Unfinished Tasks Are Not Reliably Remembered Better

The famous result disappears

Interrupt someone halfway through a task, and the unfinished task should become especially memorable. That is the popular Zeigarnik effect. Across decades of experiments, however, the predicted recall advantage nearly vanishes.

A 2025 synthesis examined 59 publications using interrupted-task methods. Across 38 recall publications, the sample-size-weighted ratio of mean recall for interrupted versus completed tasks was 0.99. In other words, recall was effectively equal.

The result was similar under calculations less vulnerable to extreme ratios. In 14 publications, interrupted tasks made up 49.43% of everything recalled. Removing Zeigarnik’s original study lowered that figure slightly, to 49.16%.

This matters because Zeigarnik’s principal statistic averaged each participant’s ratio of interrupted to completed tasks recalled. When someone recalled few or no completed tasks, that ratio could become extreme. The measure was structurally favorable to interrupted tasks, according to the 2025 analysis.

Interruption is not one clean variable

The synthesis does not prove interruption can never improve recall. Its included experiments varied in tasks, instructions, populations and interruption procedures. Some showed an advantage; others showed the reverse.

The analysis also used sample-size-weighted descriptive averages rather than a conventional random-effects model with pooled uncertainty. Many older reports lacked enough information for standardized effects, and the search was limited by databases, languages and available publications. Its conclusion applies to the literature it included, not every possible interruption.

A further problem is processing time. If an interrupted task receives less attention than a completed one, later recall compares more than completion status. The neat slogan—unfinished means memorable—compresses several different experimental conditions into one supposed law.

An intention stays easier to reach

A narrower effect has held up better. In four experiments, Goschke and Kuhl asked participants to memorize two action scripts while expecting to perform one later. During intervening recognition tests, participants responded faster to words from the intended script.

That is not free recall, and it is not evidence of conscious rumination. It is a response-time difference for material connected to an expected action.

Meilán later tested 30 Spanish psychology students with a similar procedure. Before performance, words from the intended script were recognized in 890 milliseconds, compared with 964 milliseconds for the neutral script. The advantage disappeared after the action was performed. No equivalent advantage appeared when participants expected to watch the experimenter act instead.

These studies support temporary accessibility under controlled conditions. They do not show that every unfinished task continuously occupies working memory.

A plan can quiet some intrusions

Masicampo and Baumeister studied unfinished personal goals rather than interrupted laboratory activities. In one experiment, undergraduates who wrote about unfinished tasks later reported more task-related distraction while reading than those who wrote about completed tasks: 3.00 versus 1.82 on a seven-point scale.

Participants who also made specific completion plans reported distraction of 1.77. Their reading-comprehension scores were likewise similar to the completed-task group.

The limits are important. The six experiments came from one paper, not a large independent replication literature. Planning reduced reported intrusions but did not reduce anxiety or restore contentment. The experiments also did not reproduce Zeigarnik’s interruption-and-recall procedure.

What actually survives

Here the evidence ends and the interpretation begins. “Open loops stay in your head” bundles together recall, response speed, distraction and behavior. Those outcomes are not interchangeable.

The most consistent interrupted-task result may be behavioral. Across 21 publications in the 2025 synthesis, participants resumed interrupted work in roughly 67% of observations. The procedures were too heterogeneous to treat that figure as a universal probability, but the direction recurred.

An unfinished task may not win a memory contest. Given another chance, though, people often return to it. Why recall fails while resumption persists remains the more interesting unfinished question.