The 1927 Study Behind the Effect
Bluma Zeigarnik was a graduate student in Kurt Lewin's seminar at the University of Berlin when Lewin described something he'd noticed at a café: a waiter could recite a long, complicated unpaid order from memory, then drew a blank on the same order minutes after the bill was settled. Zeigarnik turned the anecdote into an experiment. She gave 32 adults 22 short tasks - thread-winding, arithmetic, puzzles, clay modelling - each meant to run three to five minutes. Roughly half of each participant's tasks were allowed to finish; the rest were interrupted at the point of deepest engagement. An hour later, asked to recall what they'd been doing, participants remembered the interrupted tasks about 90% more often than the completed ones. She published the results in 1927 in Psychologische Forschung.
"People remember uncompleted or interrupted tasks better than completed tasks - though how reliably that holds up is still debated." - commonly cited summary of Zeigarnik's 1927 finding
Why the Effect Is Shakier Than It Sounds
This is one of psychology's most cited findings - and one of its least reliable. Replication attempts have gone both ways since the 1930s. Rosenzweig (1943) got the opposite pattern: when he told participants the tasks were an intelligence test, they remembered the completed tasks better, as if protecting their self-image from the discomfort of unfinished work. Van Bergen's 1968 replication found no significant difference between finished and unfinished tasks at all. A 2025 meta-analysis in Humanities and Social Sciences Communications reviewed the accumulated evidence and concluded that unfinished tasks carry no consistent memory advantage - what does hold up across studies is a related but separate finding, the Ovsiankina effect: people's tendency to want to go back and finish an interrupted task, independent of whether they remember it any better.
What Product Designers Are Actually Using
Most "Zeigarnik-style" UI patterns lean on that resumption tendency rather than the contested memory claim. A profile sitting at 70% complete pulls at users differently than one at 0% - not necessarily because they'll recall it better, but because an open loop nags at them until it closes.
- Progress indicators: A completion bar at 70% creates tension that 0% doesn't. Pre-fill it with data you already have (name, email from sign-up) so users start invested rather than blank.
- Streaks: Duolingo's streak counter, GitHub's contribution graph - the visible, unbroken chain is the open loop.
- Step counters: "Step 2 of 5" signals an incomplete sequence more effectively than "Step 2" alone.
- Cart-abandonment emails: "You left something behind" works on the resumption urge, not on the assumption the user forgot what they left.
The Ethical Boundary
Manufactured incompleteness that serves the product at the user's expense - infinite notification badges, fake progress that never resolves - is a dark pattern, not a design technique. Use open loops to help users finish something they already wanted to do, not to keep them circling a task that only benefits engagement metrics.
Key Takeaways
- Zeigarnik's original 1927 finding - better recall for interrupted tasks - has failed to replicate consistently since the 1940s.
- A 2025 meta-analysis found no reliable memory advantage for unfinished tasks, but confirmed people do want to resume them (the Ovsiankina effect).
- Progress bars, streaks, and step counters work on that resumption urge, not on a proven memory boost.
- Use open loops for tasks users already want to finish - not as a hook to inflate engagement.
