Ultradian rhythms: the case for 90-minute work cycles, examined

If you've read more than three productivity books in the past decade, you've encountered the ninety-minute cycle. Work in ninety-minute blocks. Rest for twenty. Repeat. The claim is usually attributed to "ultradian rhythms" — short biological cycles within a twenty-four hour day. The implication is biological, scientific, settled.

Some of that is true. Some of it is the kind of true that becomes false in transit.

1. Where the number actually comes from

Nathaniel Kleitman, the great twentieth-century sleep researcher, proposed in the 1950s and refined through the 1960s a Basic Rest-Activity Cycle (BRAC) — a hypothesis that the brain alternates between higher-arousal and lower-arousal phases on a cycle of roughly ninety minutes throughout the day, with the cycle most obvious during sleep (where it tracks the well-documented NREM-REM cycle).

Kleitman's sleep observations were rigorous. The waking-day extension — that human attention or cognitive performance also oscillates on the same ninety-minute beat — was speculative even in his own writing. He treated it as a hypothesis worth investigating, not a settled fact.

2. What the data actually supports

The sleep ninety-minute cycle is real. Across normal adults, NREM-REM cycles average eighty-five to one hundred ten minutes, with considerable individual variation (Carskadon & Dement, 2011).

Some daytime physiological measures do oscillate on roughly similar periodicities. Heart-rate variability, body temperature, certain hormones. Whether attention and cognitive performance track those oscillations cleanly is much less clear.

The strongest evidence for waking-day ultradian rhythms in cognition comes from a handful of studies in the 1980s and 1990s that found modest oscillations in tasks like vigilance and reaction time. Effects were small. Methodologies varied. Replication has been spotty. The 2015 meta-analytic literature on attention rhythms describes the picture as "suggestive but unsettled."

3. What survives is more modest

Three claims survive reasonably well:

Sustained focused attention degrades after extended periods. This is uncontroversial. Most well-designed studies find performance decrements on demanding tasks within the first ninety minutes, but the curve is dose-response, not a clean cycle.

Breaks restore performance. Also uncontroversial. The optimal break length and frequency are not robustly pegged to any specific number.

Individual variation is large. The 90/20 prescription is a population-level average projected onto an individual rhythm that may not exist in that form. Some people focus best in twenty-minute sprints. Some sustain for two hours.

4. The productivity-book translation problem

Publication after publication has cited Kleitman + BRAC + a few intermediate studies and concluded that humans should work in ninety-minute blocks. The leap from "there appears to be cyclic variation in some measures" to "the optimal work block is ninety minutes" requires several inferential steps that the literature doesn't fully support.

It's not that the prescription is wrong. It's that it's been laundered through claim attribution into something more confident than the evidence justifies. "Ultradian rhythms" is doing rhetorical work it doesn't earn.

The honest version of the advice is closer to this: take breaks before you crash; ninety minutes is a reasonable upper limit for most people on demanding cognitive work; pay attention to your own pattern. That's defensible. It's also not what sells productivity books.

5. What good advice would look like

Tracking your own performance dip is more useful than adopting a population-mean schedule. The skill is in noticing when concentration starts to fall apart — for some people that's forty-five minutes in, for others ninety, for a few two hours. The break afterward should be long enough that you genuinely return restored, not just one minute under your phone's clock.

That's harder to put on a poster. It's also closer to what the science says.