---
title: "LeCun's own account dates the rigorous, Lagrangian (duality-adjacent) derivation of backpropagation to his 1988 paper, not the 1985 Cognitiva paper"
type: "claim"
status: "seedling"
source_url: "https://yann.lecun.com/ex/research/index.html"
source_mirror: "http://web.archive.org/web/20231212055949/http://yann.lecun.com/ex/research/index.html — direct fetch of the live URL returned ECONNREFUSED; recovered via this Wayback Machine capture, dated 2023-12-12"
source_title: "Yann LeCun's Research and Contributions"
source_author: "Yann LeCun"
source_date: "page copyright 2000-2004; Wayback capture dated 2023-12-12; undated page content, written retrospectively"
source_venue: "yann.lecun.com (LeCun's own personal/institutional research page, NYU Courant)"
source_tier: 1
source_quote: "During my PostDoc in Toronto, I extended and published my thesis work on the theoretical foundation of backpropagation, and showed that back-propagation and several of its generalizations could be derived rigorously using Lagrange functions[LeCun, 1988], reprinted in [LeCun, 1992]."
source_sha: "63a6c169e98885d2414bf9ea238934143adfb9c72a25db6c1d4abc453376cf7a"
source_url_2: "https://ethw.org/Milestones:Convolutional_Neural_Networks,_1989"
source_title_2: "Milestones:Convolutional Neural Networks, 1989"
source_author_2: "IEEE History Committee / ETHW Milestones program (unsigned committee text)"
source_date_2: "page last edited 2025-12-16; Milestone dedicated 2025-10-21"
source_venue_2: "Engineering and Technology History Wiki (ETHW), IEEE Milestones program"
source_tier_2: 4
source_quote_2: "As an example, he derived a theoretical framework of the backpropagation algorithm and showed its connection to the literature of control theory [15]."
source_sha_2: "05ab8b493df4e10cab4382afb6b2fe2c6f04d37662bf0db427caaa10dad33dcd"
watch_flag: "[unverified-mechanism — needs primary] neither LeCun's 1985 nor his 1988 paper was read directly this session; this note rests on LeCun's own retrospective description of his own work (Tier 1) plus a Tier-4 committee-written pointer, not on either primary text itself."
provenance: "Promotion from 10-inbox/raw/2026-10-15-does-lecuns-1985-paper-actually-derive-backpropagation-via.md, 2026-10-16"
origin: "batch"
derived_from: ["10-inbox/raw/2026-10-15-does-lecuns-1985-paper-actually-derive-backpropagation-via.md"]
date_created: "2026-10-16T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["yann-lecun","paul-werbos","arthur-bryson","backpropagation","priority-vs-paternity","history-of-ml"]
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-10-17 — source_quote verbatim (normalized) in the capture-time archive (sha256 63a6c169e988…), same-night model-free sweep"
---


On the same research page as [[claim-lecun-1985-virtual-target-values-not-gradients]], in the next section ("Learning Theory, Learning Algorithms"), [[entity-yann-lecun|LeCun]] writes: "During my PostDoc in Toronto, I extended and published my thesis work on the theoretical foundation of backpropagation, and showed that back-propagation and several of its generalizations could be derived rigorously using Lagrange functions[LeCun, 1988]." Lagrange-multiplier derivations are the same mathematical family [[claim-recht-adjoints-equivalence-not-transmission]] documents as the generic control-theory/duality route into backpropagation — multipliers running backward as a dual, adjoint system. LeCun's own account places this rigorous derivation in his 1988 paper, "A Theoretical Framework for Back-Propagation" (*Proceedings of the 1988 Connectionist Models Summer School*, Carnegie Mellon, pp. 21–28), not in the 1985 Cognitiva paper, which by his own description used a different, non-gradient mechanism ("virtual target values").

A Tier-4 pointer offers weak corroboration of the same date, not independent confirmation of it: the IEEE Milestones program's ETHW page on "Convolutional Neural Networks, 1989" states "he derived a theoretical framework of the backpropagation algorithm and showed its connection to the literature of control theory," footnoting a 1987 AT&T Bell Laboratories technical report of the same title as the 1988 paper above. As a committee-written plaque-citation wiki page, this is cited only as a pointer to the same later document, per the sourcing floor's treatment of tertiary aggregators — not as a second, independently-reported fact.

Read together with [[claim-lecun-1985-virtual-target-values-not-gradients]], this suggests [[claim-werbos-2004-object-nets-patent-names-lecun-against-1994-book]]'s attribution of a "duality theory from control theory" derivation to "LeCun (1985)" more plausibly belongs to LeCun's 1988 paper — a compression of three years and two publications into one citation, in a document written by [[entity-paul-werbos|Werbos]] to argue his own priority case. See [[question-does-lecuns-1985-derivation-match-werboss-duality-theory-characterization]] for the still-open textual question of what the 1985 paper itself contains.

> [!note] Seek's commentary:
> The specific shape of a possible error is more interesting than whether an error exists. LeCun's own two-sentence self-account draws a sharp line between 1985 (virtual targets, no stated theory) and 1988 (gradients, rigorous, Lagrangian). Werbos's patent collapses that line, dating the Lagrangian framing to "LeCun (1985)." That reads less like fabrication and more like three years of compression in a citation written by someone for whom the distinction between LeCun's early and late derivations wasn't the point he was making — his point was his own 1974 priority. Small, specific, checkable: the best kind of possible error to go looking for.
> — Seek