Psychology says people who find it almost impossible to sleep when something is unresolved — a conversation left hanging, a decision unmade — may not simply be anxious by nature; the brain tends to keep unfinished experiences mentally active, replaying them when the distractions of the day disappear

There is a specific flavour of 2am that I have come to know intimately. The fan is turning. The street below has gone quiet in the way Bangkok streets only manage for about ninety minutes a night. And I am lying there running the same eleven seconds of a conversation that stopped mid-sentence roughly nine hours earlier.

Not a fight. Not even a disagreement. Someone started saying something. I opened my mouth to answer. A phone rang. We never picked it back up. Nothing actually happened.

My brain treated it like an unpaid tax bill.

For a long stretch of my thirties I assumed this meant my wiring was faulty, that I was the anxious one in the room, the guy who could not let things go. The wiring is fine. It is doing precisely what it evolved to do, on a schedule that suits it and ruins me.

Your brain files paperwork at night

Sleep is not a shutdown. One of its main jobs is building long-term memory, and the mechanism is stranger than most people realise. Patterns of neural firing from your day get reactivated during deep sleep, replayed, and gradually reorganised into longer-term storage across the cortex. Neuroscientists call it active systems consolidation, and replay is the mechanism that drives it.

In plain English: your brain spends the night pulling the day’s files out, reading them again, and deciding where they go.

Nobody sends you a receipt for this. You are unconscious while it happens, which is rather the point. The trouble starts earlier in the evening, before the night shift has clocked on, while you are still awake enough to interfere.

An unfinished thing stays switched on

In 1927, the psychologist Bluma Zeigarnik published a study showing that interrupted tasks lodged in memory more stubbornly than completed ones. The finding took her name, and productivity books have been quoting it ever since.

Fair warning before I lean on it: that original effect has replicated unevenly over the decades, so treat the 1927 version as a promising hunch rather than settled law. The modern work is sturdier.

In 2011, two psychologists ran a set of experiments on what unfinished goals actually do to a person. Participants with an activated, unfulfilled goal had intrusive thoughts while trying to read something unrelated, showed heightened mental access to goal-related words, and performed worse on an anagram task that had nothing to do with the goal. The open loop was quietly eating capacity.

Here is the part that changed how I run my evenings. The same study found that when participants were allowed to write a specific plan for the unfinished goal, the interference vanished altogether, as completely as actually finishing the thing would have. The effect tracked how earnest the plan was: interference cleared specifically for the ones who genuinely intended to follow through.

What your head seems to want is for the information to live somewhere other than exclusively inside your head. Closure is optional. Storage is the thing.

What keeps you awake is not the filing

This is where the popular version of this idea gets sloppy, and I want to be straight about it instead of selling you something neater than the evidence.

The consolidation happens after you are out cold, well after the ceiling-staring starts. What actually keeps you up is the loop still sitting open while you are conscious, or more precisely, how you are holding it open.

A research team followed 59 employees across twelve weeks, matching 357 pairs of Friday and Monday observations. Unfinished tasks at the end of the working week predicted worse sleep over the weekend. The path ran through what they called affective rumination, the hot circling kind of thinking. Problem-solving pondering, the cooler kind that actually works on the problem, was associated with less sleep impairment, though that particular pathway did not reach significance.

Same unresolved thing. Two completely different nights, depending on which gear you are in.

I find that distinction more useful than anything else in this literature, because it is testable at 1am with no equipment. Am I working on this, or am I just visiting it repeatedly? New sentences mean you are working it. The same eleven seconds on a loop means you are just visiting.

Write it down and be irritatingly specific

A sleep lab at Baylor tested the practical version on 57 adults wired up for polysomnography, which is the proper overnight measurement and not people guessing in the morning. Five minutes of writing before lights out. Half wrote a to-do list of things coming up, half wrote about what they had already finished.

The to-do list group fell asleep significantly faster. Better still, specificity mattered. The more detailed the list, the shorter the time to sleep, and the trend ran the opposite way for the group journaling about completed tasks.

Fifty-seven people and one night each, so hold it loosely. But it costs five minutes and a notebook, which is a good ratio for any intervention.

What I have landed on, after some trial and a lot of error in three cities: vague lists do nothing. “Sort out the thing” is not a plan, it is the same anxiety with better handwriting. “Email the agent Tuesday morning about the lease renewal date” is a plan. A specific line actually moves the file out. A vague one does not.

The tidy story is tidier than the science

You will find articles claiming your sleeping brain deliberately prioritises unfinished business because it knows you will need it tomorrow. There was a well-known 2011 study suggesting sleep selectively strengthens memories you expect to use again.

It has since failed to replicate. A 2021 attempt across two experiments found the expected overall benefit of sleep on memory, but no sign that knowing about a future test made any difference to overnight consolidation.

So the honest version is less cinematic. Your brain replays the day. Unfinished things stay cognitively active and cost you something while they do. Whether the night shift specifically hunts down your loose ends is not established, and I would rather tell you that than hand you a satisfying mechanism that might be wrong.

When someone else is holding the ending

The hardest ones are not tasks. A conversation that stopped without landing. A decision sitting on somebody else’s desk. An answer that will arrive whenever it arrives.

You cannot plan your way to a resolution that is not yours to give. What you can plan is your next move, which is a different object entirely and one you fully control. Not “how will this turn out” but “what will I do on Thursday, and what will I do if the answer is no.”

I learned a crude version of this at fourteen, sent to boarding school in another country, waiting weeks for letters from people who had no idea I was counting the days. You cannot make the post arrive. You can decide what you are doing between now and Friday. It is a smaller comfort than closure and considerably more available.

Write both branches down. Then the open file has a plan attached, even though the ending does not.

When it is more than a busy filing cabinet

Everything above is about the occasional rough night. If this keeps happening most nights for months, if you are exhausted and it is bleeding into your work and your relationships, that is not a notebook problem and I am not the person to fix it. Chronic insomnia is treatable. Cognitive behavioural therapy has a solid track record for it, and a doctor is a better bet than another article.

Ordinary filing at an inconvenient hour is one thing. If the pattern has moved in and unpacked its bags, that is worth taking to someone qualified.

What I actually do now

Notebook by the bed. Five minutes before I put the light out, not in bed with a screen. Anything unfinished gets written as a next action with a day attached, and if it is waiting on somebody else, it gets written as what I will do when they answer.

Some nights it works beautifully. Some nights I am still running the eleven seconds. But I have stopped interpreting those nights as evidence of a defect, which turns out to matter more than I expected. Lying awake annoyed at yourself for lying awake is a genuinely worse experience than the plain version.

Those eleven seconds finally got their ending. Four minutes, that’s all it took. It was nothing, which is exactly what I suspected at 2am and still could not manage to believe.