Dreyfus & Dreyfus's 1988 Daedalus paper frames Rosenblatt's connectionism — not symbolic AI — as the program their phenomenological critique vindicated

In "Making a Mind versus Modeling the Brain" (Daedalus 117(1), 1988, p. 34), Hubert and Stuart Dreyfus name Frank Rosenblatt's connectionist perceptron program — not artificial intelligence in general — as the research program their long-running phenomenological critique of AI supports. In their own words:

"Frank Rosenblatt's intuition that it would be hopelessly difficult to formalize the world and thus to give a formal specification of intelligent behavior had been vindicated. His repressed research program (using the computer to instantiate a holistic model of an idealized brain), which had never really been refuted, became again a live option."

This resolves the apparent "double role" the vault had queued around Stuart Dreyfus — the man who both derived a 1962 chain-rule optimization result feeding the backpropagation lineage (see claim-dreyfus-1973-lineage-link-uncorroborated and claim-backprop-special-case-kelley-bryson-formula) and co-authored the 1986 anti-AI polemic Mind Over Machine. There is no contradiction to reconcile: connectionism is the paradigm Stuart's own chain-rule mathematics trains, and the brothers cast it as their critique's ally, not its casualty.

The framing inverts the folk reading in which Rosenblatt's perceptrons were the thing "refuted" — a reading the vault tracks through the Minsky–Papert "sterile" conjecture in claim-perceptrons-multilayer-sterile-was-conjecture and myth-perceptrons-book-killed-connectionism. The Dreyfuses treat Rosenblatt's program as merely repressed, never actually refuted, and now revived. The distinct question of what the critique targets is carried by claim-dreyfus-critique-targeted-symbolic-ai-not-neural-nets. See moc-backpropagation-origins.