Light-Switched Neurons and Second-Generation Planets

The 2026 Nobel Prize in Physiology or Medicine goes to Karl Deisseroth (Stanford), Peter Hegemann (Humboldt University of Berlin) and Georg Nagel (University of Würzburg) for optogenetics. The award follows the 2013 Brain Prize and the 2021 Lasker Award for the same work. Berkeley neuroscientist Hillel Adesnik calls it "a matter of when, not a matter of if."

How it works

In the early 2000s, Hegemann and Nagel discovered channelrhodopsins in single-celled algae. These proteins are ion channels that open when light hits them. Other light-activated proteins were already known, but Ofer Yizhar of the Weizmann Institute says none offered the millisecond control that channelrhodopsin does. In a 2005 Nature Neuroscience paper, since cited more than 4,700 times, Deisseroth showed that inserting these proteins into neurons lets researchers fire those neurons with flashes of light.

The key is genetic targeting. Researchers can put the proteins only into cells with a particular genetic marker, or only into cells that project to a particular brain region. Before this, electrical stimulation gave clear results only in neatly layered structures such as the hippocampus, cerebellum and retina. Adesnik says that in the cortex, where axons run everywhere, "you can't just put a microelectrode... and really know what you're stimulating." The toolbox now holds dozens of variants: fast and slow versions, red-shifted ones that respond to yellow light so two cell populations can be controlled at once, and silencers that block neurotransmitter release at axon terminals.

When Hubble observed the white dwarf HS 0209+0832 in 1999, its spectrum contained roughly 100 absorption features that nobody could identify. Jamie Williams, a doctoral candidate at the University of Warwick, went back to that archival data with an updated database of atomic lines. He found that many of the mystery features matched niobium. The result, published in Nature Astronomy, is the basis for the team's claim that the star may host a second-generation planet: one that formed from material the star shed as it died, not from the leftovers of its birth.

Why niobium points to a dying star

A white dwarf is the exposed core of a low-mass star that has used up its fuel and lost its outer layers. Niobium is heavier than iron. Elements in that class are not made by ordinary fusion in stellar cores. As team member Nicholas Stone of the University of Wisconsin–Madison puts it, they form only "in the exotic conditions that briefly emerge inside dying stars." Co-author Boris Gaensicke says niobium had not been reported in any white dwarf analyzed before. So a large amount of it suggests the matter came from the star's own late-life ejecta.

Volume electron microscopy (EM) images tissue in 3D at very high resolution and across large volumes. It is the method behind connectomics. Davi Bock's lab at HHMI's Janelia Research Campus produced the 21 million images behind FlyWire, the complete wiring diagram of the Drosophila brain published in 2024. Few labs can afford the equipment. Last month the U.S. National Science Foundation gave Bock, now at the University of Vermont, $19 million to design a "National Connectomics and Ultrastructure Volume Observatory." Outside investigators would request projects and receive the data.

What changes

Bock says the handful of groups worldwide with high-throughput volume EM have mostly pushed toward ever-larger single volumes: from a whole fly brain, to a whole nervous system, to a planned whole mouse brain. He wants to spend the same throughput on "n of many samples," meaning smaller volumes compared across time points, experimental conditions, species or stages of a disease.

The grant pays for a five-year design-and-testing phase. The team can then apply for a second five-year award to put the program into operation. Four pilot projects will test the pipeline:

• how synaptic changes from learning fit into hippocampal circuits
• retinal structure in sharks from murky versus clear water
• Toxoplasma gondii crossing the blood-brain barrier
• volume EM combined with expansion microscopy to map protein distribution in protists

Bock's argument is that a pipeline is a separate engineering job from a record-setting volume. Each fly connectome took many years. A pipeline should routinely produce datasets that are segmented and come with analysis tools, which requires a dedicated team. The access model copies co-PI Mark Ellisman's National Center for Microscopy and Imaging Research at UC San Diego, where anyone can submit an inquiry through a web portal. Data would likely be cheap or free during the implementation award. Long-term pricing is undecided: fee-for-service, or costs written into users' grants. Bock prefers "increasing access" to "democratize." He points out that the infrastructure stays expensive and only the data gets cheaper.

Roughly 5,000 years ago, shortly after Upper and Lower Egypt were unified, the first rulers of the new kingdom may have taken their servants with them in death. Roselyn Campbell, an anthropology researcher at Purdue University, re-examined skulls from burials around royal tombs at Abydos. In a study published October 5 in Frontiers in Environmental Archaeology, she reports that many of the dead show trauma consistent with being killed.

An old theory, finally tested on the bones

Flinders Petrie excavated the site in the early 20th century. Hundreds of graves surround the tombs of King Djer, King Djet and Meretneith, a woman who probably ruled Egypt at times. Petrie concluded that these graves held retainers who were killed when their king died so they could keep serving him in the afterlife. His main evidence was that some gravestones carried titles such as "seal-bearer." He never found fatal injuries on the skeletons, though, so the sacrifice idea has been debated ever since. He reburied most of the remains, but 46 skulls ended up in the Duckworth Collections at the University of Cambridge.

Campbell looked at those 46 skulls with modern forensic methods and found that 18 showed trauma that was likely fatal. The key evidence is linear fractures, which Petrie had put down to damage after burial. In the study, Campbell explains that these fractures form when the skull hits, or is hit by, a fairly large object with low overall force. Because the skull is round, the force spreads across the vault of the skull. The fractures don't always kill straight away. They often cause bleeding or lesions inside the skull, which can lead to unconsciousness and death if no surgery relieves the pressure.

In June, mathematicians met in six cities (London, Toronto, Los Angeles, New York, Berkeley and Cambridge, Mass.) to put together a list of 50 important open problems. The request came from Epoch AI, which benchmarks artificial intelligence systems, and it had one condition: there had to be a way to check any proposed solution automatically. That condition makes the list a test for AI systems as well as for people.

The timing is deliberate. Epoch's Greg Burnham notes that AI models have recently been making progress on major problems. OpenAI says it has solved the Navier-Stokes Millennium Prize Problem, and it says an unreleased internal model has settled more than 100 open questions. Burnham guesses that "at least a few" of those are on Epoch's list. By late September, some problems had been solved, including some landmark ones.

A sample of the list

Sum of three cubes (open). The question is whether x³ + y³ + z³ = 114 has integer solutions. Researchers strongly suspect it does, but the smallest solutions probably have around 30 digits. In 2020 Andrew Booker and Andrew Sutherland settled the case of 42 using 1.3 million computing hours on volunteers' home computers. Burnham estimates that the same brute-force approach for 114 would cost $100 million or more. Checking an answer, though, is just arithmetic.

Apéry-style irrationality (rumored solved). In 1978 Roger Apéry proved that ζ(3), the sum of 1/n³, is irrational. He did it by squeezing ζ(3) between two infinite series that converge at exactly the right rate. He never explained how he found those series. The odd zeta values (ζ(5), ζ(7) and so on) are still unproven. Burnham says a "credible" proof for ζ(5) is circulating, but it has not been confirmed.