Fish Under Lifelong Surveillance Reveal That Aging Happens in Sudden Steps, Not Gradual Decline

By midlife, an animal's everyday behaviors can signal how long it is likely to live. That is the central finding of a study from Stanford's Wu Tsai Neurosciences Institute, published in Science, that placed 81 short-lived fish under continuous surveillance from early adulthood until death and then asked what their daily routines revealed about the process of aging.

The answer was more structured than anyone expected. Aging, at least in these fish, does not unfold as a smooth decline. It proceeds through sudden transitions between relatively stable stages - more like a tower of blocks that periodically collapses and restructures than a steady downhill slide.

The Truman Show, but for fish

Most aging studies compare groups of young animals with groups of old ones. These snapshots blur how aging unfolds within individuals and how differences between them emerge over time. Postdoctoral scholars Claire Bedbrook and Ravi Nath wanted something different: continuous observation across entire adult lifespans.

They turned to the African turquoise killifish, one of the shortest-lived vertebrates studied in laboratories. With lifespans of four to eight months, killifish compress the aging process into a timeframe that makes lifelong surveillance feasible. Despite their brevity, they share key biological features with longer-lived vertebrates, including a complex brain.

The researchers built an automated system in which individual fish lived in separate, camera-monitored tanks. Every moment of their lives was recorded. The result: billions of video frames from 81 animals. From those recordings, the team extracted detailed data on posture, speed, rest patterns, and movement, identifying 100 distinct behavioral syllables - short, recurrent actions that form the basic vocabulary of how a fish moves and rests.

Diverging paths visible by early midlife

After following each fish through its entire lifespan, the researchers grouped animals by how long they ultimately lived and then looked back to see when behavioral differences first appeared. By early midlife - 70 to 100 days of age - fish destined for shorter or longer lives were already behaving differently.

Some of the clearest signals involved sleep. Short-lived fish began sleeping during the day as well as at night relatively early in life. Long-lived fish kept their sleep mainly nocturnal. Long-lived fish also swam with greater vigor during daylight hours and reached higher speeds when darting around the tank.

These differences were not merely descriptive. Machine-learning models trained on just a few days of behavioral data from middle-aged fish could forecast lifespan. The behavior itself contained predictive information about future health and longevity.

Rapid transitions, not slow drift

Perhaps the most surprising finding was the temporal structure of aging. Most fish underwent two to six rapid behavioral transitions, each lasting only a few days, separated by longer stable periods lasting weeks. Animals generally progressed through these stages in sequence rather than bouncing back and forth.

Bedbrook describes the pattern as the opposite of what the team expected. Rather than a gradual, continuous process, aging involves long stretches of stability punctuated by brief periods of rapid change. The researchers liken it to a Jenga tower: many blocks can be removed with little visible effect, until one removal forces a sudden restructuring.

This stepwise pattern echoes emerging evidence from human studies showing that molecular features of aging shift in waves, particularly during midlife and older adulthood. The killifish data add a behavioral dimension to what may be a conserved feature of how vertebrate aging works.

Molecular hints from the liver

To connect behavior to biology, the researchers examined gene activity across eight organs in adult fish at an age when behavior could reliably predict future lifespan. Rather than focusing on individual genes, they looked for coordinated changes across gene groups involved in shared biological processes.

The clearest differences appeared in the liver, where genes involved in protein production and cellular maintenance were more active in fish on shorter aging trajectories. These molecular patterns provide a hint that internal biology is shifting alongside the behavioral changes, though the study does not establish which drives which.

From fish tanks to wearable data

The killifish results raise an obvious question: could the same principles apply to humans? The study's senior authors - geneticist Anne Brunet and bioengineer Karl Deisseroth - note that daily behaviors like movement patterns and sleep quality, now routinely captured by wearable devices, might contain similar predictive signals about how aging is unfolding in individuals.

But significant caution is warranted. Killifish live in isolated tanks under controlled laboratory conditions with minimal genetic diversity. Human aging unfolds across decades in endlessly variable environments, shaped by diet, stress, social relationships, medical interventions, and thousands of other factors. The specifics of killifish aging - which behavioral syllables, which transition timing, which organ-level changes - are unlikely to translate directly to people.

What may translate is the principle: that aging has a structured architecture with detectable early signatures, rather than being an undifferentiated decline. Testing that principle in longer-lived, more complex organisms will be the work of coming years. Both Bedbrook and Nath will open their own laboratories at Princeton University in July 2026 to pursue these questions.

For now, the study offers something rare in aging research: a complete, individual-level behavioral record of how 81 vertebrates aged and died, and a demonstration that behavior alone - observed early enough - can predict the outcome.