It’s 2:34pm on a Wednesday. You’ve been at the screen since 1pm. Before that, you were at the screen from 9 to noon with a fifteen-minute walk to grab lunch, which you ate at the desk while reading email. You are now staring at a paragraph you’ve read three times. You can feel your eyes moving across the words but the words aren’t going anywhere. You think you might be hungry, but you also ate at 12:30, so probably not. You consider getting coffee. Instead you open a new tab and check the news for ninety seconds, then close it and look at the paragraph again. It still doesn’t land.
This isn’t a discipline problem. This is your nervous system telling you, with the only language it has, that the focus tank is empty and you have been running on the warning light for forty minutes.
The 90-minute cycle isn’t optional
In the 1950s, sleep researcher Nathan Kleitman noticed something while studying sleep stages: the brain doesn’t only run on a 24-hour circadian cycle. It also runs on a much shorter cycle of roughly 90 minutes, alternating between higher and lower arousal states throughout the day, not just at night. He called this the Basic Rest-Activity Cycle (BRAC), and laid it out in detail in Sleep and Wakefulness (University of Chicago Press, 1963). The original observation was about sleep architecture — REM and non-REM cycle every ~90 minutes during the night — but Kleitman proposed, correctly, that the same ultradian rhythm continues during the day. It just expresses as fluctuating focus and energy instead of dream stages.
What this means in practice: roughly every 90 minutes during the day, your brain wants to shift out of high-focus mode. Not because you’re weak. Because the underlying physiology is cycling. Pushing through the trough doesn’t extend your focus window — it just degrades the quality of work in the trough and burns into the next cycle, so the cycle after that is worse than it should have been.
The performance research lines up with this neatly. K. Anders Ericsson’s seminal 1993 paper “The Role of Deliberate Practice in the Acquisition of Expert Performance” in Psychological Review, studying elite violinists at the Berlin Music Academy, found that the top performers practiced in focused blocks of about 60–90 minutes, with deliberate breaks between blocks, and rarely exceeded 4–5 hours of true deliberate practice in a day. Ericsson’s data showed that the elites were not the ones grinding for ten-hour stretches. They were the ones practicing intensely for ~90 minutes, recovering, and then practicing intensely again — and they accumulated more total quality work because each block was at near-peak quality.
The mechanism is mundane. Focused cognitive work depletes neurotransmitter pools — dopamine, norepinephrine, glutamate — faster than they replenish during the work itself. The replenishment requires a low-activity window: a walk, a non-screen pause, a conversation that isn’t work, even a few minutes of looking out a window. When you skip the window and keep working, you’re not getting more work done. You’re getting more time spent at the desk, which is not the same thing. The actual quality of cognition past the 90-minute mark falls off measurably, and the residue of the depleted state persists into the next block, which is now also degraded.
The 2:30pm wall is not a personal failing. It’s the second BRAC trough of the day arriving on time, and it’s telling you that 1pm to 2:30pm at the desk should have been one block, not two-thirds of a longer one.
What to schedule tomorrow morning
The Pomodoro technique (25 minutes on, 5 off) is a real tool, but it’s the wrong tool for this. Pomodoro is for tasks you don’t want to do — a forced timer to bully you through admin work, undergraduate reading, anything where the resistance is to starting. The 25-minute block is calibrated to be short enough to overcome the resistance.
Ultradian work (~90 on, ~15-20 off) is the opposite tool. It’s for work you actually want to do well — strategic thinking, writing, analysis, design, the things where the quality of the output matters more than the quantity of minutes. The 90-minute block is calibrated to match the natural focus cycle, not to bully you past it.
The practical version: tomorrow morning, set one 90-minute block on a single task. Calendar it. Treat it like a meeting you can’t move. After the block, take a real 15–20 minute break — and “real” matters. A break is not “scrolling on the phone at the desk.” That’s just switching the task while keeping the cognitive system active. A real break has three properties: stand up, leave the screen, and let the mind drift. Walk around the block. Make coffee slowly. Sit on the floor and stretch. Stare out a window. Talk to your roommate or partner about nothing important. The point is to give the depleted neurotransmitter pools a chance to recover and to let any unresolved task threads from the block settle on their own.
Doing two real 90-minute blocks before lunch will produce more quality output than five hours at the desk would. This is not a motivational claim. It’s what the deliberate-practice data shows in domains where the output is measurable.
A few common mistakes when people first try this:
Skimping on the break to “save time.” This produces a worse second block and you end up netting less work, not more. The break is doing work — physiological recovery — even though it doesn’t look like work.
Using the break to do other cognitive work. Answering Slack, reading the news, checking a long thread — these all keep the cognitive system loaded. The recovery requires actual disengagement. If you can’t trust yourself to step away from the laptop, leave the laptop in the other room while you take the break.
Choosing the wrong task for the block. The 90-minute block is for work that benefits from sustained attention. Email is not that work. Status meetings are not that work. The block is for the one thing on your list where uninterrupted thinking actually changes the outcome.
Trying to do four blocks in a day right away. Two is plenty for a starting point. Most people who haven’t held this kind of focus in years can do one good ultradian block before lunch and one in the early afternoon, with a real lunch in between. Four is an Ericsson-elite day. Don’t aim there in week one.
If your focus block keeps getting hijacked by Slack pings and quick lookups, the foundational fix is closing every tab except the one for the task and treating switching as the actual enemy — without that, the 90-minute number is theoretical. And if 90 minutes feels too long because you can’t get a single uninterrupted block in your job, the realistic adaptation is the 50-minute defended block at a predictable time, which is closer to what most office environments will actually let you keep.
Tomorrow morning. One 90-minute block on the task that matters most. One real 15-minute break after. See what the rest of the day looks like.