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

neurosciencesystems-neurosciencesound-localizationjeffress-modelequilibrium-dynamicspreprintunverifiedtheory-confirmation-lag

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.