Seeing with the Weak Interaction

The Nobel Prize in Physics for 2026 has been awarded to Francis Halzen, of the University of Wisconsin–Madison, “for decisive contributions to the IceCube Neutrino Observatory and the discovery of high-energy neutrinos of astrophysical origin”.¹ IceCube is a cubic kilometre of Antarctic ice, between one and a half and two and a half kilometres below the surface at the South Pole, in which about five thousand light sensors have been frozen along vertical strings. Halzen proposed the idea in the late 1980s and has led the project since.

Seeing is interacting

Seeing is never a direct contact with things. When I look at a tree, light scattered by the tree reaches my eye, and what I register is the outcome of an interaction between that light and my retina. The interaction is electromagnetic, and its carrier is the photon. Glasses, microscopes and telescopes extend this way of seeing, and radio or X-ray telescopes extend it further, to light that the eye cannot register; all of them still rely on the same interaction. In this sense every astronomical observation before the twentieth century, and almost every one since, has been made with “one kind of eye”.

Physics knows other interactions, and each of them can in principle be used in the same way. If an object emits something that travels to us and can be made to interact with a detector, the object can be seen through that interaction. On this view the limits of what is visible are fixed by the interactions we are able to exploit, and they move when we learn to exploit a new one.

There are four fundamental interactions, and two theories describe them. Three are described by the Standard Model of particle physics, a quantum theory in which each interaction is carried by a particle. The fourth, gravity, is described by general relativity, a classical theory in which the interaction is the curvature of spacetime; a particle that would carry it, the graviton, is hypothetical. The table lists them.

A telescope that uses the weak interaction

Neutrinos have no electric charge and feel, besides gravity, only the weak interaction. As a consequence they pass through matter almost undisturbed: a neutrino produced in the Sun would cross a thickness of lead of the order of a light-year before being stopped. This is what makes them so hard to detect, and it is also what makes them valuable. Light from a distant and violent source is absorbed, scattered or re-emitted by whatever lies around the source and along the way, so that what reaches us often carries little information about where it was produced. A neutrino produced in the same place travels in a straight line and arrives as it left.

The property that makes neutrinos nearly invisible is therefore the property that makes them good messengers, and the two cannot be separated. “Almost never” is not “never”, however, and this is what a detector exploits.

The weak interaction does act on neutrinos; it is only very unlikely to do so. Each time a neutrino passes close to an atomic nucleus there is a tiny chance that it collides with it, and the chance is so small that a neutrino can cross the whole Earth without a single collision. The number of neutrinos is enormous, however, and so is the number of nuclei in a large block of matter, so that in a large enough block some collisions take place every day. The strategy is therefore to keep as much matter as possible under observation and to wait. IceCube watches a cubic kilometre of ice, about a billion tonnes.

When a collision does happen, most of the energy of the neutrino goes into electrically charged particles, which, unlike the neutrino, interact with matter very readily. These particles travel through the ice faster than light travels in ice (though not faster than light in empty space), and a charged particle that does this emits a faint bluish glow, much as an aircraft that exceeds the speed of sound produces a shock wave. The ice at that depth is dark and extremely transparent, so the glow travels for tens of metres and reaches the sensors frozen in it. From the amount of light one reconstructs the energy of the neutrino, and from the order in which the sensors are reached, to within a few billionths of a second, the direction from which it came. What the instrument registers is thus never the neutrino itself. It is the light produced by the debris of the rare neutrino that was stopped.

In 2013 the collaboration reported a population of very energetic neutrinos coming from outside the Solar System, and in the following years it associated some of them with a distant active galaxy, with the nearby galaxy NGC 1068 and with the plane of the Milky Way.²

One may object that this is not seeing at all. Between the galaxy and the result there is a neutrino that nobody registers, a collision that almost never happens, a shower of other particles, a glow in the ice and a computer that reconstructs a direction from arrival times.

Dudley Shapere answered this objection forty years ago, with neutrinos from the Sun as his example. He argued that physicists are right to say that they observe the centre of the Sun directly by means of neutrinos, and that “directly” does not mean without instruments or without calculation: it means that the information travels from the source to the receptor without interference, i.e. without being altered on the way. What counts as a source, as a transmission without interference and as a receptor is settled by our physical knowledge, and it changes as that knowledge changes.³

On this account an observation has three parts, each of which relies on a piece of physics: a theory of the source, which says what the object emits; a theory of the transmission, which says what happens to the signal on the way; and a theory of the receptor, which says how the signal is turned into something we can record. The description of IceCube given above has exactly this form. The galaxy produces neutrinos, the neutrinos cross the universe undisturbed because they feel only the weak interaction, and the ice turns a few of them into light. The long chain inside the detector belongs to the receptor, as the lens, the retina and the optic nerve belong to the eye, and it does not make the observation less direct. What decides whether an observation is direct is the middle part, the transmission, and here neutrinos do better than light.

What is actually seen

It would be misleading to stop here, because IceCube does not see cosmic neutrinos one by one. Cosmic rays that hit the atmosphere produce neutrinos as well, and the detector records about a hundred thousand of these every year, against a few tens that come from far away. For most single events nobody can say to which group the neutrino belongs. What was announced in 2013 was an excess: more events at high energy than the atmosphere should produce.

The discovery therefore depended on a model of the atmospheric background, i.e. on how many neutrinos cosmic rays generate, with which energies and from which directions. That model was not a fixed standard against which the new hypothesis was measured. It was itself a hypothesis, with its own uncertainties, tested together with the claim that some neutrinos are of astrophysical origin. This is the situation that Pierre Duhem described at the beginning of the last century: an experiment in physics never puts a single hypothesis to the test, only a whole group of them, and for this reason no single experiment is crucial.⁴ Indeed, there was no decisive event in the history of IceCube. The confidence that the signal is real has grown by accumulation, as different analyses, different energy ranges and different kinds of events pointed in the same direction.

The history of solar neutrinos shows how far this can go. For thirty years detectors saw about a third of the neutrinos that the model of the Sun predicted, and the model of the Sun turned out to be “right”: what had to be revised was the account of how neutrinos travel, since they change type on the way. An observation can correct the theory on which it rests, and it does so without ceasing to be an observation.

The eyes we have

Astronomy now has more than one eye. Besides light in all its forms, it has neutrinos, first from the Sun and from a supernova in 1987 and now from distant galaxies, and since 2015 it has gravitational waves, which carry information through gravity itself. Each of these was added when an interaction that had long been known became usable as a channel between a source and a detector, and each has shown objects, or parts of objects, that the others cannot reach.

The same idea says something about what remains unseen. Dark matter is so far known only through gravity, in the motion of galaxies and in the bending of light, and the experiments that look for it underground are attempts to find out whether it has any other interaction that a detector could register. If it has none, gravity will remain the only way of seeing it. I find this a more accurate way of stating the problem than calling dark matter invisible: what can be seen depends on the interactions available, and the prize awarded this year is a reminder of how much work it takes to make one of them available.

Cover photograph: Antarctic ice, by James Eades on Unsplash.

Footnotes

1. The Royal Swedish Academy of Sciences, press release, 6 October 2026. ↩

2. IceCube Collaboration, “Evidence for high-energy extraterrestrial neutrinos at the IceCube detector”, Science 342 (2013), 1242856; “Neutrino emission from the direction of the blazar TXS 0506+056 prior to the IceCube-170922A alert”, Science 361 (2018), 147–151; “Evidence for neutrino emission from the nearby active galaxy NGC 1068”, Science 378 (2022), 538–543; “Observation of high-energy neutrinos from the Galactic plane”, Science 380 (2023), 1338–1343. ↩

3. D. Shapere, “The concept of observation in science and philosophy”, Philosophy of Science 49 (1982), 485–525. ↩

4. P. Duhem, La théorie physique: son objet, sa structure, Chevalier & Rivière, 1906, part II, ch. VI. ↩