Doorway effect

Why does walking into the next room so often erase the reason you went there?

The ready explanation is distraction: your mind wandered on the way. Hold it against what happened. The walk took four seconds down a corridor you have used ten thousand times, nothing occurred, nobody spoke to you. And look at the shape of the failure — you have not lost everything. You are standing in the kitchen entirely certain that you came here for something. Distraction should not leave a neat hole with its edges intact. Something specific was taken and something specific left, and that pattern is the clue.

Start with an assumption worth doubting: that experience is laid down continuously, like tape. It is not. When people watch films of ordinary activity — someone making a bed, someone assembling a piece of furniture — and mark where one thing ends and another begins, they agree with each other to a striking degree, and the points they pick are where what happens next stops following from what came before.

Why would the mind carve experience up at all? Because what it is really maintaining is not a record but a working model of the current situation: who is here, what is where, what I am in the middle of doing. That model earns its keep by predicting the next moment cheaply. When prediction starts failing, the model has stopped fitting the world, and the efficient move is not to patch it but to close it and open a new one. That closing is an event boundary.

Now put your intention into that picture. "Go and get the scissors" is, in its first seconds, not filed anywhere durable. It is a component of the model of the present situation — one of the things being actively held. And a doorway is about as strong a boundary as domestic life provides: the scene changes wholesale, every visible thing is replaced at once, and the model has to be rebuilt from the new room.

So here is the reframe. Nothing was forgotten. The intention changed address. It went from being part of what is currently held to being something retrievable from the previous event — and retrieval across a boundary is slower, less certain, and sometimes fails outright. What you experience as a blank is a lookup that has not returned.

That also explains the intact edges. The fact of having an open goal is held differently from the content of the goal; an unfinished intention stays generally accessible, which is why you know perfectly well that you came for something. The wrapper survived and the contents are on the other side of the boundary.

Here is the finding that decides between explanations, and it is not the one intuition predicts. If this were ordinary context-dependent memory — the room's cues supporting recall — then walking back should restore what was lost. In follow-up work it did not: returning to the original room failed to remove the decrement, which is the main reason the effect is read as the model having been updated rather than the cues left behind. Crossing back is a second boundary, not an undo. Everyday success at "going back" may owe to bumping into the specific object you had been looking at, a far more particular cue than the room itself.

One honest qualification, and it is substantial. Attempts published in 2021 to reproduce the doorway result found it inconsistently, and mainly when memory was already loaded or the surroundings were sparse. The event-segmentation machinery underneath is well supported and not in doubt; whether a doorway alone reliably produces measurable forgetting is contested, and the current reading is that it surfaces when the intention was fragile to begin with — which fits the everyday case, where the thought was formed in passing and never rehearsed once.

And it is not really the door. Virtual doorways in a simulated house produced the effect too, and a change of topic, of task, or of scene can serve the same function. The door is a convenient physical marker of a boundary, not the cause of one.

Three corrections tie the reasoning together. The first is the distraction account, which fails not because attention is irrelevant but because it predicts the wrong thing: it says a busier corridor should be worse, when what matters is whether a boundary was crossed. The second is the doorway itself, a marker rather than a mechanism — treat it as the cause and you will be puzzled that a change of subject does the same job.

The third is the most useful. This is not a malfunction. A system that carried every detail of the last situation into the next would be worse, not better; the reason you can stand in the kitchen and think clearly about the kitchen is that the study has been closed off. The doorway effect is the visible edge of a mechanism doing what it should, priced occasionally at one forgotten errand.

• Why an interrupted task keeps returning to mind while a finished one does not, and how that general accessibility sits alongside the boundary that hides the content.

Every term the collection defines is gathered in the glossary.