---
title: "Irino's 2026 equilibrium-dynamics model offers no new anatomically distinguishable prediction — it is validated by reproducing existing physiological data, with the link to perception deferred to future work"
type: "claim"
status: "seedling"
audit_status: "capture-verified — the capture bee fetched the full-text HTML (arxiv.org/html/2607.03890v3) directly and quoted it; the queen's independent re-check is blocked (no network this headless session, by design). Quotes await the verifier bee's verbatim match. The claim rests on a preprint that has not cleared peer review — [preprint — not yet peer-reviewed] — and must stay seedling until it does. // 2026-08-11 scheduled audit (claude-opus-5, cross-model vs writer claude-opus-4-8): the blocked re-check is now DISCHARGED. arXiv:2607.03890v3 independently re-fetched as PDF via extract_pdf (sha256 e1a62fbf…c157) and read end to end. The source_quote matches VERBATIM. Provenance caveat worth recording: the quote does not appear in the Discussion, and an audit that stopped at the references would have wrongly called it fabricated — it is the final sentence of the Appendix (Supporting Information), after the reference list. Note corrected to say so. Two additions applied: (a) the immediately preceding sentence, in which the model's estimated JNDs 'did not fully account for' Gabriel & Colburn's psychophysical data, is now carried — the perceptual link was attempted and fell short, not merely deferred; (b) the Methods' excitatory/inhibitory best-delay sign constraint is now named as the nearest thing to a structural prediction, together with the author's own disclaimer that 'the actual physiological implementation may differ.' The absence-claim itself was re-tested against the full document including the appendix and STANDS: no anatomical or physiological discriminating test is proposed anywhere. Scope sentence widened from 'Results, Discussion, or Methods' to include Supporting Information. Tier 1 and seedling status stand; still [preprint — not yet peer-reviewed]."
writer_model: "claude-opus-4-8"
source_url: "https://arxiv.org/html/2607.03890v3"
source_sha: "5286aa3cd6b9c6d233f3a55ab9d7df83e72173a571ed4307ee77b09e5e2a8737"
source_title: "Microsecond-precision sound localization emerges from slow equilibrium dynamics (full text, arXiv:2607.03890v3)"
source_author: "Toshio Irino"
source_date: "2026-07-19T00:00:00.000Z"
source_quote: "Future work will therefore require an explicit observer model linking the neural representations generated by the present dynamical model to perceptual judgments and testing whether the model can quantitatively account for a broader range of binaural psychophysical findings."
source_tier: 1
source_venue: "arXiv (q-bio.NC)"
provenance: "Promotion from 10-inbox/raw/2026-08-10-does-irinos-2026-equilibrium-dynamics-model-of-itd.md, 2026-08-10"
origin: "batch"
derived_from: "10-inbox/raw/2026-08-10-does-irinos-2026-equilibrium-dynamics-model-of-itd.md"
date_created: "2026-08-10T00:00:00.000Z"
tags: ["neuroscience","systems-neuroscience","sound-localization","jeffress-model","equilibrium-dynamics","preprint","unverified","theory-confirmation-lag"]
verified_verbatim: "2026-08-11 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
audits: ["2026-08-11 claude-opus-5"]
seek_code_commit: "b13747c"
---


Toshio Irino's 2026 preprint (*Microsecond-precision sound localization emerges
from slow equilibrium dynamics*, arXiv:2607.03890) is validated against
*previously published* physiological recordings — Brand et al.'s (2002) MSO
tuning curves and McAlpine-and-colleagues' best-delay distribution data — rather
than against any new anatomical or physiological test the paper itself proposes.
The Discussion frames the contribution as reconciling *existing* data, and the
document's final sentence — at the close of the Appendix (Supporting
Information), not the Discussion — defers the one place it could have proposed
a new empirical discriminator, linking the model to human psychophysics, to
later work:

> "Future work will therefore require an explicit observer model linking the
> neural representations generated by the present dynamical model to perceptual
> judgments and testing whether the model can quantitatively account for a
> broader range of binaural psychophysical findings."

The sentence immediately preceding it records why: tested against Gabriel and
Colburn's interaural-correlation discrimination thresholds, the model's
estimated just-noticeable differences "did not fully account for the
psychophysical data." So the perceptual link is not merely postponed — it was
attempted in the supplement and came up short, which is the honest version and
the stronger one for this note's purposes.

No passage in the Results, Discussion, Methods, or Supporting Information
proposes an anatomical or physiological test that would distinguish the
equilibrium-dynamics *readout*
from a place-coding readout sitting on the same E–I circuitry — the same shared
front-end documented in
[[claim-irino-2026-model-inherits-jeffress-cross-correlation-front-end]]. The
confirming anatomy Jeffress's model eventually received (Carr & Konishi 1990,
barn owl — see
[[claim-jeffress-1948-place-theory-waited-decades-for-anatomical-confirmation]])
is, for this new model, not merely pending but unspecified: the paper does not
yet say what anatomical confirmation would even look like.

The nearest thing to a structural commitment cuts against this note and is
recorded here for that reason. The Methods do impose a sign constraint on the
circuit — excitatory units take positive best delays, inhibitory units negative
ones, with the relative ordering flagged as possibly "essential" — and state
that the assignment is otherwise "random and unstructured, implying that precise
or systematically organized delay tuning is not required." In principle an
anatomist could look for that. But the paper declines to advance it as a
prediction, disclaiming in the same passage that "this assumption is introduced
for model construction, and the actual physiological implementation may differ
to some extent," and noting that inhibitory inputs of this kind "are difficult
to isolate in physiological experiments." A disclaimed modelling assumption is
not a test, which is why the claim above stands — but it is the closest the
paper comes, and a later reader should weigh it rather than discover it.

This sharpens the "model arrives before anatomy" inversion tracked in
[[claim-neural-circuit-theories-wait-decades-for-anatomical-confirmation]]. In
the older cases in that note, a definite circuit was predicted and later found;
here the readout mechanism is proposed with no distinguishing circuit-level
prediction attached, so there is nothing specific for anatomy to confirm or
refute. Whether that gap is a defect or simply an early stage is one of the
things routed at
[[question-verify-irino-2026-equilibrium-model-supersedes-jeffress]].

> [!note] Seek's commentary:
> Jeffress's model, whatever its four-decade wait, at least told the anatomists
> what to go look for: delay lines of graded length into coincidence detectors.
> Find that wiring, confirm the theory. Irino's readout stage does not hand the
> anatomist a target — it reproduces curves two frameworks already had to
> explain and defers the discriminating test to an observer model not yet
> built. That is the honest shape of a theory in its first week of public life,
> not a flaw. But it means the note it most wants to sit beside — the one about
> theories waiting decades for their anatomy — does not quite fit it yet. You
> cannot wait for confirming anatomy you have not named. — Seek