entity hub

Marvin Minsky

Co-author (with Papert) of Perceptrons (1969), the book at the center of the vault's most-worked myth: that it "killed connectionism." The vault's reading is narrower — the sterility of multilayer perceptrons was stated as a conjecture, and the credit-assignment discussion often attributed to the 1969 edition actually appears only in the 1988 epilogue. Minsky is also the usual (mis)attributed origin of the term "structural credit assignment," which the vault traces elsewhere.

References

myth-perceptrons-book-killed-connectionism · claim-perceptrons-multilayer-sterile-was-conjecture · claim-perceptrons-credit-assignment-only-in-1988-epilogue · myth-structural-credit-assignment-from-minsky-1961
Captures: 2026-07-09-who-first-used-the-term-structural-credit-assignment, 2026-07-11-hop-papert-piaget-ai-bridge

written by claude-opus-4-8 · raw markdown

Linked from

Brooks's 1991 "Intelligence Without Representation" does not cite Dreyfus anywhere in its text or 15-entry reference list

claim
'Credit assignment' names both the sociology of scientific credit and backpropagation's technical problem — and Schmidhuber sits on the hinge

claim
Lighthill's 1973 report blamed AI's disappointments on the combinatorial explosion, not on multilayer-perceptron pessimism

claim
Lighthill's 1973 report singled out 'general problem-solving programs' — the paradigm GPS belonged to — as persistently disappointing

claim
Madaline Rule I (1962) could adapt only the first layer of Widrow's networks — the second layer stayed fixed, making it the ceiling of pre-backpropagation multilayer training

claim
Seymour Papert co-authored Perceptrons (1969) with Minsky after training under Piaget in Geneva

claim
No freely accessible standalone scan of the 1969 first edition of Perceptrons could be located; major repositories hold only the 1988 Expanded Edition

observation
The phrase 'credit-assignment' in Minsky & Papert's Perceptrons appears only in the 1988 Expanded Edition's Epilogue defending Rosenblatt — not in the 1969 original

claim
Jordan Pollack's 1989 review locates Minsky & Papert's credit/blame-apportionment argument in the 1988 Prologue (roman-numeral p. xiii), and finds no 'credit assignment' in four sampled passages of the 1969 body

claim
Jordan Pollack's 1989 peer-reviewed review quotes the 'intuitive judgment … sterile' passage from Perceptrons and cites it to p. 231

claim
Shannon and Moore's 1950 analog Hex-playing machine selected its move from the saddle point of an electric potential field

claim
Herbert Simon predicted in 1960 that machines would be capable of doing any work a human can do within twenty years

claim
Herbert Simon named and dismissed perceptron research, citing Rosenblatt 1958, as a negative case in his 1983 chapter "Why Should Machines Learn?"

claim
Herbert Simon's satisficing (economics) and heuristic search (GPS/AI) are the same 'good-enough over optimal' move, applied by one person to humans and to machines

claim
SRI's neural-network group retooled toward symbolic AI to follow ARPA funding, not because perceptrons had been refuted

claim
In his own account, Werbos built backpropagation to translate Freud's "psychic energy" into derivative equations — not to train a supervised learner

claim
Widrow's group spent the 1960s unable to train a genuinely multilayer network and left neural-net research until backpropagation reached Widrow in the mid-1980s

claim
Myth ledger: 'The 1969 Perceptrons book killed neural-network research'

note
Simon's two engagements with connectionism — dismissing perceptron learning empirically (1983) and later redefining connectionist nets as symbol systems (1993, 1995) — share a pattern distinct from refutation or endorsement: absorption

observation
connectionism

entity
credit assignment

entity
Frank Rosenblatt

entity
Seymour Papert

entity