For most of the twentieth century, medical students opened their neuroanatomy textbooks to the same sentence, phrased a hundred different ways: the adult human brain does not make new neurons. Santiago Ramón y Cajal, the Spanish histologist who won the 1906 Nobel Prize for mapping the nervous system’s cellular architecture, had written the doctrine into the field’s foundation. Once development was ended, the founts of growth dried up irrevocably. In the adult, the nerve paths are immutable. Everything may die, nothing may be regenerated.
That sentence held for the better part of a century. Then, in 1998, a paper appeared with an author list led by Peter Eriksson, a Swedish neuroscientist at Sahlgrenska University Hospital in Gothenburg, and Fred Gage of the Salk Institute in La Jolla. The paper reported, in careful language, that the adult human hippocampus makes new neurons.
What the 1998 paper actually looked at
The tissue came from cancer patients, not healthy volunteers. Patients with advanced cancer had received injections of bromodeoxyuridine, or BrdU, a synthetic compound that slots into the DNA of any cell that is copying its genome to divide. Oncologists used BrdU at the time to measure how fast tumors were growing. The patients had consented to donate their brains after death.
Eriksson and his colleagues asked a different question of the same tissue. If BrdU labels any dividing cell, then any neuron born in the weeks or months before death should carry the tracer in its nucleus. The team stained sections of the hippocampus, the seahorse-shaped structure buried in each temporal lobe that handles memory and spatial navigation, and looked for cells that were BrdU-positive and also carried markers of mature neurons.
They found them. Small numbers, in the dentate gyrus, a curl of tissue inside the hippocampus. The cells were freshly dated. They had the shape and molecular signature of neurons. The reporting in The Guardian years later noted that the authors were careful about what the finding did and did not prove. It showed newborn cells. It did not show that those cells were doing anything.
Why the dogma had been so hard to shift
The idea that adult mammalian brains could grow neurons was not new in 1998. Joseph Altman had reported it in rats and guinea pigs in the 1960s using tritiated thymidine autoradiography, an older tracer method. His papers were ignored, then dismissed. Fernando Nottebohm showed that adult male canaries regrow brain regions involved in song production each spring. Birds were one thing. Mammals were another. Primates, and particularly humans, were assumed to be a closed case.
Pasko Rakic at Yale had argued that primate brains were built for stability. New neurons would disrupt circuits that had taken years to wire. The evolutionary logic seemed sound. The experimental evidence in monkeys was mixed.
Eriksson’s paper did not settle the debate. It reopened it, using human tissue and a tracer whose presence in the DNA of a mature-looking neuron was hard to explain any other way.
What the tracer could and could not show
BrdU is not a clean tool. Rakic pointed out, then and later, that the compound can label cells that are repairing damaged DNA rather than dividing, and that dying neurons synthesize DNA in their final hours. A BrdU-positive cell is not automatically a newborn cell. The 2007 review in Nature Reviews Neuroscience by Charles Gross and colleagues walked through these objections in detail, alongside the accumulating counter-evidence from rodent and primate work.
Eriksson’s team had tried to close the loophole by double-staining. A cell had to carry BrdU and also express NeuN or calbindin, proteins found in mature neurons, before it counted. That combination is harder to produce by accident. It is not impossible.
The finding was a first observation, using a tracer with known limitations. It was not a settled account of adult human neurogenesis. It was the beginning of one.
What came after, and where the field went sideways
For roughly two decades, the Eriksson result held up. Other groups reported adult neurogenesis in the human hippocampus using different methods. Jonas Frisén at the Karolinska Institute measured atmospheric carbon-14 from Cold War nuclear tests, which had been absorbed into the DNA of cells born during and after the atmospheric test-ban era, and estimated that new neurons are added to each adult human hippocampus daily. Maura Boldrini and colleagues at Columbia University reported in Cell Stem Cell in 2018 that the process continues into the eighth and ninth decades of life, though at reduced rates.
Then, in the same year, a group at UCSF published a paper in Nature reporting that they could not find newborn neurons in the adult human hippocampus at all. They used different markers, different tissue preservation methods, and a larger sample. Their conclusion was that adult hippocampal neurogenesis, if it exists in humans, is vanishingly rare.
The field split. Two research groups, looking at similar tissue, reached opposite conclusions in the same journal in the same year. Coverage in Discover and elsewhere framed the disagreement as a methods problem, which it largely is: how brains are fixed after death, how long they sit before staining, and which antibodies are used all affect what a microscope can see.
What the more recent work suggests
The methods problem has not fully resolved, but the balance of evidence has tilted back toward the finding Eriksson reported. A 2026 study in Nature using single-nucleus RNA sequencing on post-mortem hippocampal tissue identified transcriptional signatures of immature neurons in adult brains, including in people over 80. Reporting in The Jerusalem Post noted that so-called superagers, adults whose memory in their eighties matches that of people three decades younger, appeared to carry roughly twice the density of these signatures compared with typical older adults.
This is one study, not settled consensus. Single-cell sequencing has its own limitations, and the difference between a cell that expresses immature-neuron genes and a cell that is functionally integrating into a memory circuit remains a real gap. But the trend line in the recent literature suggests that Eriksson and Gage were looking at something real, even if the numbers involved are smaller than the early enthusiasm implied.
National Geographic covered the superager work as part of a broader piece on cognitive aging, and the framing there is worth borrowing: the interesting question is no longer whether adult neurogenesis happens in humans, but how much, where, under what conditions, and whether it matters for anything a person would notice.
What the finding does not mean
The Eriksson paper, and the twenty-eight years of work that followed it, do not license the popular claim that exercise or meditation or a particular diet will grow you a fresh brain. Rodent studies show that voluntary wheel-running increases hippocampal neurogenesis in mice. Human studies show that aerobic exercise is associated with larger hippocampal volume on MRI. These are not the same measurement. A volume change in a scan can reflect many things, including vascular changes, glial cell numbers, and dendritic complexity, not just new neurons.
The honest position is that adult humans appear to make some new neurons in the hippocampus, that the numbers are small, that the rate declines with age, and that the functional significance for memory, mood, or resilience against neurodegeneration is still being worked out. None of that is a prescription.
The tissue Eriksson examined in Gothenburg is still, in a real sense, the reason the question got asked at all.
Cajal wrote that the adult nerve paths were immutable. The slides from the 1998 paper, stained brown at the dentate gyrus, were the first pieces of human tissue to say otherwise.