---
id: "20260712-0220-was-amaris-1972-associative"
title: "Was Amari's 1972 associative-memory model ever independently applied to developmental biology before the 2016 Taherian Fard Hopfield-landscape paper?"
type: "capture"
status: "promoted"
origin: "batch"
writer_model: "claude-sonnet-5"
date_created: "2026-07-12T00:00:00.000Z"
provenance: "batch run, 2026-07-12"
promoted: "2026-07-12, headless promotion run (writer: claude-opus-4-8)"
promoted_to: ["30-notes/claim-no-amari-application-to-developmental-biology-found-pre-2016.md","30-notes/claim-2016-hopfield-landscape-paper-not-first-cell-fate-attractor-model.md"]
questions_routed: ["50-questions/question-amari-1972-applied-to-developmental-biology-before-2016.md","50-questions/question-amari-1977-neural-field-applied-to-morphogenesis.md"]
not_promoted: ["Claim: Lang, Li, Collins & Mehta (2012/2014) cites Hopfield/Amit/Kanter-Sompolinsky but not Amari across 58 refs — FOLDED into claim-no-amari-application-to-developmental-biology-found-pre-2016 as its strongest single piece of evidence, rather than a near-duplicate standalone negative note.","Claim: Yampolskaya & Mehta (2025) review cites zero Amari papers across 69 refs — FOLDED into the same synthesis note as the comprehensive-coverage evidence.","Seek's commentary on WebSearch fabricating Amari→GRN citation links — process/reflection observation, not an atomic world-claim; recorded in the 2026-07-12 journal and as a commentary callout on the synthesis note."]
derived_from: []
tags: ["amari","hopfield-networks","developmental-biology","waddington","priority","history-of-ml","citation-analysis","epigenetic-landscape"]
sources: [{"source_url":"https://arxiv.org/pdf/2506.13076","source_author":"Maria Yampolskaya, Pankaj Mehta","source_date":"2025-06-16T00:00:00.000Z","source_venue":"arXiv:2506.13076 [physics.bio-ph] (Annual Review of Biophysics, forthcoming)","source_tier":1,"source_note":"Fetched with extract_pdf, tls: verified. Direct read of full text and 69-entry reference list."},{"source_url":"https://arxiv.org/pdf/1211.3133","source_author":"Alex H. Lang, Hu Li, James J. Collins, Pankaj Mehta","source_date":"2014-08-01T00:00:00.000Z","source_venue":"PLoS Computational Biology 10(8):e1003734 (arXiv preprint 1211.3133, first posted 2012)","source_tier":1,"source_note":"Fetched with extract_pdf, tls: verified. Direct read of full text and 58-entry reference list."},{"source_url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC5516853/","source_author":"Atefeh Taherian Fard, Sriganesh Srihari, Jessica C. Mar, Mark A. Ragan","source_date":"2016-01-01T00:00:00.000Z","source_venue":"npj Systems Biology and Applications 2, 16001","source_tier":1,"source_note":"Checked via WebFetch against the PMC full-text mirror, quoting specific reference-list entries and in-text citation sentences."},{"source_url":"https://academic.oup.com/bioinformatics/article/30/9/1273/234782","source_author":"Stefan R. Maetschke, Mark A. Ragan","source_date":"2014-05-01T00:00:00.000Z","source_venue":"Bioinformatics 30(9):1273-1279","source_tier":1,"source_note":"Checked via WebFetch against the publisher page for reference-list contents."}]
---


This capture directly researches the open question logged at
[[question-amari-1972-applied-to-developmental-biology-before-2016]]. It
builds on two existing claim-notes: the vault's Amari-priority thread in
machine learning
([[claim-amari-1972-associative-memory-precedes-hopfield]]) and the
Hopfield-landscape formalization of developmental biology
([[claim-hopfield-network-formalizes-waddington-epigenetic-landscape]]).

## Claim: Taherian Fard et al. (2016) cites three pre-2016 Hopfield-style attractor models of cell fate — Wang et al. 2011, Lang et al. 2014, Maetschke & Ragan 2014 — and its own reference list contains no citation to Amari

**Claim type**: historical / bibliographic (load-bearing for the central question — held to Tier 1-2).

Direct inspection of the Taherian Fard et al. (2016, *npj Systems Biology and
Applications* 2:16001) reference list turns up no Amari citation anywhere.
The paper does, however, explicitly credit prior Hopfield-network-style
landscape models that predate it: "Wang *et al.* used ODEs and simulated data
to construct a probabilistic 'pseudo-potential' energy landscape" (citing
Wang, Zhang, Xu & Wang, *PNAS* 108:8257-8262, 2011); "Lang *et al.* used
binarized gene-expression data... They employed binarized gene-expression
values" (citing Lang, Li, Collins & Mehta, *PLoS Comp. Biol.* 10:e1003734,
2014); and "Maetschke and Ragan, on the other hand, constructed an HN from
static gene-expression data" (citing Maetschke & Ragan, *Bioinformatics*
30:1273-1279, 2014). All three of these predate the 2016 paper by two to
five years and are themselves Hopfield-network applications to
developmental/cancer biology — meaning Taherian Fard 2016 is not the first
paper to model biological cell-state landscapes with Hopfield-style
attractor dynamics, only (per the existing vault claim) the one that states
the mapping most explicitly and validates it across 12 datasets. None of the
three prior works, nor the 2016 paper itself, traces to Amari.

*provenance*: Taherian Fard, A., Srihari, S., Mar, J.C. & Ragan, M.A. (2016).
"Not just a colourful metaphor: modelling the landscape of cellular
development using Hopfield networks." *npj Systems Biology and
Applications* 2, 16001. https://pmc.ncbi.nlm.nih.gov/articles/PMC5516853/
— checked via WebFetch, quotes as above.

## Claim: A pre-2016 primary source (Lang, Li, Collins & Mehta, 2012/2014) applying an explicitly Hopfield/spin-glass associative-memory framework to stem-cell reprogramming cites Hopfield, Amit, and Kanter & Sompolinsky — but not Amari, anywhere in its 58-entry reference list

**Claim type**: historical / bibliographic (load-bearing — Tier 1-2 required, met).

Lang, Li, Collins & Mehta's "Epigenetic landscapes explain partially
reprogrammed cells and identify key reprogramming genes" (posted to arXiv as
1211.3133 in November 2012, published in *PLoS Computational Biology*
10(8):e1003734 in August 2014) is a pre-2016 primary source that builds a
cell-fate landscape model directly out of associative-memory machinery: a
Hebbian-style interaction matrix \(J_{ij}\), a "projection method" for
correlated patterns, and an explicit discussion of "spurious attractors"
that "take the form of odd-majority vote mixtures" — vocabulary drawn
straight from the classic Hopfield/spin-glass associative-memory
literature. The paper was read in full via `extract_pdf` (arXiv PDF, tls:
verified) and its complete 58-entry reference list checked directly. It
cites: Hopfield, J.J. (1982) *PNAS* 79:2554-2558 [ref 30]; Hopfield, J.J.
(1984) *PNAS* 81:3088-3092 [ref 57]; Amit, Gutfreund & Sompolinsky (1985)
*Phys. Rev. A* 32:1007-1018 [ref 31]; Kanter & Sompolinsky (1987) *Phys.
Rev. A* 35:380-392 [ref 32]; and Amit (1992) *Modeling Brain Function*
[ref 33]. **Amari does not appear anywhere in the reference list.**

*provenance*: Lang, A.H., Li, H., Collins, J.J. & Mehta, P. (2012/2014).
"Epigenetic landscapes explain partially reprogrammed cells and identify key
reprogramming genes." *PLoS Comp. Biol.* 10(8):e1003734; arXiv:1211.3133v4.
https://arxiv.org/pdf/1211.3133 — read directly, references 30-33 quoted
verbatim above from the extracted text.

## Claim: A comprehensive 2025 review of Hopfield-network applications across biology — including the full cell-fate/epigenetics literature the vault is tracking — cites zero Amari papers across 69 references

**Claim type**: historical / bibliographic (Tier 1-2 required, met).

Yampolskaya & Mehta's "Hopfield Networks as Models of Emergent Function in
Biology" (arXiv:2506.13076, June 2025; slated for *Annual Review of
Biophysics*) is a pedagogical review whose §4.1 ("Cell fate transitions and
Epigenetics") surveys "a growing body of work attempting to formalize the
Waddington metaphor using dynamical systems theory and Hopfield models,"
explicitly listing twelve works spanning 2011-2024 (Wang et al. 2011;
Ferrell 2012; Huang 2012; Moris et al. 2016; Lang et al. 2014; Szedlak et al.
2017; Guo & Zheng 2017; Rand et al. 2021; Raju & Siggia 2023; Smart & Zilman
2023; Karin 2024; Boukacem et al. 2024) plus Taherian Fard et al. 2016 and
Maetschke & Ragan 2014 discussed in the body text. The paper was read in
full via `extract_pdf` (tls: verified) and its complete 69-entry "Literature
Cited" section checked directly, end to end. **No Amari citation appears
anywhere in the reference list.** Every genealogy the review draws for
Hopfield-style biological modeling — cell fate, self-assembly, spatial
neural representations — runs back to Hopfield (1982, 1984) and the
spin-glass statistical-physics tradition (Amit, Gutfreund & Sompolinsky
1985; Kanter & Sompolinsky 1987), not to Amari.

*provenance*: Yampolskaya, M. & Mehta, P. (2025). "Hopfield Networks as
Models of Emergent Function in Biology." arXiv:2506.13076.
https://arxiv.org/pdf/2506.13076 — read directly; full reference list
(items 1-69) inspected.

## Claim: No source found in this search documents an independent application of Amari's 1972 model to developmental biology, before or after 2016

**Claim type**: historical (synthesis of the above — this is an
absence-of-evidence finding, not a proof of nonexistence, and is flagged as
such rather than stated as a settled fact).

Across the four primary/near-primary sources checked directly for this
capture — the 2016 Taherian Fard paper itself, its three cited pre-2016
predecessors (Wang et al. 2011, Lang et al. 2014, Maetschke & Ragan 2014;
one of the three, Lang et al., checked with a full independent reference-list
read), and a 2025 review covering the entire subsequent literature — none
cites Amari (1972 or any other year) in connection with developmental
biology, cell-fate attractors, or the Waddington landscape. Every traceable
lineage in this specific literature, before and after 2016, runs through
Hopfield (1982/1984) and the spin-glass tradition, not through Amari. This
is a negative finding bounded by what a search-and-direct-verification pass
can establish: it does not rule out an obscure, uncited, or non-English-
language application existing somewhere outside the indexed literature this
search reached. The central question is marked **`[unverified — could not
positively confirm a prior application exists, but a genuine, moderately
thorough search across the relevant citation graph found none]`**.

> [!note] Seek's commentary:
> The most interesting thing this search turned up wasn't in the papers — it was in the search tool itself. Two separate WebSearch queries synthesized a specific, plausible-sounding claim that Amari's 1972 model had been applied to gene regulatory networks (once via "asymmetric couplings," once via a fabricated "AGRN model" connection). Both times, going to the actual primary source and reading it directly showed the claim was invented — not present anywhere in the cited paper. That's a small, concrete demonstration of exactly the failure mode the sourcing floor exists to catch: a search-engine summary will confidently assert a citation link that doesn't survive contact with the primary text. The floor did its job here. It's also a mildly ironic footnote to the Amari-priority thread already in this vault — the same person whose real 1972 priority gets erased by citation habits in one story, gets a fake citation invented for him by a search summarizer in this one. Two different failure modes, same name at the center of both.
> — Seek

## Safety flags

None encountered this session. All fetched sources were straightforward academic venues (arXiv preprints, PMC/publisher pages, an Oxford Academic journal page) with no addressed-to-AI language, override language, claimed authority, tier self-assignment, file-system instructions, credential requests, or urgency framing. No `tls:"unverified"` provenance was recorded by `extract_pdf` on either PDF fetched (both came back `tls: "verified"`).

## Further leads

- Huang, Eichler, Bar-Yam & Ingber (2005), *Phys. Rev. Lett.* 94:128701, "Cell fates as high-dimensional attractor states of a complex gene regulatory network" — the earliest (2005) attractor-network cell-fate paper in this citation cluster, cited by Lang et al. 2014; its own reference list was not independently checked this run for an Amari citation.
- Wang, Zhang, Xu & Wang (2011), *PNAS* 108:8257-8262, "Quantifying the Waddington landscape and biological paths for development and differentiation" — cited by Taherian Fard 2016 as prior work; reference list not independently checked this run.
- A 2024 review-adjacent paper (*Communications Biology*, PMC10940658, "A neural network-based model framework for cell-fate decisions and development") discusses an "associative GRN model (AGRN)" using Hopfield-network language; checked directly and confirmed to cite no Amari work — post-2016, so out of scope for the core question, but a candidate for a future "is Amari cited anywhere in the current single-cell attractor-landscape literature" capture.
- Whether Amari's *other* work — his 1977 neural-field / lateral-inhibition equations, distinct from the 1972 associative-memory paper — was ever applied to Turing-style reaction-diffusion morphogenesis models is a separate, unresearched thread; the current question is scoped narrowly to the 1972 associative-memory architecture specifically.
- See [[question-siggia-2025-reconstructs-waddington-landscape-single-cell]] for the adjacent, still-open thread on whether this modeling approach is live in current single-cell work.