Toshio Irino's new model explains microsecond sound localization from slow neural dynamics

A new paper by Toshio Irino, submitted to Science on July 4, 2026, presents a radical reinterpretation of how the brain achieves microsecond-precision sound localization. The classical Jeffress model from 1948 posits that neurons act as precise delay lines to detect interaural time differences (ITDs). But this struggles to explain how slow neural dynamics (millisecond timescales) can yield microsecond sensitivity. Irino's model replaces place-coding with a dynamical systems approach: ITD is encoded as a stable equilibrium state of a neural population. Excitatory and inhibitory interactions across frequency channels generate a population signal that evolves toward an equilibrium corresponding to the true ITD. The model does not require explicit delay lines or precisely timed inhibition.

Remarkably, despite relying on relatively slow temporal dynamics, the model achieves microsecond-level precision — matching the best psychophysical and physiological data. It also reproduces frequency-dependent best-delay distributions observed in animals. The work suggests that the brain can exploit equilibrium dynamics to extract hyper-precise timing information from sluggish neurons. This could have implications for understanding other temporal processing tasks (e.g., speech, echolocation) and inspire bio-inspired audio processing algorithms.