What is the Einstellung effect?

The Einstellung effect is the tendency to solve a problem using a familiar method even when a simpler or better solution is available. First demonstrated by Abraham Luchins in 1942, it happens because a familiar-looking problem triggers a "mental set" that directs your attention toward the known approach and away from alternatives—often without you realizing it.

What's the difference between the Einstellung effect and functional fixedness?

Functional fixedness is getting stuck on an object's usual purpose—failing to see that a box of tacks could be a shelf. The Einstellung effect is getting stuck on a familiar problem-solving method. Both are forms of mental set, but Einstellung is about the approach you reach for first, which makes it the bigger risk in knowledge work and decision-making.

Why are experts more vulnerable to the Einstellung effect?

Expertise is a deep library of familiar patterns, and any of those patterns can capture your attention before you consciously choose. In a 2008 chess study, the presence of a familiar-but-worse move dropped expert performance to the level of players three standard deviations weaker. Nobody is immune, and the more patterns you know, the more often the trap is baited.

Does the Einstellung effect apply to AI?

Yes—arguably more than to anything else. Generative AI returns the most statistically likely answer first, which is exactly the familiar stimulus that triggers fixation. A 2024 CHI study found AI examples increased fixation and reduced idea variety, and Microsoft and Carnegie Mellon researchers found that the more workers trust AI, the less they explore alternatives.

How do you overcome the Einstellung effect?

Force a search the first answer skips. Generate multiple options before evaluating any of them, assign someone to argue for the alternative the AI crowded out, and add a required "what's a different approach?" step to AI-assisted decisions. Luchins fixed half his subjects with a two-word reminder—"Don't be blind"—so even a small, deliberate nudge to keep looking goes a long way.

The Einstellung Effect: Why AI Traps Your Best People

In a lab at Oxford, chess masters sat in front of a board with a checkmate hiding in plain sight. There were two ways to win. One was a famous pattern they had drilled for years. The other was faster, cleaner, and just one move shorter.

The masters found the famous one. Most never saw the better one. And here is the unsettling part: eye-tracking showed their gaze drifting back, again and again, to the squares of the move they already knew—while they swore out loud that they were still searching for something better.

That trap has a name. It is the Einstellung effect, first measured in 1942, and it is the most important psychology your team has never heard of. Because the very tool you rolled out to make people smarter—generative AI—is the most efficient Einstellung machine ever built. This piece explains what the effect is, why your strongest people are the most exposed, how AI quietly deepens it, and what to actually do about it.

What the Einstellung effect actually is

The Einstellung effect is your brain's tendency to solve a problem the familiar way, even when a simpler or better way is sitting right there. Einstellung is German for "setting" or "attitude"—a mental setting that locks into place the moment a problem looks familiar.

The classic proof comes from psychologist Abraham Luchins and his water-jar experiments. Subjects had to measure out an exact amount of water using three jars of fixed sizes. The training problems all worked with the same three-step formula: fill the big jar, pour off the others. People learned it cold. Fill jar B (127 units), pour off A (21), then pour off C (3) twice, and you land on 100.

Then Luchins slipped in a trap. A new problem could still be solved the long, learned way—but it also had an obvious two-step answer staring the subject in the face. Most people who had learned the formula marched right past the easy solution and used the long one anyway. A fresh control group, with no formula burned in, spotted the simple answer instantly. The "set" group had mechanized their own minds.

The fix Luchins found is almost funny. When he scrawled two words on the test sheet—"Don't be blind"—over half of the stuck subjects suddenly saw the shorter path. The knowledge was always there. Attention was the bottleneck. This is why the Einstellung effect is best understood as a mental set, or problem-solving fixation: not a lack of ability, but a spotlight pointed at the wrong spot.

Why experts get stuck on familiar solutions

Here is the counterintuitive part, the part that should worry every team lead. The Einstellung effect does not mostly hit beginners. It hits people who are good at their jobs.

Go back to those chess masters. In a landmark 2008 study, researchers Bilalić, McLeod, and Gobet put strong players in front of a position that contained a familiar winning pattern (a well-known mate) plus a shorter, better solution. When the familiar pattern was present, players found it and stopped—missing the faster win. When the researchers removed the familiar pattern from an otherwise identical board, the same players found the optimal solution easily.

The size of the damage is the headline. The mere presence of a familiar-but-worse option dropped expert performance to the level of players roughly three standard deviations weaker—a master suddenly playing like a decent club amateur. And the eye-tracking told the real story: their eyes stayed locked on the region of the board tied to the move they knew, even as they reported hunting for alternatives. They were not lazy. They were blind to their own blindness.

Why are experts so exposed? Because expertise is a library of familiar patterns. The more patterns you have, the more triggers there are to capture your attention before you have consciously chosen anything. The skill that makes you fast is the same skill that snaps you shut. (The very best players in the study could still break free—so expertise is not a curse, exactly. But nobody is immune, and the better you are, the more often the trap is baited for you.)

This is different from a related trap called functional fixedness—Duncker's famous candle problem, where people can't see that a box of tacks could become a candle holder. Functional fixedness is getting stuck on what an object is for. The Einstellung effect is getting stuck on the method you reach for first. In knowledge work, the second one is far more expensive, because the "object" you're fixated on is an idea.

AI is an Einstellung machine

Now add the accelerant.

A generative AI tool is, by design, a familiarity engine. Ask it anything and it returns the most statistically probable response—the first plausible solution, delivered with confidence, formatted to look finished. That is precisely the stimulus the Einstellung effect needs. The AI doesn't just hand you an answer; it hands you a familiar-feeling answer and points your spotlight straight at it.

We don't have to theorize about this. A 2024 study presented at CHI, the top human-computer interaction conference, tested it directly. Researchers had people brainstorm with and without AI-generated examples. The finding, in their words: "support from an AI image generator during ideation leads to higher fixation on an initial example." The AI-assisted group "produced fewer ideas, with less variety and lower originality" than people working without it. That is the Einstellung effect, reproduced on demand.

Zoom out from one person to a whole population and it gets worse. In a 2024 study in Science Advances, 293 writers produced stories with and without AI help. Individually, AI made stories more novel—up to 8.1% more. But collectively, the stories became measurably more similar to each other. AI raised the floor for less-creative writers (their quality jumped 26.6%) while barely helping the strongest ones. Everyone converged toward the same middle. A team can feel more creative person by person while becoming more uniform as a group—a monoculture you can't see from your own desk.

Researchers at Microsoft and Carnegie Mellon found the human mechanism behind it: in a 2025 survey of 319 knowledge workers across 936 real tasks, the more people trusted the AI, the less critical thinking they did—and they produced "a less diverse set of outcomes" on the same task. The trust is the off-switch for the search. This is the same family of risk as plain old AI over-reliance and automation bias, but it's sneakier: you're not failing to check the answer, you're failing to look for a different one.

The slowdown you won't feel

The reason this matters now—not in some abstract future—is that the cost is invisible to the people paying it.

In mid-2025, the research group METR ran a clean experiment: 16 experienced open-source developers, 246 real tasks, randomly assigned to use AI tools or not. The developers expected AI to speed them up by 24%. Instead, the tasks took 19% longer with AI. And the kicker: even after living through the slowdown, the developers still believed AI had sped them up by 20%. They were slower and convinced they were faster.

That gap between felt productivity and real productivity is the Einstellung effect's signature. Accepting the first good-enough answer feels like speed. Searching for the better one feels like friction. So we optimize for the feeling and lose the outcome.

The pattern shows up in the work product, too. An analysis by GitClear of millions of lines of code found that copy-pasted, duplicated code climbed sharply as AI tools spread—while the rate of refactoring (the act of finding a cleaner solution) fell by more than half. And a 2023 Harvard–BCG study of 758 consultants found that AI flattened the gap between strong and weak performers, but on tasks just outside what AI does well, consultants using it were 19% less likely to reach the right answer. The tool that makes the easy stuff effortless makes the hard stuff invisible.

Meanwhile, adoption is near-total: Gallup found that by late 2025, 51% of U.S. employees were using AI at work, with 26% using it weekly. The Einstellung effect used to be a quirky lab result. It is now the default operating condition of knowledge work.

How to overcome the Einstellung effect

The point is not to rip out the AI. Used well, it is genuinely powerful—and you can read our take on routing the right question to the right model. The point is to rebuild the search that the first answer skips. A few plays that work:

Write "Don't be blind" into your process. Luchins fixed half his subjects with a two-word nudge. Make "What's a second, completely different approach?" a required field on any AI-assisted decision. The prompt to find an alternative has to be louder than the relief of having an answer.
Diverge before you converge. Generate three options before you evaluate any of them. The Einstellung effect strikes at the moment of the first plausible solution, so the only defense is to refuse to evaluate until you have more than one.
Red-team the AI's first answer. Assign someone to argue against it—not to check whether it's correct, but to find the answer it crowded out. This is especially important because your most senior people are the most fixated, and they're the ones who usually go unchallenged.
Keep the thinking visible and shared. Most AI tools are built to collapse a messy conversation into a single summary, a single action list, a single suggested reply—"conversation to completion." That's convergence as a feature, and it's exactly what feeds fixation. The antidote is to keep the alternatives, the disagreements, and the discarded options where the whole team can see them.

That last point is why we built Coommit the way we did. When the AI works on a shared canvas alongside the conversation—instead of quietly resolving everything into one tidy output—the discarded paths stay on the board. The team can see the move it didn't make. That doesn't immunize anyone against the Einstellung effect, but it does what Luchins's note did: it points the spotlight back at the space the first answer made you stop searching.

Conclusion

The Einstellung effect has been quietly costing organizations good decisions for 80 years. What's new is the trigger. Generative AI is the most fluent producer of familiar-feeling first answers ever invented, and it has landed in nearly every workflow at once—pointed, by default, at the people whose expertise makes them the most susceptible.

The teams that win the next few years won't be the ones that adopt AI fastest. Everyone is doing that. They'll be the ones that pair it with a deliberate search for the better answer—who build the second-solution habit into how they work. The first good answer is almost never the best one. The trick is remembering to look. Start by asking your team one question this week: what's the move we're not seeing?