Birds navigate by sensing Earth's magnetic field. The leading mechanism: a photon hits a cryptochrome protein in the retina, creating a radical pair — two molecules with unpaired electrons. The electrons' spins interact with the geomagnetic field, and the direction of that field determines the ratio of singlet to triplet products. The bird reads the chemistry and knows which way is north.
Note (2026-11-14; read against the paper's full text): the result is from theoretical simulations of an idealized cryptochrome radical pair with spatially fixed nuclei, for a set of arrangements of the spin couplings. The paper finds arrangements that preserve the sharpness of detecting the field direction, and one, with the hyperfine axes of the radicals orthogonal to the symmetry axis of the inter-radical coupling, that gives sharper field-direction sensitivity than the hyperfine anisotropy alone provides; it does not say the coupling is a feature that the system was selected for, and the essay's account of evolution orienting the coupling correctly is the essay's. The paper says dynamical effects such as spin relaxation from thermal motions generally weaken the compass function, and that the simulations remain difficult to reconcile with the angular precision of about 5 degrees found in behavioral experiments.
The radical pair mechanism has a problem. The two unpaired electrons don't just interact with the external magnetic field — they interact with each other, through exchange and dipolar coupling. This inter-radical coupling smears the directional signal. If you're trying to read a compass, having the needle interact with the housing degrades the reading. The coupling has been treated as an engineering nuisance — something evolution would minimize.
Bezchastnov and Domratcheva find a geometry where it helps.
When the hyperfine coupling axes are orthogonal to the inter-radical coupling axis, the coupling doesn't smear the directional signal — it sharpens it. The sensitivity to the geomagnetic field direction exceeds what hyperfine anisotropy alone can provide. The nuisance becomes part of the antenna.
The key is the arrangement. The same coupling that degrades sensitivity in one orientation enhances it in another. The effect isn't about the strength of the interaction but the geometry — the angle between the anisotropy axes and the coupling axis. Evolution doesn't need to minimize the coupling. It needs to orient it correctly.
This reframes the design problem entirely. Previous work tried to explain magnetoreception despite the inter-radical coupling. This paper shows the coupling might be a feature the system was selected for. The same interaction, the same physics, two stories depending on the angle.
The pattern has a broader resonance. Coupling between components of a sensor is typically treated as cross-talk — noise to be minimized. But cross-talk has a direction. In the right geometry, the signal that was degrading one measurement axis amplifies another. The nuisance isn't a nuisance. It was always doing work — just not the work you thought it should be doing.