A little more than two years ago, a junior scientist walked into Paola Arlotta's office at Harvard with images of organoids the lab had left growing. By the rules of the field, organoids that old should not have been alive. The scientist, Irene Faravelli, had analyzed them anyway, and Arlotta told STAT: "I was actually pleased that somebody had the courage to do this."
The images were of cultures more than two years old, and they became the seed of a study published this week in Nature: cortical organoids kept alive more than five years, the longest reported to date. The team grew 34 organoids from stem cells obtained from a blood draw, eraser-sized balls of tissue modeling the cerebral cortex. Some survived nearly six years, and the lab has others at seven, though too few remained for reliable analysis. Most organoid cultures previously died within months, and the field assumed they could only ever mimic early, fetal-like stages of development. The record here is not merely survival. It is that the tissue kept maturing, on a schedule researchers could predict from human brain development itself.
The cells keep their own calendar
The study's centerpiece is a clock. The organoids showed gene-expression patterns that tracked human developmental stages in order: cultures 15 days to 2 months old resembled first-trimester fetal cortex, cultures 3 to 6 months old resembled the second trimester, and past a year they took on newborn-like features. The five-year organoids carried transcriptional and methylation profiles similar to the cerebral cortex of a typical 4-year-old child. Epigenetic clocks trained on human cortical tissue predicted the organoids' ages almost exactly; methylation-based age correlated tightly with time in culture. In the paper's own framing, the cells record the passage of time. The National Institutes of Health, which funded the work, describes the finding as maturation driven by a lifelike developmental clock.
The most striking experiment shows the clock lives inside the cells. When the lab mixed neural progenitors from 9- to 12-month-old organoids with cells from 15-day-old ones, the old cells bypassed the early developmental steps the young cells were performing and produced mature neuronal fates in about two weeks, a sequence that normally takes about two months. They behaved as if they knew how much time they had already spent. The young cells, in the same dish, behaved normally.
Getting there took engineering. Electrical signals in the original five-year organoids began to decline around the one-year mark, until the lab switched to an activity-permissive culture medium with lower glucose and physiological ion concentrations. Organoids grown in the modified medium produced more mature excitatory neurons, greater structural complexity, and network bursting that persisted past two years. Arlotta describes the organoids as avatars of the patients the cells came from, and she sees them as a platform for postnatal brain development and for diseases like autism, schizophrenia, and epilepsy, which older models captured poorly.
Outside researchers called the longevity a feat and noted that the cells aged at roughly the same pace as a person's brain. The skepticism about the approach is practical rather than scientific: as Alysson Muotri of UC San Diego put it, "Nobody's excited to keep them that long." Waiting five years is not a workflow, and the field wants to compress the clock, not honor it. Muotri, who grows organoids in his own lab, also cautioned that RNA measurements do not always reflect protein levels, a reminder that these profiles are molecular fingerprints, not reports from inside the tissue.
Five years does not make them a brain
Everyone in this story agrees on what the organoids are not. The researchers state plainly that the organoids are not capable of consciousness and have no sensory input, and that their structure remains far simpler than a real brain's. Electrical bursting in a dish is not experience: there is no organism, no body, no behavior, no world to perceive. Even the closest thing to a sensing pathway produces no feeling. Sergiu Pașca's group built a four-part assembloid that mimics a pain-sensing circuit and reacts to capsaicin, the molecule that makes chili peppers hot, but feeling pain requires a second pathway that registers the unpleasantness, and the assembloid lacks it. A reflex is not suffering.
Philosophers add a category warning. Attributing consciousness to a clump of cortical tissue in a dish commits what philosophers call the mereological fallacy: treating a part as if it had the properties of the whole. On the view many philosophers of mind hold, that consciousness belongs to whole organisms and not to detached tissue, the question of organoid consciousness is close to a category error before it is an empirical one.
The molecular resemblance to a four-year-old cortex is exactly what makes the organoids valuable as models of postnatal development, and it is also the property that makes some ethicists uneasy, because resemblance of function is not the same as feeling. That distinction is the contested core of the debate, and both sides of it are argued below on their own terms.
The debate was built for organoids that die in months
The ethics conversation about brain organoids is older than any organoid that has lived a year. In 2018, Nita Farahany, Henry Greely, and colleagues warned in Nature that as lab-grown brain models drew closer to replicating brain function, difficult questions would follow. The assumptions underneath the guidance written since then are that organoids are small, short-lived, immature, and fetal-equivalent, with properties bounded by the dish. The International Society for Stem Cell Research's guidelines assert that today's organoids are devoid of consciousness, a 2024 editorial in Nature Reviews Bioengineering noted, before arguing that building global policy on that dismissal would rest on unstable ground and warning that organoids sit in a regulatory grey area between human-participant research, laboratory animal research, and stem cell research, where laws remain scarce.
The governance gap is concrete. Greely, who helped organize a November 2025 meeting at Asilomar on organoid ethics, told The Scientist, "There's no regulatory setup that's looking at the organoids as organoids." Organoids fall between the human-subjects review that covers research on people and the animal review that covers research on animals. Consent procedures were designed for tissue that stops, and Farahany argues that research needs mechanisms of continual governance rather than one-time broad consent, because organoids can live for years and donors cannot anticipate every future use.
Every ethical comfort of the past decade was tied to a limit: brevity, immaturity, fragility, fetal-equivalence. The five-year record breaks all four at once. That is the event the field has not yet metabolized.
The two strongest cases, both on the table
The research community's case, stated as its advocates would state it, is that there is no evidence of consciousness in organoids and no plausible mechanism for it, that the architecture is missing, and that over-attribution is a category error with real costs: misleading patients waiting on treatments, eroding public trust, and letting inflated "mini brain" claims do long-term harm, a concern researchers raised pointedly at Asilomar. The burden of proof, on this view, belongs to those making the positive claim, and the empirical properties remain far from it. The Asia Pacific Neuroethics Working Group, a consensus of scientists, ethicists, and legal scholars, urged the debate to track actual properties, describing the differences between organoids and ordinary cell cultures as quantitative changes on a spectrum rather than categorical steps, and warning against ethical red herrings.
The ethicists' counter-case, stated as its advocates would state it, is that the burden question runs the other way. The entity is on a trajectory, and every reassurance is time-indexed: "not conscious" describes what the organoid is today, not what it will be at six years or ten. The properties that justified comfort are precisely the ones being outgrown. Detection is the hard problem. Andrea Lavazza and Marcello Massimini proposed in the Journal of Medical Ethics adapting the perturbational complexity index, a measure used in clinics to assess consciousness in patients who cannot communicate, to organoids. Later work in the Cambridge Quarterly of Healthcare Ethics concluded that no validated measure exists for judging whether an organoid's measurable activity indicates consciousness. If we cannot yet tell whether a line has been crossed, the precautionary argument holds, then the time to decide what would count as crossing it is now, not later. And the question will not wait for the laboratories: survey research found that the more people attribute consciousness to organoid-based systems, the more they support obligations of care for them, so public perception may answer the question if policy does not.
The evidence is thin in both directions, and it should be said plainly. No one has produced evidence of anything like consciousness in organoids, and the burden-shifting case is built on extrapolation from molecular profiles that even the researchers caution are not the whole picture.
This analysis takes no position on whether brain organoids could ever warrant moral status, or on whether any research practice should change. Both cases above are stated as their advocates would state them. The observation here is narrower and mechanical: neither side has a schedule.
Nobody has scheduled the checkpoint
The field knows the gap. At Asilomar, held 50 years after the meeting that produced the first guidelines for genetic engineering, the organizers deliberately made no rules; participants agreed that the public belongs in the discussion, and proposals ranged from soft regulation with periodic review to a standing international monitoring system, which a Science opinion piece the same month called the minimum needed. The Asia Pacific group's recommendations likewise call for proportionate responses and adaptive frameworks that track the evidence.
What does not exist is a trigger. No one has defined what would count as the moment to revisit the assessment, who would decide, what evidence would move the consensus, or which body would act. Review is event-driven: it happens when a paper like this one lands, not when a plan says it should. The five-year organoid is the closest thing to a checkpoint the field has produced, and it arrived with no process attached. The debate is being resumed because the organoids got older, not because anyone built the mechanism the debate needs.
The cells keep time, and the governance does not. Whatever one believes about organoid consciousness, both sides should want the checkpoint: researchers need it to protect the trust their field depends on and to know where the line actually is, and ethicists need it to test their frameworks against data instead of hypotheticals. And the clock will not wait. Five years is now the floor, not the ceiling, and the field is already working on making maturation faster rather than slower.
Primary sources
- STAT, "Brain organoids, kept alive more than five years, matured like human brains," Megan Molteni, August 19, 2026, for the opening scene, the Arlotta quote, and the fact of the Nature publication.
- The study, "Human brain organoids record the passage of time over multiple years in culture," Faravelli, Antón-Bolaños, et al., Nature 2026, for the growth of the organoids, the developmental staging, the methylation clock, and the chimeroid findings, and the National Institutes of Health news release for the funding and developmental-clock framing.
- The 2018 Nature essay by Farahany, Greely, Hyman, et al. for the founding ethics framing, the Nature Reviews Bioengineering editorial for the ISSCR position and the regulatory grey area warning, and The Scientist's report for the Greely quote, the Farahany governance argument, and the Pașca assembloid detail.
- Science's report on the Asilomar meeting, Lavazza and Massimini's Journal of Medical Ethics article for the perturbational-complexity proposal, the Cambridge Quarterly of Healthcare Ethics article for the absence of validated measures, and the Asia Pacific Neuroethics Working Group recommendations for the spectrum framing and proportionate-responses position.