I started this study because I kept losing the middle of things. Films, lectures, arguments with myself in the shower — the beginning stayed, the ending stayed, and the middle quietly evaporated. This is not a new observation. It has a tidy name (the serial position effect) and a tidy curve. What I wanted to know was whether the curve holds up when the thing being remembered is a whole ordinary day.
Method (or: what I actually did)
For 118 days I logged one sentence every waking hour. No editing, no retrospective tidying. At the end of each week I tried to reconstruct the days from memory alone, then compared the reconstruction against the log and scored each hour as recalled, partially recalled, or gone.
- 118 days, 1,642 logged hours
- Weekly recall attempts, always on Sunday evening
- Scoring done blind by hiding timestamps until after the attempt
- One month lost to a phone that fell in a river (documented in Appendix B)
Recall is highest at the edges of the day and collapses through the early afternoon — the same U-shape, drawn in ordinary life instead of a lab.
What the middle actually contains
The forgotten hours were not empty. They were, overwhelmingly, hours of maintenance: replying, tidying, waiting, moving between rooms. The mind seems to treat maintenance as overhead and declines to file the receipt.
| Category | Share of lost hours | Example log line |
|---|---|---|
| Maintenance | 44% | "moved laundry, thought about nothing" |
| Transit | 21% | "bus, rain, someone's headphones too loud" |
| Admin | 18% | "emails. emails. emails." |
| Genuine work | 11% | "draft section 3, going badly" |
| Unclassifiable | 6% | "???" |
Categories assigned after the fact from the one-sentence logs.
“A day is not remembered as a line. It is remembered as two bright ends and a long soft nothing in between.”
A caveat I keep returning to
This is a study with one participant, and that participant knew exactly what the hypothesis was. Treat it as a carefully documented anecdote rather than evidence. I include it because the raw logs are more honest than the conclusion, and because I think self-study deserves to be published messily instead of not at all.
If you have run something similar on yourself, I would genuinely like to see it — even the part where the phone fell in the river.