Five Labs Ran the Same Psilocybin Test. Most Effects Did Not Replicate

Five neuroscience labs at three Bay Area universities ran identical psilocybin experiments on mice and pooled the results, and the effects that survived the comparison were narrow: a short-lived spike in anxiety and a brief dip in fear response, both measurable within minutes of a dose and gone the next day. The findings were published Nov. 2 in Nature Neuroscience by researchers from UC Berkeley, UCSF and Stanford.

The consortium found no replicable reduction in anxiety- or depression-like behavior 24 hours after a single dose, no measurable effect on fear responses a day later, and no impact on social interaction. It also tested whether psilocybin reopens a developmental window for social reward learning in adult mice, an effect reported in earlier single-lab work, and found no evidence that it does.

What the collaboration was built to catch

The study was designed around a problem that has dogged psychedelic research at every level, from rodent behavior to human trials: small experiments produce striking results that later fail to appear. The Berkeley-led team tried to answer a narrower question before anyone tries to explain a mechanism — which of psilocybin's behavioral effects show up in five different rooms, run by five different groups, with different hands handling the animals.

Each lab followed the same protocol. After pooling data across the sites, the picture that emerged was of a drug with reliable acute effects and much weaker persistent ones.

"One of our main findings is that, at least with our experiments and doses used, the persistent effects of psilocybin across these behaviors was modest, if anything," said Odilia Lu, a Berkeley doctoral student and co-first author of the paper. She said the result points to the value of multi-lab collaborations in estimating a true effect size and reducing publication bias, in which positive results from small single experiments skew the literature.

The acute effects held up

All five labs recorded increased anxiety and avoidance behavior in mice 10 minutes after injection, along with a decrease in the expression of fear. Those are the findings the authors say are worth building on, because they are reproducible and therefore can be traced to something in the brain.

"Initial very small clinical studies have shown that psilocybin could be a potential treatment for these diseases like depression and PTSD, but so far we know very, very little about what it actually does to the brain," said Stephan Lammel, a UC Berkeley professor of neuroscience who runs one of the participating labs. "Once we understand what are really the reproducible effects of psilocybin on behavior, then we can study the mechanism and how it affects behavior in the brain."

Andrea Gomez, a Berkeley assistant professor of molecular and cell biology who leads another of the five labs, said the acute results demonstrate that something is happening in the animals' brains, while leaving the location and the downstream consequences open. "Where in the brain, we don't know," she said.

Where the labs disagreed with each other

The individual experiments did not agree, even under identical protocols. Pooled data showed no lasting effect on depression-like behavior, yet two of the labs recorded a significant effect on one of the two assays most often used to judge antidepressants in rodents. The authors treat that disagreement as evidence about the assays rather than about the drug: in the standard forced-swim test, a mouse is scored as depressed when it stops swimming, which is a crude proxy for a condition that in humans involves rumination, anhedonia and months of altered functioning.

"I don't necessarily think we need to move away from focusing on depression, because it seems like psilocybin is most promising for the treatment of the disorder," Lu said. "A lot of the behavioral tasks that we use in mice to assay depression just can't capture the really complicated human."

The field's well-known dependence on set and setting — the idea that a psychedelic experience depends heavily on mindset and environment — cannot be modeled in animals at all, a limitation the authors acknowledge rather than argue around.

Why this lands in the middle of a policy push

The replication paper arrives as psilocybin moves from research curiosity toward regulated medicine. The FDA has been reviewing psilocybin-assisted therapy for treatment-resistant depression, Oregon and Colorado have built state-licensed programs, and a 2026 executive order directed federal agencies to accelerate work on psychedelic treatments, including ibogaine, for serious mental illness. Research on psilocybin for treatment-resistant depression has expanded faster than the underlying behavioral science has been pinned down.

Lammel framed the result as a course correction rather than a verdict on the drug, and as an argument that psychedelic researchers should adopt parallel designs as a matter of routine. "It's an urgent call to the field," he said.

For clinicians and patients, the practical reading is narrower than either the hype or the disappointed headlines. Nothing in the study tests psilocybin in humans, and nothing in it contradicts the small clinical trials that generated the current interest. What it does undercut is the assumption that a single mouse study can establish what psilocybin durably does — a standard that applies equally to the rodent work now cited in grant applications and investor decks for depression and PTSD treatments, where the human evidence remains early.

The paper's authors argue that parallel, multi-site designs should become standard for psychedelic research, on the theory that a field moving this fast can afford to learn which of its findings are real before it builds treatment protocols on them.