Does It Hold Up?

Do open loops work because people remember unfinished things?

The advice

Start a video with a question you don't answer yet, and viewers stay to find out. The reason that comes with the advice is that the brain remembers unfinished things better than finished ones. That claim has a name, the Zeigarnik effect, and it traces back to a study Zeigarnik published in 1927.

What the original studies did

Zeigarnik gave participants a series of tasks, interrupted some of them partway through, and afterward asked them to list the tasks they could remember. She reported that they recalled more of the interrupted tasks than the finished ones.

A year later, Ovsiankina used a similar setup and looked at something else: what people did after the interruption. She reported that they tended to go back and finish the interrupted tasks without being asked. That's a separate claim about behavior, and it's called the Ovsiankina effect.

What happened when the studies were combined

In 2025, Romain Ghibellini and Beat Meier gathered every study they could find that tested either effect and combined the results, weighting each study by how many people were in it. A study that pools other studies this way is called a meta-analysis, and it's harder to fool than a single study because one fluke result gets diluted by all the others. They ended up with 59 publications: 38 on the memory claim, 20 on resuming, and 1 on both.

The memory claim didn't hold up. On the measure most of those studies used, interrupted tasks were recalled at a ratio of 0.99 to finished tasks across 38 publications, where 1.00 means no difference. Interrupted tasks made up 49.16 percent of what people remembered, which is about what you'd get if interruption made no difference at all. The authors also showed that the ratio Zeigarnik used in 1927 is thrown off by a few extreme scores, which inflated her original result.

The resuming claim did hold up. Across 21 publications, people went back to an interrupted task 67 percent of the time. The authors say to read that number as a rough guide rather than an exact rate, because the studies measured resuming in different ways.

They also offer a reason the memory effect showed up in older studies: an experimenter with authority in the room, pressure to perform well, and how invested people were in the task. The authors write that those conditions were more common in the past and are rarer now, which could explain why the effect has been hard to repeat.

Where this stops applying

The studies in that meta-analysis gave people tasks to work on and then interrupted them. They didn't test videos, captions, or feeds. So the research supports a narrow statement: when people are interrupted partway through a task, they tend to go back and finish it. It doesn't show that an open question at the start of a video raises watch time, and it doesn't rule that out either.

What to do with it

If you explain your open loops to your audience, swap the reason. Say people tend to go back and finish something that was interrupted, and name the 2025 meta-analysis, instead of saying the brain remembers unfinished things. The first version has 21 publications supporting it. The second one has a ratio of 0.99.

Next step

This kind of check is what The Lab does once a month: one piece of content advice, the research it came from, and whether that research held up when other people tried to repeat it. You can see what's in it at datapsych.net/the-lab.

Sources