A man in his late fifties came to see me about his memory. He was still working full days, but he told me that nothing new seemed to stick. What caught my attention was not the memory complaint itself. It was the way he described his week. A colleague had been short with him in a meeting, a client had not returned a call, and his wife had gone to bed early, and by the time he finished recounting all of it the three events had merged in his telling into a single impression that people were pulling away from him. He had carried a diagnosis of major depressive disorder for six years and had been through three antidepressants.
I have come to think that this blending of experience is not a side effect of low mood. It is closer to the center of what depression does to the brain, and a paper published on August 21 in Nature Medicine gives us the cellular picture behind it.
What the Columbia Team Actually Measured
Madeleine Peng, Jialin Jiang, and colleagues working in the laboratory of Maura Dupont at Columbia University Vagelos College of Physicians and Surgeons and the New York State Psychiatric Institute examined 495,037 cell nuclei from the human hippocampus, drawn from 11 donors with major depressive disorder and 19 control donors and collected shortly after each donor's death. For every nucleus they recorded the activity of every gene, checked whether the cell's proteins had been altered, and mapped where in the hippocampal circuit that cell was positioned.
What they found is that adult neurogenesis, the production of new neurons in the dentate gyrus, stalls in major depressive disorder. The neurogenic lineage is present and the cells that should be maturing into new granule neurons are there, but they are not progressing the way they do in donors without depression.
Two details govern how far this can be carried. The impressive number here is nuclei, not people; thirty donors is a small cohort by any standard. And the depressed donors were unmedicated at the time of death, which was a deliberate design choice rather than an accident of sampling. Antidepressant exposure is itself known to alter hippocampal neurogenesis, so what this paper describes is untreated depression, and applying it to a patient already on medication is an extension the study was built to avoid.
This is the first time the stall has been shown in human tissue at single-nucleus resolution, and it lands inside an argument the field has been having for close to a decade. In 2018, Sorrells and colleagues reported that they could find no immature neurons in the human dentate gyrus past early adolescence, while Boldrini and colleagues at Columbia reported finding them in donors into their seventies. That second group is worth naming carefully, because Maura Boldrini now publishes as Maura Boldrini Dupont and is the senior author of the new paper. The 2026 dataset does not settle the 2018 argument, and it comes from one of the parties to it. What it does instead is describe the lineage well enough to say what goes wrong with it in a specific disease.
Why Pattern Separation Links New Neurons to Mood
The hippocampus performs a computation called pattern separation. It stores experiences that closely resemble one another as distinct memories rather than letting them collapse into a single averaged impression. Work in rodents has established that this computation depends on adult-born neurons. In people the evidence is thinner: patients who received focal radiation to the hippocampus for a benign meningioma, a treatment that likely depletes adult neurogenesis without anyone being able to measure it directly, developed a pattern-separation deficit that was both one-sided and temporary, recovering by twenty-four months.
Dupont describes the clinical consequence about as directly as it can be described. You are out to lunch with a friend who is tired and does not say much. With pattern separation intact, you remember it as its own event. With pattern separation impaired, that lunch mixes with prior memories of feeling rejected, and the conclusion you arrive at is that she is upset with you. She is careful about how far this can be pushed, and says in the same breath that we do not yet know the complete mechanism, particularly in humans.
That may be the mechanism behind what my patient described, though nothing in this study lets me confirm it in him, and he sits outside its cohort in the one way that matters most. What I can say is that when a patient reports that the only version of an event available to them is the painful one, I have started to treat that as information about how the hippocampus encoded the event rather than as a cognitive distortion to be argued with.
Depression May Not Be One Disease
The neurogenic stall was not the only finding. The trisynaptic circuit, which is the hippocampus's main route for laying down emotional memories, showed molecular disruption across several systems at once: genes governing the formation of new connections between neurons, genes handling cellular energy, and genes managing the transport of cargo inside cells. The depressed tissue also carried signatures of inflammation and cellular stress. Some of the dysregulated genes carry variants that genetic studies have already tied to major depression, while others had been altered epigenetically, their activity turned up or down by exposures such as stress, aging, and chemicals rather than by inherited sequence.
Dupont's reading of that breadth is that depression may not be just one disease, and her stated goal is to reclassify it by molecular features the way oncology reclassified cancers by cellular characteristics instead of by anatomical site. That reclassification is what made targeted cancer therapy possible.
Serotonin deficiency was never a satisfying account of depression, and what has been missing is a replacement specific enough to act on. Impaired neurogenesis, disordered energy metabolism in hippocampal neurons, and neuroinflammation are three separate candidate targets, each of them modifiable, and a given patient may have one or all three. I have written before about the metabolic and microbial contributions to depression, and this paper puts a cellular address on that argument.
What This Changes in the Clinic
We cannot yet measure hippocampal neurogenesis in a living person, and that should be said clearly. Nothing in this paper produces a test I can order tomorrow.
What it changes is where I look. The mechanisms thought to regulate neurogenesis in the dentate gyrus are largely the ones we already assess when a patient with a mood disorder and cognitive complaints comes through our Intensive Brain Health Program: inflammatory markers, homocysteine and the methylation pathway, thyroid function, sleep architecture, insulin resistance, and the exposures that govern BDNF, aerobic conditioning chief among them. Each of these has a documented relationship with hippocampal plasticity in preclinical models, the human work is uneven and in several cases absent, and each of them is modifiable in a patient today. That combination is what makes them worth measuring even while the mechanism is unsettled.
The nutritional side belongs in the same conversation, though it needs to be stated carefully. Omega-3 status and B-vitamin sufficiency bear on the metabolic conditions a maturing neuron needs in order to integrate into a circuit, which is part of why we pay attention to brain nutrition alongside pharmacology, and I should disclose that Action Potential Supplements is a sister company to The Neurogenesis Project. NAD+ availability is a plausible further lever on those same conditions, but the human work is early, and the one controlled trial in mild cognitive impairment improved hippocampal perfusion without improving memory. None of this is a treatment for depression and I would not present it as one. It is the substrate that any treatment has to work in.
Cognition is the asset that produces nearly everything else a person builds, which is the premise underneath the Neuroeconomy. A brain that has stopped making new neurons has lost part of the machinery it uses to update itself, and that loss shows up in a patient's life as an inability to see a new situation as new. Restoring that capacity is a more precise goal than lifting mood, and this paper is the clearest map we have yet of where to aim.
Illustration created with AI image tools.
This article was drafted with the assistance of AI writing tools, then reviewed, edited, and approved by Dr. Sean C. Orr, M.D., who holds full editorial responsibility for its content.
Sources: Peng MS, Jiang J, Polizzi L, et al. Dysregulated adult hippocampal neurogenesis in major depressive disorders. Nature Medicine. Published August 21, 2026. DOI: 10.1038/s41591-026-04571-8 · Columbia University Irving Medical Center. Depression Stalls Formation of New Brain Cells. August 21, 2026. · Sorrells SF, Paredes MF, Cebrian-Silla A, et al. Human hippocampal neurogenesis drops sharply in children to undetectable levels in adults. Nature. 2018;555:377-381. · Boldrini M (Boldrini Dupont M), Fulmore CA, Tartt AN, et al. Human hippocampal neurogenesis persists throughout aging. Cell Stem Cell. 2018;22(4):589-599.e5. · Krotkova OA, Galkin MV, Danilov GV, et al. Delayed, transient, lateralized deficits in pattern separation and pattern completion after focal hippocampal irradiation in humans. Scientific Reports. 2025. DOI: 10.1038/s41598-025-29852-z (PMID 41398354)