Ask yourself a simple question: how was your last week, really?
Notice what your mind just did. It didn't replay seven days. It grabbed two or three scenes — probably the most intense moment and whatever happened most recently — and stitched them into a verdict. If Thursday held an awkward meeting and Sunday ended with a small argument, the whole week files itself under bad, even if forty of its waking hours were quietly pleasant. You didn't remember your week. You summarized it, and the summary was written by an editor with a strong agenda.
Psychologists have a name for the rule that editor follows. Once you see it, you can't unsee it — and you can start keeping a second set of books.
The peak-end rule, briefly
In the early 1990s, Daniel Kahneman and his collaborators began asking a strange question: when an experience is over, what do we actually keep of it?
In one experiment, Kahneman, Barbara Fredrickson, and colleagues had participants hold a hand in painfully cold water for sixty seconds. In another trial, the same participants endured the same sixty seconds plus an extra half minute during which the water was warmed slightly — still unpleasant, just less so. Then they were asked which trial they'd rather repeat. Most chose the longer one. Objectively, it contained more total discomfort. But it ended better, and the memory of it was kinder. The paper's title says it plainly: "When More Pain Is Preferred to Less."
Donald Redelmeier and Kahneman then took the question into a hospital, asking patients undergoing colonoscopies (in an era before routine sedation) to rate their pain in real time, minute by minute. Afterward, they asked for an overall rating. The retrospective judgments barely reflected how long the procedure lasted. They tracked two data points instead: the worst moment and the final moments. Everything else — the duration, the accumulated experience — was largely discarded. Fredrickson and Kahneman called this companion effect duration neglect: memory keeps the shape of an experience, not its length.
Peak, plus end, divided by two. That's roughly the formula your memory uses to grade your experiences — including your weeks.
Two selves, two sets of books
Kahneman drew a distinction from this work that became one of the most useful ideas in modern psychology: the experiencing self and the remembering self.
The experiencing self is the one living your life in real time — the one who felt fine at 10 a.m. on Tuesday, enjoyed lunch, got mildly bored at 3 p.m. It lives in moments a few seconds wide and keeps almost nothing. The remembering self is the one who answers questions like "how was your week?" It writes the official history. And crucially, it's the one that decides things: whether the job is good, whether the routine is working, whether last year was a good year.
The trouble is that the remembering self wasn't there for most of it. It works from the peak-end summary. Kahneman called this the tyranny of the remembering self: we make choices on behalf of a narrator who systematically misreports what the experiencing self actually went through.
Why this distorts ordinary weeks, not just lab experiments
A week contains roughly a hundred waking hours. The peak-end rule compresses them into a verdict built from a handful of minutes, and the compression has a predictable tilt.
Quiet, fine hours vanish first. Duration neglect means a long stretch of okay — focused work, an easy dinner, a walk — contributes almost nothing to the remembered week. It wasn't a peak and it wasn't the end, so it's as if it didn't happen.
Negative peaks punch above their weight. Human attention has a well-documented negativity bias — bad events capture and hold processing resources more strongly than good ones — so the "peak" your memory selects is disproportionately likely to be the argument, the criticism, the mistake.
And endings color everything upstream. A rough Sunday evening can retroactively sour a week that was, hour by hour, mostly good. The same mechanism runs in reverse, too: a season that was genuinely draining can be remembered warmly because it ended with a win.
None of this would matter much if the remembered week were just an anecdote. But it's the input to real decisions. Is this job wearing me down? Is the new schedule helping? Was cutting back on evening plans a good idea? When you consult your memory to answer, you're consulting an unreliable narrator — sincere, confident, and wrong in a consistent direction.
The case for a contemporaneous record
Researchers stopped trusting retrospective self-report decades ago, and they built their methods around that distrust. Experience sampling and ecological momentary assessment — pinging people during their actual day rather than asking them to summarize it later — became standard in pain research and wellbeing science precisely because the summaries diverge so much from the moments. Kahneman himself co-developed the Day Reconstruction Method, published in Science in 2004, to get closer to how days actually feel rather than how they're remembered.
The civilian version of all this apparatus is almost embarrassingly simple: a daily log. Two or three lines, written the same day, about what happened and how it actually felt at the time. Not literature. Not therapy. Testimony — a witness statement taken while the witness still remembers.
A log doesn't stop the peak-end rule from operating; nothing does. What it gives you is a second witness whose account your memory can't quietly overwrite.
How to keep a log your memory can't overrule
Write it the same day. A Sunday recap of the week is already the remembering self talking — you'll transcribe the distortion instead of correcting it. Same-day, even same-hour, is the point.
Record the ordinary on purpose. One line about the unremarkable fine stretch — "good deep-work morning, uneventful afternoon" — feels too boring to write. Write it anyway. Those are precisely the hours duration neglect will delete, so they're the most valuable thing you can capture.
Name the feeling at the time. One word suffices: fine, tense, light, flat. You're logging the experiencing self's actual readings, not composing a mood essay.
Run the comparison. At the end of the week, before reading anything back, write your gut verdict: "this week was ___." Then read the seven entries. The gap between the verdict and the record is the finding. It's often startling.
One honest caveat: a log written at 9 p.m. has its own small peak-end tilt — the evening's mood leaks into the entry. Writing a line or two at different points in the day, when you can, spreads the sampling and softens that bias. But even an imperfect same-day record beats a week-old summary by a wide margin.
What people find in the gap
Sometimes the record acquits the week. The version in memory — stressful, scattered — turns out to have been three tense hours embedded in days that were, entry by entry, genuinely good. That discovery changes how much weight you give the next bad Thursday.
Sometimes the record convicts. The job you keep telling yourself "isn't that bad" shows up as drained, tense, flat five days running, week after week. That's not a bad ending distorting a fine stretch; that's the experiencing self filing the same report every day. It's much harder to argue with — and much better grounds for a real decision — than a feeling.
Either way, you've replaced a verdict with evidence. You're no longer the only unreliable narrator in the room.
A record is only as good as its friction
Everything above depends on one fragile thing: actually writing the line, today, while the hour is still warm. Any tool that makes you wait — a spinner, a sync delay, a maze of folders — hands the job back to the remembering self, because "I'll write it later" is exactly how later's editor takes over. That's the problem Pagebox was built around: it opens in under a second, your daily journal is right there waiting, and because it's local-first, there's nothing between you and the entry — it syncs quietly afterward. Two lines a day is a small habit, but it's enough to keep an honest witness on retainer. If you're curious what your remembering self has been leaving out, you can start your record at pagebox.lumenlabs.works.