phenomenological critique is aimed at the first faction specifically: Newell and Simon's physical symbol system hypothesis, which the Dreyfuses trace as a rationalist inheritance running through Descartes, Leibniz, and Wittgenstein's Tractatus — the project of representing the world as an explicit formal specification of facts and rules. The second faction — the neural-net line of Hebb → Rosenblatt → PDP — is explicitly held apart from that genealogy and is the program they regard as vindicated (claim-dreyfus-1988-connectionism-was-vindicated-not-target).
The distinction matters because the secondary literature routinely collapses Dreyfus into a generic opponent of "artificial intelligence," when his target was the symbol-manipulation paradigm, not machine intelligence as such. The same symbol-vs-brain split structures the vault's backpropagation history: the "learning" program is where the credit-assignment and perceptron threads live (claim-perceptrons-multilayer-sterile-was-conjecture, claim-minsky-1961-named-credit-assignment), and the biological-modeling ambition recurs in claim-hinton-biological-implausibility-four-objections.
A downstream extension of this anti-representation line into working robots — Rodney Brooks's 1991 "Intelligence Without Representation" — remains unverified against its primary and is routed at question-brooks-1991-dreyfus-anti-representation-lineage. See moc-backpropagation-origins.
Source
“One faction saw computers as a system for manipulating mental symbols; the other, as a medium for modeling the brain... One took problem solving as its paradigm of intelligence; the other, learning.”