Sterile neutrinos: an explanation must pass several tests

A sterile neutrino lacks the ordinary weak interaction

A sterile neutrino field carries no Standard Model gauge charge, so it lacks the direct weak interaction that active neutrinos have. Mixing can still connect it to processes we can detect. The name describes a property in theory; it is not proof that such a particle exists.

Neutral singlet fields appear in explanations of neutrino mass at widely different mass scales. A very heavy state in a seesaw model and a light extra state proposed for short-baseline oscillations are different experimental hypotheses. How they are produced and what they leave in a detector can differ greatly.

The three known active flavors already explain a large body of oscillation measurements. An extra state has to improve the description of a specific observation without spoiling the rest. Adding a particle to a diagram does not by itself make inconsistent observations agree.

Sources: [1] · [2]

The anomalies differ from one another

LSND reported electron antineutrinos appearing in a source of muon antineutrinos. MiniBooNE saw an excess of electron-like events in a Cherenkov detector. The two experiments differ in event selection, backgrounds and how well they identify particles, so the shared word excess covers two different observations.

Gallium experiments calibrated with radioactive sources, including BEST, extracted fewer capture products than predicted. Reading that deficit as oscillation requires knowing the source activity, the chemical recovery and the capture cross section. BEST's two target zones also test whether the deficit depends on position, not only its overall size.

The reactor rate anomaly rests on a predicted antineutrino yield. Segmented short-baseline detectors can instead look for spectral changes that vary with distance. A revised flux calculation can change the overall deficit without producing an oscillation pattern. These mismatches cannot simply be added up as interchangeable votes for a new state.

Sources: [3] · [4] · [5] · [6]

Appearance and disappearance are linked

In the simplest model, three mostly active states plus one extra mass state, short-baseline appearance depends on how much electron and muon flavor mix into the extra state. The same mixing also changes how many electron and muon neutrinos survive. Any proposed appearance signal therefore predicts disappearance that can be checked separately.

The appearance amplitude is the product of the two mixing fractions. For small mixing, it is tied to the product of the two disappearance amplitudes. A large appearance signal with no disappearance in either channel is impossible in this model, because both depend on the same parameters.

Global fits compare these linked predictions with the data, including flux, cross-section and detector uncertainties. More elaborate models can change the relation, but they must name the added states or interactions and make their own testable predictions. Changing the theory to escape an exclusion does not confirm the new version.

Sources: [2] · [1]

Independent experiments test the same hypothesis

MicroBooNE's 2025 search, published in Nature, excludes at 95 percent confidence the explanation of the LSND and MiniBooNE anomalies by a single light sterile neutrino. It uses two accelerator beams with different flavor mixes. With a single beam, electron-flavor appearance and disappearance can mask each other and weaken the test.

The result applies to that specific oscillation explanation, not to sterile particles of every mass and mixing. Reactor experiments such as STEREO and PROSPECT test electron-antineutrino disappearance with different sources, distances and backgrounds. Agreement among such different tests says more than repeating one detector's analysis.

KATRIN looks for a heavier mixed state as a change in the shape of the beta spectrum, a test based on kinematics instead of travel distance. Comparing its result with the gallium or reactor regions requires the same convention for mass splitting and mixing, and each experiment's own statistical method.

Sources: [7] · [8] · [9] · [10]

The original anomalies still need an explanation

The original data remain after a candidate model is ruled out. The discrepancy may come from a background process, an error in the prediction or new physics with different signatures. Each possibility has to be tested against the actual measurements, not against a memorable anomaly name.

Every exclusion has a limited reach. Very fast oscillations may average out, very slow ones may barely develop, and very small mixing may be too weak to see. Outside that reach, seeing nothing is not a measurement of absence. The reach depends on the energy response and the spread of baselines as well as on sample size.

As of 11 October 2026, when the sources for this article were checked, the cited searches do not establish a sterile neutrino. The useful question is still which linked predictions survive independent data. Theoretical motives and open anomalies are worth discussing, but confirmation needs a reproducible signal with a coherent explanation.

Sources: [7] · [2] · [9]

Try it in the Lab

Primary sources & revision

1. Particle Data Group · Neutrino masses, mixing and oscillations (2025)
2. Gariazzo et al. · Global one-extra-state short-baseline analysis (2017)
3. LSND Collaboration · Electron-antineutrino appearance evidence (2001)
4. MiniBooNE Collaboration · Electron-like excess (2018)
5. Barinov et al. · Results from the Baksan Experiment on Sterile Transitions (2022)
6. Giunti, Li, Ternes & Xin · Reactor antineutrino anomaly in light of recent flux model refinements (2021/2022)
7. MicroBooNE Collaboration · Light sterile search with two beams (Nature 2025)
8. STEREO Collaboration · Uranium-235 spectrum and sterile-neutrino test (Nature, 2023; arXiv v3, 2024)
9. PROSPECT Collaboration · Final PROSPECT-I Short-Baseline Oscillation Search (2024 submission; 2025 publication)
10. KATRIN Collaboration · Sterile search from 259 days (Nature 2025)

First published 2026-10-11; last revised 2026-10-11. Original explanatory text and diagrams by Sandbox Physics. Illustrations are schematic; no experimental event records are reproduced here. This is an independent educational publication, not an official Nobel or experiment collaboration publication.