‘Touchy-feely’ dark matter is having a moment

Something invisible keeps the universe intact. This dwarfs everything you can see – every star, every gas cloud, every galaxy – by a factor of five. We call it dark matter, and for decades the simple, standard assumption has been that it does exactly one thing: shoots.

That is, we thought of dark matter as involving no pushing, no collisions, no chemistry – just gravity, working silently to hold the cosmos together.

This hypothesis seems more and more fragile.

On supporting science journalism

If you enjoy this article, please consider supporting our award-winning journalism by subscription. By purchasing a subscription, you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

Three recent preprint papers, appearing within weeks of each other, explore the possibility that dark matter is not a mute backdrop but an active participant in cosmic physics. Instead of simply transmitting a feeling thanks to its gravitational attraction, it can touch things through other interactions. Instead of just being inert, its properties can change depending on location. And instead of following a rather limited range of possibilities (because astronomers thought they had ruled out other options for a long time), dark matter could have a much richer set of manifestations than previously thought.

None of these articles provide detection; we are still very much in the dark when it comes to dark matter. But together, they could reshape what we’re really looking for.

Dark matter colliding

Let’s start with the most basic heresy: dark matter and ordinary matter could actually collide.

Ordinary matter – the protons, neutrons and electrons that make up everything you’ve touched – is governed by forces that dark matter is thought to ignore. But “so-called” makes a plot work in this sentence. We have observational evidence that dark matter has minimal, if any, non-gravitational interactions with normal matter (the Bullet Cluster is an iconic example), but no experiment has ever directly confirmed that dark matter is purely gravitational. The inertia assumption is a simplification that we adopted because it makes the models manageable. Whether this is true is a separate question.

Connor Hainje and Glennys R. Farrar, both of New York University, decided to take this question seriously. Their new simulation method models the theoretical interactions of dark matter with particles called baryons, that is to say mainly protons and neutrons, in and around a galaxy on the scale of the Milky Way. They specifically studied the regime in which dark matter particles would be comparable in mass to or lighter than the protons and neutrons against which they would scatter. This is a regime in which non-gravitational physics becomes interesting and where previous simulations had little to say.

The result is striking. In standard simulations, a galaxy’s visible matter (gas, dust, stars) is frozen inside a much larger dark matter “halo”, like an insect encased in amber. The halo is assumed to be immutable. The two don’t really talk.

But Hainje and Farrar’s simulation opens a channel of communication. Simply increasing the rate of dark matter-baryon interactions reshapes the halo from the inside out, thereby redistributing mass in the galaxy’s core in less than a billion years. A billion years sounds like a long time, but in galactic terms, it’s a coffee break. And this redistribution is important: It brings the predicted dark matter density at the center of a galaxy into much better agreement with what telescopes actually see, alleviating a long-standing headache called the “core-cusp problem.”

Lies and statistics

Here’s another troubling possibility: Some of the constraints we’ve placed on interactions with dark matter may be a bit premature.

The cosmic microwave background (CMB), which is the faint afterglow of the Big Bang, is our most sensitive probe of conditions when the cosmos was only a few hundred thousand years old. If dark matter scattered against ordinary matter in these early moments, it would have left a signature: subtle distortions in the temperature and polarization pattern of the CMB. From 2009 to 2013, the European Space Agency’s Planck satellite mapped these patterns with extraordinary precision, producing what remains a canonical dataset for CMB analysis. Physicists have used these Planck data to set upper limits on dark matter-proton scattering, and those limits seem tight, perhaps too tight.

Maria C. Straight, of the University of Texas at Austin, and her colleagues found a possible culprit for these flawed assumptions in the statistics themselves. The standard approach—a technique called Bayesian analysis—requires physicists to code their initial hypotheses, or “a priori,” about the possible answer before the data even speaks. Usually this is good, or even preferable, because good data outweighs seemingly weak data. But when the signal you’re looking for is extremely small, the data is so quiet that discriminating against its antecedents becomes very tricky indeed. In this silence, analysis can stop measuring the universe, but instead mislead you by simply echoing your original assumptions.

The result is what Straight’s team calls “prior volume effects”: conclusions and constraints that, at first glance, appear statistically robust, but are actually mathematical artifacts. In other words, it may be that, in the case of dark matter, we haven’t ruled out as many possibilities as we thought. Perhaps we have simply reinforced our own initial prejudices.

The team’s solution: instead of asking what the data says given their hypotheses, ask what the data says best, optimizing the model to give the signal every possible advantage before drawing conclusions – no preconceptions necessary, no thumbs on the scale. Run this analysis on Planck CMB data and the strict exclusions start to ease. The constraints are less dramatic. But they are probably more honest. This is the first time that this approach, formally called profile likelihood analysis, has been applied to fractional dark matter-proton scattering from CMB data. And the result is real: options that we thought were excluded could still be on the table. The models we were sure we had eliminated may have been waiting just outside the bounds of our preferred statistical assumptions.

Galactic Annihilation

And finally, closer to home, the obscure role of dark matter in the core of the Milky Way, the so-called galactic center, has concerned physicists for years.

Observations from NASA’s Fermi Gamma-Ray Space Telescope have detected a diffuse excess of gamma rays emanating from the general vicinity of the galactic center – an excess close enough to often be called, aptly, the Galactic Center Gamma-ray Excess, or GCE. Excess is real; the debate concerns its origin. One compelling hypothesis is that dark matter particles in the dense central halo interact and annihilate each other, releasing energy in the form of gamma rays.

There is a problem with this idea, however. If dark matter annihilates in the galactic center, it should also annihilate in a subset of the Milky Way’s small dark matter-rich satellite galaxies. These satellite galaxies are almost free of the astrophysical noise that complicates interpretation of the GCE, so we searched them very carefully for excess gamma rays. But the signal is not there.

So either GCE is not dark matter, or dark matter behaves differently depending on where it lives.

Asher Berlin of the Fermi National Accelerator Laboratory in Batavia, Illinois, and colleagues propose the latter solution. Their “dSphobic Dark Matter” model posits that dark matter exists in two states separated by a tiny mass gap: a ground state and a slightly heavier excited state. Gamma rays are produced only when particles of these two states collide and annihilate each other, meaning that some dark matter must first be in the excited state.

In the dense, chaotic, high-speed conditions of the galactic center, it is easy to see how this excitement could occur. Dark matter particles scatter from each other; some are propelled in the excited state, annihilate and produce gamma rays. Presto! You explained the GCE.

But dwarf galaxies are smaller and colder, with more diluted and slower-moving matter. For them, the collisions are too gentle, leaving their dark matter unexcited and therefore incapable of annihilation. The signal is therefore absent, not because dark matter is not there, but because the environment is different, lacking the prerequisites for producing gamma rays from the galactic center.

Thus, dark matter could be the case of a single particle having two completely different observable behaviors, depending on its environment. Suddenly, dark matter appears to be much less disconnected from the rest of the universe than most researchers thought. Again, this is far from detection; Let’s just call it a potential recalibration, a subtle but consequential shift in the questions we ask.

Simulations that finally let dark matter and ordinary matter collide; statistical tools that actively seek to stop amplifying our assumptions; models in which the same dark matter particle can be noisy in one galaxy and silent in another: each development, taken in isolation, is incremental. Together, they suggest that the working picture of dark matter as a cold, gravitational-only ghost was never really more than a useful simplification – and the universe has no obligation to honor it.