In 1998, a landmark study by Eriksson and colleagues demonstrated what many neuroscientists had long considered impossible: the adult human hippocampus was generating new neurons. The finding, based on BrdU labeling in terminal cancer patients who had received the thymidine analog for tumor staging, upended a central dogma of neuroscience—that the adult brain was a post-mitotic organ, incapable of producing new neurons after early development. For two decades, adult hippocampal neurogenesis became one of the most productive research programs in neuroscience, spawning thousands of studies linking the birth of new granule cells in the dentate gyrus to learning, memory, stress resilience, and antidepressant action.

Then, in 2018, a carefully executed study by Sorrells and colleagues in Nature reported finding virtually no evidence of young neurons in the adult human dentate gyrus. The paper sent a tremor through the field. Subsequent studies offered contradictory results, with some groups reaffirming robust neurogenesis into the eighth decade of life and others corroborating the sharp postnatal decline. The dispute has become one of the most consequential methodological controversies in contemporary neuroscience.

What makes this debate so instructive is not merely the disagreement over cell counts, but what it reveals about the limits of our tools, the dangers of cross-species extrapolation, and the fragility of conclusions drawn from post-mortem tissue. Understanding where the evidence actually stands requires a careful dissection of animal model findings, the technical challenges unique to human studies, and the functional implications that hinge on the resolution of this controversy.

The case for adult neurogenesis rests on an extraordinarily well-developed body of evidence in rodents and, to a lesser extent, non-human primates. In mice and rats, the subgranular zone of the hippocampal dentate gyrus reliably produces new granule cells throughout the lifespan. These cells can be labeled with BrdU, retroviral vectors, or genetic reporter lines, and their maturation can be tracked through well-characterized stages—from radial glia-like progenitors to doublecortin-expressing neuroblasts to functionally integrated mature neurons. The phenomenon is not subtle; under optimal conditions, thousands of new cells are generated daily in the rodent dentate gyrus.

Critically, animal studies have established that adult neurogenesis is not merely constitutive but dynamically regulated. Voluntary running robustly increases the proliferation of neural progenitors. Chronic stress and elevated glucocorticoids suppress it. Environmental enrichment promotes the survival of newborn neurons that would otherwise undergo apoptosis. These findings are among the most replicated in behavioral neuroscience, and they have been demonstrated across multiple laboratories, strains, and experimental paradigms.

The functional integration of new neurons has been confirmed through electrophysiology. Young adult-born granule cells in rodents display a period of heightened synaptic plasticity—lower thresholds for long-term potentiation, higher input resistance, enhanced excitability—that distinguishes them from mature granule cells. This critical period, typically spanning the first four to six weeks after cell birth, has been proposed as a mechanism through which neurogenesis contributes to hippocampal circuit function. Ablation studies using irradiation, antimitotic drugs, and transgenic approaches consistently show impairments in specific hippocampal-dependent tasks when neurogenesis is suppressed.

However, the leap from rodent to human is substantial and often underappreciated. The rodent hippocampus is structurally simpler, the dentate gyrus proportionally larger relative to total brain volume, and the timescales of neuronal maturation fundamentally different. Non-human primate studies, conducted primarily in macaques, show neurogenesis that is present but dramatically reduced compared to rodents—by some estimates, one to two orders of magnitude lower in rate. This primate attenuation already raises questions about how much further the decline might extend in the human brain, which is far larger and has a much longer lifespan.

The animal literature, then, provides a proof of principle but not a direct prediction for humans. It establishes the biological plausibility of adult neurogenesis, identifies the molecular pathways that govern it, and suggests what functions new neurons might serve. What it cannot do is tell us whether the human dentate gyrus retains a meaningful neurogenic capacity beyond childhood, a question that requires entirely different methods and confronts entirely different technical obstacles.

Takeaway

Robust evidence across species establishes that adult neurogenesis is biologically real and functionally significant in animals, but the magnitude of cross-species decline from rodents to primates should temper any assumption that human brains behave identically.

The central difficulty of studying neurogenesis in the living human brain is straightforward: we cannot label dividing cells with BrdU or retroviral tracers in healthy individuals. The Eriksson et al. study circumvented this by exploiting a clinical circumstance—patients receiving BrdU for cancer diagnostics—that is no longer ethically replicable. Subsequent human studies have relied almost entirely on immunohistochemistry of post-mortem tissue, staining for markers of immature neurons such as doublecortin (DCX) and polysialylated neural cell adhesion molecule (PSA-NCAM). And it is here that the controversy ignites.

The 2018 Sorrells et al. study examined 59 human brains spanning fetal development to adulthood. Using rigorous stereological methods and multiple immature neuron markers, the team reported a precipitous decline in dentate gyrus neurogenesis during childhood, with virtually no young neurons detected in adults over 13 years of age. The findings were striking in their starkness and methodological care. Just one year later, however, Moreno-Jiménez and colleagues published a study in Nature Medicine reporting thousands of DCX-positive cells in the dentate gyrus of neurologically healthy adults up to age 87. The critical variable, they argued, was tissue processing—specifically, the duration of fixation in formaldehyde, which progressively degrades the epitopes that antibodies target.

This fixation-time hypothesis strikes at the heart of the debate. Extended post-mortem intervals and prolonged formalin fixation are known to reduce immunoreactivity for many antigens, and DCX is particularly vulnerable. Sorrells and colleagues used brain bank tissue with highly variable fixation durations, some spanning years. Moreno-Jiménez used tissue fixed for less than 48 hours. The disagreement, therefore, may not be about biology at all, but about whether the technical conditions of the two studies are comparable. Each group has accused the other of methodological confounds—either false negatives from degraded antigens or false positives from nonspecific antibody binding.

Adding further complexity, a 2019 study by Boldrini et al. in Cell Stem Cell found persistent neurogenesis in the human hippocampus using a different methodological approach, reporting stable numbers of progenitors and immature neurons across adulthood but declining angiogenesis and quiescent progenitor pools in aging. Carbon-14 birth-dating studies by Spalding and colleagues, leveraging the atmospheric spike in radiocarbon from nuclear bomb testing, estimated a turnover rate of approximately 700 new neurons per day in the adult human hippocampus—a number that is large in absolute terms but vanishingly small relative to the billions of existing neurons.

The field is thus in a genuinely unresolved state. The positive and negative findings cannot be trivially reconciled, and both sides present internally consistent data. What has become clear is that the answer hinges as much on methodological standardization—fixation protocols, antibody validation, quantification approaches, and the interpretation of marker expression in aging tissue—as on any underlying biological question. This is a case where the tools of investigation have themselves become the primary subject of debate.

Takeaway

When opposing findings are both methodologically defensible, the controversy is no longer about what is true in the brain—it is about what our techniques are capable of revealing, a distinction that demands epistemic humility.

Even if adult hippocampal neurogenesis persists in humans, a separate and equally important question concerns its functional relevance. In rodents, adult-born neurons are thought to contribute to pattern separation—the computational process by which the dentate gyrus transforms overlapping inputs from entorhinal cortex into orthogonal, distinct representations. Young granule cells, with their heightened excitability and sparse connectivity, are particularly suited to encoding novel features that distinguish similar experiences. This is not merely theoretical; ablation of neurogenesis in mice specifically impairs discrimination between overlapping spatial or contextual cues while leaving other forms of hippocampal memory intact.

The implications extend to mood regulation. Virtually all classes of antidepressants, from SSRIs to electroconvulsive therapy, increase hippocampal neurogenesis in rodent models, and some behavioral effects of antidepressants are abolished when neurogenesis is blocked. The neurogenesis hypothesis of depression posits that reduced production of new neurons contributes to the hippocampal volume loss observed in major depressive disorder, and that restoring neurogenesis is at least one mechanism of antidepressant action. This hypothesis remains influential but has been difficult to test directly in humans, precisely because of the measurement challenges described above.

Memory flexibility—the ability to update, recombine, and contextualize stored representations—is another proposed function. Computational models suggest that ongoing neurogenesis introduces a controlled form of forgetting into hippocampal circuits, clearing old patterns to make room for new learning. This is not pathological forgetting but adaptive reorganization, a mechanism that prevents catastrophic interference between old and new memories. Akers et al. demonstrated in 2014 that increasing neurogenesis in adult mice induced forgetting of previously acquired contextual fear memories, lending empirical support to this counterintuitive idea.

But here is the critical question: do these functions scale? If adult humans generate 700 neurons per day—Spalding's estimate—this represents roughly 0.004% of the total granule cell population turning over annually. Can such a small fraction of cells exert a meaningful influence on circuit computation? Some modelers argue yes, that the unique physiological properties of young neurons allow them to punch above their numerical weight. Others contend that the human dentate gyrus has evolved alternative mechanisms for pattern separation that do not depend on neurogenesis at all, rendering the addition of new cells functionally redundant.

The resolution of this debate carries significant clinical weight. If adult neurogenesis is functionally meaningful in humans, it represents a therapeutic target for depression, age-related cognitive decline, and neurodegenerative disease. If it is vestigial—a remnant of a process that was once important in smaller-brained ancestors but has been superseded by other plasticity mechanisms—then the extensive pharmacological interest in promoting neurogenesis may be misplaced. The stakes are not merely academic; they shape research priorities, clinical trial design, and the direction of translational neuroscience.

Takeaway

The question is not simply whether new neurons exist in the adult human brain, but whether they exist in sufficient numbers and with sufficient functional specificity to matter—a distinction that separates a biological curiosity from a therapeutic opportunity.

The neurogenesis debate is a microcosm of a broader challenge in human neuroscience: reconciling what we can demonstrate in animal models with what we can detect—and what matters—in the human brain. The rodent evidence is clear, the primate evidence suggests decline, and the human evidence remains genuinely contested on technical grounds that neither side has fully resolved.

What is needed is not another immunohistochemistry study but a convergence of methods—single-nucleus RNA sequencing of fresh human tissue, advanced imaging modalities capable of detecting immature neurons in vivo, and computational frameworks that can model the functional impact of very low rates of cell addition. Until these tools mature, honest uncertainty is the most defensible position.

The neurogenesis story reminds us that paradigm-defining claims require paradigm-defining evidence, and that the gap between a beautiful hypothesis and a confirmed human reality is often measured not in logic but in the precision of our instruments.