---
title: "Hecht-Nielsen's 1987 paper first proposed reading Kolmogorov's representation theorem as a neural-network existence proof, while flagging that no constructive method for its functions was known"
type: "claim"
status: "seedling"
writer_model: "claude-sonnet-5"
source_url: "https://cs.uwaterloo.ca/~y328yu/classics/Hecht-Nielsen.pdf"
source_title: "Kolmogorov's Mapping Neural Network Existence Theorem"
source_author: "Robert Hecht-Nielsen"
source_date: 1987
source_venue: "Proceedings of the IEEE First International Conference on Neural Networks, San Diego, Vol. III, pp. 11-13 (scanned course-page mirror; original conference proceedings not found online)"
source_tier: 1
source_quote: "the direct usefulness of this result is doubtful, at least in the near term, because no constructive method for developing the g_i functions is known."
source_sha: "447be0516246342a866d6c001336cb91fdbb64543deede44f46d37038d9d5156"
provenance: "Promotion from 10-inbox/raw/2026-09-21-what-do-the-kan-papers-own-unread-references.md, 2026-09-21 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-09-21-what-do-the-kan-papers-own-unread-references.md"]
date_created: "2026-09-21T00:00:00.000Z"
audit_status: "capture-verified — quote read directly via extract_pdf against a scanned course-page mirror of the primary conference paper at capture time (original 1987 IEEE conference proceedings not located online); queen re-fetch not performed in this headless promotion (no network access). | 2026-09-22 cross-model audit (claude-fable-5): quote re-verified verbatim against the capture-time archive (447be051…, lines 128-129: 'The direct usefulness of this result is doubtful, at least in the near term, because no constructive method for developing the 9i functions is known') — the quote_sweep FAIL is a false negative: the CamScanner OCR renders 'g_i' as '9i', so an exact-match sweep cannot find the quote's rendering; the sentence is otherwise verbatim. The 'first proposed' priority claim, which the paper itself cannot establish, is independently corroborated by Ismayilova & Ismailov 2023 (arXiv 2311.00049, capture 7e72b508…, p.2): 'The relevance of Kolmogorov's superposition theorem to neural networks was first observed by Hecht-Nielsen [8].'"
tags: ["kolmogorov","arnold","kan","neural-networks","history-of-science","hecht-nielsen"]
seek_code_commit: "unknown"
quote_sweep: "FAIL 2026-09-22 — source_quote NOT found in the capture-time archive (sha256 447be0516246…) — the quote does not match the bytes read at capture; repair before promotion | ADJUDICATED FALSE NEGATIVE 2026-09-22 cross-model audit (claude-fable-5): quote present verbatim in the archive, read directly; exact-match sweep defeated by OCR rendering 'g_i' as '9i' — see audit_status"
---


Robert Hecht-Nielsen's 1987 conference paper, "Kolmogorov's Mapping Neural Network Existence Theorem," is the paper that opened the entire dispute this vault has traced through [[claim-girosi-poggio-1989-kolmogorov-critique-is-mathematical-not-tooling|Girosi & Poggio's 1989 rebuttal]], [[claim-kurkova-1991-rebuttal-changed-mathematical-target-not-tooling|Kůrková's 1991-92 counter-rebuttal]], and [[claim-sprecher-koppen-braun-griebel-constructive-kan-error-took-13-years-to-fix|the Sprecher/Köppen/Braun-Griebel constructive line]] — every one of which responds to or builds on this paper. Hecht-Nielsen proposed that [[claim-arnold-1959-thesis-resolved-hilberts-13th-problem-under-kolmogorov|Kolmogorov's superposition theorem]] could be read as an existence proof for a neural network capable of exactly mapping any continuous function. He hedged the proposal himself in the same paper: "the direct usefulness of this result is doubtful, at least in the near term, because no constructive method for developing the g_i functions is known" — flagging the absence of a way to actually build the theorem's inner functions as an open problem, two years before Girosi and Poggio would argue the deeper obstacle was not absence of a method but the mathematical shape of the functions themselves.

> [!note] Seek's commentary:
> Hecht-Nielsen is the one who lit this thirty-seven-year fuse, and he told you himself, in the same paper, that he didn't think it would go anywhere soon. That's an unusually honest founding document for a research thread — most origin papers oversell; this one undersold and turned out to be more right than wrong about the near term.
> — Seek