What Actually Slices a Memory in Two

A small team of researchers in the Netherlands and Sweden ran an experiment last year that I keep coming back to.

They put people in a virtual reality headset and asked them to play a salesperson. The setup looked like a small market with several booths. Customers walked up. Each customer asked for something specific. The participant had to remember the order of customers and what each one wanted.

The clever part was the two different kinds of boundary the researchers built into the task.

A spatial boundary was when the booth changed — the participant moved from one stall to a different stall, with different visuals, different layouts. Same job, different room. Like walking from your kitchen into your living room.

A conceptual boundary was when the customer's request changed — switching from someone asking about flowers to someone asking about books. Same booth, different goal. Like sitting in the same chair but suddenly switching from writing an email to scheduling a doctor's appointment.

The two kinds of boundary could be tested independently or together. A booth-change with no goal-change. A goal-change with no booth-change. Both at once. Neither at all. Then they tested what people remembered about the customers and the order they had appeared.

Only one of the two boundaries mattered

The result was clean and a little startling.

Spatial boundaries — the booth-change, the room-change, the literal version of walking through a doorway — barely mattered for sequence memory. People remembered roughly the same whether they had moved booths or not. The wall, by itself, did almost nothing.

Conceptual boundaries — the goal-change, the what am I doing now? shift — mattered a lot. The moment a participant's task changed, their memory of the order of recent customers got worse. Their confidence about the sequence dropped. The brain treated the goal-change as the start of a new event, regardless of whether the booth had changed at all.

Confidence in correct memories was highest when participants stayed in the same booth and continued the same kind of task — a small island of stable thinking. The moment either the goal or the location shifted, confidence dropped. But only the goal shift actually moved the accuracy. The location shift moved how it felt, not how it was.

The boundary that fragments a memory isn't where you are. It's what you're trying to do.

That is a quiet finding with loud implications.

It means the doorway in your house is not really the thing rearranging your kitchen-bound thoughts. The thing rearranging them is the change of intent — going from "I am sitting at my desk thinking about that email" to "I am getting up to do something with my hands". The room change is along for the ride. The intent change is the boundary.

It also means something specific about what your brain does when you change tasks. It doesn't just shift gears. It quietly closes a small file. Whatever was loose in working memory at the moment of the shift — the email half-formed in your head, the worry you hadn't quite put down — gets harder to fully retrieve. You can still remember it. But the link to the current moment is weaker.

You can use that. On purpose. Most of us don't.

That is what the next chapter is for.

References

• Li, Y., Johansson, M., & Nikolaev, A. R. (2025). Hierarchical event segmentation of episodic memory in virtual reality. npj Science of Learning, 10, 25.
• Bailey, H. R., Kurby, C. A., Sargent, J. Q., & Zacks, J. M. (2024). Toward an integrative account of internal and external determinants of event segmentation. Psychonomic Bulletin & Review, 31(2), 484–506.