The Self · Berlin, 1926
Nine people order. He writes nothing down. He is perfect — until the bill is paid, and then the whole table drops out of his head. A young woman at that table spent the next year turning his memory into an experiment. A hundred years later we know which half of it was real.
Take a seat
Act one · The table
You are sitting with your back to the window in a Berlin café, some time in 1926. The table is too small for the number of people at it and everybody is arguing. This is Kurt Lewin's table — he brings his students here from the Psychological Institute, which is housed, absurdly, in the old Imperial Palace, and they talk for hours over coffee that goes cold.
The waiter comes. Eight or nine orders go around the table. He writes nothing down. He is not showing off; this is simply the job. An hour later he returns and recites the whole table back — who had the coffee, who had the cake, who changed their mind — and takes the money.
Then someone at the table, a few minutes later, asks him what they had ordered. And he cannot say. Not "let me think" — genuinely gone, like a word rubbed off a slate. The bill was paid. The table closed. The whole thing went.
He was not remembering the orders. He was remembering the ones that were still open.
Sitting somewhere at that table is a twenty-six-year-old from Prienai, in what is now Lithuania, who has come to Berlin with her husband and a very good ear. Her name is Bluma Zeigarnik. She is about to spend a year building the waiter into a laboratory.
A caution before we go further: this story does not appear anywhere in Zeigarnik's own 1927 paper. It surfaces later, secondhand, in at least two different versions — in one the waiter forgets the total, in the other he forgets who ordered what. Which is a nice joke on everyone involved, since the thing psychology was about to learn is that memory rebuilds rather than replays.
For the next minute you are the waiter. Eight guests order, one after another, and you may not write anything down. Then four of them will pay and leave. Then we will ask you what everyone had.
No hurry, and no penalty. Guess if you have to; guessing is data too.
Every part of this page is skippable. Nothing here is scored except the quiz at the end, and nothing here is stored anywhere you did not put it.
Act two · The laboratory
Zeigarnik's design is beautifully plain. Thirty-two adults, one at a time. Twenty-two little jobs, each meant to take three to five minutes: wind thread onto a spool, fold a sheet of paper into a shape, build something out of cardboard, count backwards, do a multiplication, draw a vase, string beads.
Half of the jobs she lets you finish. The other half she stops — and she does not stop them at a tidy moment. She stops them, in her words, just when you are most engrossed. Then the sheet is swept away and the next job begins.
At the end, with no warning at all, she asks you to name every task you did. Not to do them again. Just to say what they were.
The people in her room named the interrupted ones about twice as often as the finished ones. She wrote the result as a quotient — interrupted recalled divided by completed recalled — and it came out around 2.0, and stayed near 2.0 across four experiments. Lewin had a theory ready for it: starting a task opens a small tension system in you, a Quasibedürfnis, a quasi-need. Finish and it discharges. Get stopped and it stays charged, humming, easy to reach.
A finished task is a closed account. An unfinished one is still owed.
Eight tiny jobs, about fifteen seconds each. Four of them you will be allowed to finish. Four of them a bell will cut off — and we will not tell you which is which until it happens. Do them properly; the effect, if you have it, depends on you actually caring for a moment.
One person doing eight jobs is not a study. Zeigarnik had thirty-two people and twenty-two jobs, and even that turned out not to be enough. This is here so the thing happens in your hands, not so it proves anything.
Sixteen names below. Eight of them happened. Eight never existed. Tick the ones you remember doing.
Act three · The open door
A year after Zeigarnik, another of Lewin's students asked a different question about the same interrupted table. Maria Ovsiankina — born in Chita in Siberia, in Berlin since 1920, later a professor in Connecticut — did not ask what people remembered. She asked what they did.
Same set-up: give someone a small job, interrupt it, take it away. But then simply leave them alone in the room with nothing in particular to do, and watch.
They went back to it. Around 86% of her participants picked the unfinished thing back up without being asked, and the accounts of those sessions are wonderful: people asking "so I don't have to do this one any more?", people noticing exactly how much was left, people hunting for the pieces after the experimenter had quietly hidden them.
Zeigarnik's unfinished tasks sat in the memory. Ovsiankina's got up and walked back across the room.
This one only opens once you have run the bench above — it needs your own interrupted jobs to offer you.
Act four · What the numbers actually say
Here is where a lesson about yourself has to be honest, because the Zeigarnik effect is one of the most confidently repeated facts in popular psychology, and it has been in trouble for seventy years.
The doubts start early. By the 1950s researchers are reporting that they cannot reliably get it. In 1968 Annie van Bergen published a book-length attempt at an exact replication and got a quotient of 0.88 — pointing the wrong way. Twenty of her thirty participants remembered the finished tasks better. She then reviewed the literature and found that fewer than a third of the studies meant to show a Zeigarnik effect actually did.
In 2025 Ghibellini and Meier pooled fifty-nine publications. For the memory effect: a mean-to-mean ratio of 0.99 across thirty-eight papers, an interrupted share of recall of 49.16%, a standardised effect of dz = 0.15. Their sentence is flat and worth reading twice: "The current findings do not support a memory advantage for interrupted tasks."
For Ovsiankina's question, the same meta-analysis found something quite different: across twenty-one samples, 67% of people go back to an interrupted task. Their conclusion — "The Ovsiankina effect represents a general tendency, whereas the Zeigarnik effect lacks universal validity."
The half everyone quotes came apart. The half nobody quotes held.
Which is, if anything, the more useful finding about you. Unfinished things are not primarily a memory problem. You are not walking around with a longer list than you think. You are walking around with a stronger pull than you think — toward the things you were in the middle of.
No waiter appears in the 1927 paper. The anecdote reaches us through Edwin Boring in 1957 and through Alfred Marrow's 1969 biography of Lewin, and the two versions disagree about what the waiter forgot. Treat it as a good story that points at a real question, not as evidence.
Most retellings credit Lewin with spotting the waiter and Zeigarnik with building the experiment. Some hand the observation to Zeigarnik. Nobody at that table wrote it down at the time, so nobody can settle it.
The 2025 meta-analysis found the atmosphere of the experiment moved the result: relaxed rooms gave a ratio of 1.07, neutral 0.96, achievement-flavoured rooms 0.88. In other words, whether unfinished things stick may depend on whether being stopped felt like a rest or a failure.
Her original sessions ran high. The pooled figure across twenty-one samples is 67%. Both are real; the second is the one to carry around.
The plan-making result you will meet in a moment comes from six experiments by Masicampo and Baumeister in 2011, published before psychology's reforms around pre-registration. Their own studies 5A and 5B found plans removed the intrusive thoughts but not the anxiety. Useful, not settled.
Whatever number the bench gave you, it is one person on one morning. If it came out at 2.0 you have not confirmed Zeigarnik, and if it came out at 0.5 you have not refuted her. You have felt the shape of the question, which is the point.
Act five · Turn it around
Everything above is about strangers in a room in Berlin. Now the instrument turns around.
You are carrying open loops right now. Not the big frightening ones — just the ordinary sediment of a life in progress: the message you started and didn't send, the shelf that is still leaning, the language you got eleven lessons into, the drawer, the email, the person you meant to call, the book with the receipt still in it at page 90.
Most of them don't cost much. Some of them cost more than they are worth. And a surprising number of them, once written down, turn out to be finished already — they just never got a stamp.
Add up to ten. Small ones welcome — the leaning shelf counts. For each one, set how hard it tugs at you, then decide where it goes: back on the bench to be finished, into a plan with a when and a where, or stamped closed and off the list.
Skippable, obviously. And keep it to things you'd be happy to say out loud — this is a page about ordinary unfinished business, not a place for the heavy things.
This was his actual picture of a person: not a container of traits but a field, with the unfinished things pulling from wherever they sit. The dot in the middle is you. Every loop you wrote pulls on it with the weight you gave it. Plan one and the pull goes thin and dashed. Stamp one closed and it lets go.
In 2011 E. J. Masicampo and Roy Baumeister ran six experiments on exactly this. Give people an unfinished task and their minds keep going back to it — in their first study, more intrusive thoughts and measurably worse reading comprehension on an unrelated passage afterwards.
Then they added one step. Before the reading, some participants wrote a specific plan for the unfinished task: not "I'll get to it", but when, and where, and how. They did not do the task. They only planned it.
The intrusions dropped to the level of people who had actually finished. The comprehension came back. Something about naming a time and a place appears to let a task stop shouting — the pull is discharged not by completion but by a credible appointment.
With the caveat from the wobbles above: their own last two studies found the anxiety stayed. A plan quiets the loop. It does not necessarily quiet you.
Skip this one freely. And if the thing that came to mind is heavier than a leaning shelf — a person, a loss, something you have been going around for years — that is not a task, and it is not yours to solve alone before Thursday. Saying it out loud to someone you trust is how most people get anywhere with those.
Everything below is yours, played back. It is not a score, not a type, and not a verdict — the page has no idea what kind of person you are and is not going to guess. Read it, and if the pattern is wrong, say so; the last word is yours.
The graded part
This is the only thing on this page with right answers. It is about the people, the dates and the findings — never about you. Your shelf is not marked. Berlin is.
How it travelled
She left Berlin for Moscow in 1931 and never worked on the effect again. She built Soviet pathopsychology instead — thinking disorders, brain-injured soldiers in a wartime hospital in the Urals with Alexander Luria, seven hundred and ten patients in her second doctorate. In 1940 her husband was arrested and sentenced to ten years as a supposed foreign agent; he died in the camp. She raised two children and kept working. In 1983 the American psychological establishment gave her the Kurt Lewin Memorial Award, named for her old supervisor at the small crowded table. She was not permitted to travel to collect it.
The thing with her name on it turned out to be the least of what she did, and half of it did not survive contact with the next hundred years. That is not a sad ending. That is what a real finding looks like from the inside: something you were sure of, checked by strangers, and reduced to the part that holds.
So: of everything still open on your shelf — how much of it are you actually going to go back to, and how much of it has been finished for a while, waiting for you to stamp it?