Functional Fixedness: Why You Cannot See the Second Use

An object's design function is part of how it is stored. Asked what a box is, people answer with what it is for, not with its material, rigidity or dimensions, and the problem-solving process draws on the same representation. When a solution requires the dimensions and not the purpose, the purpose has to be got out of the way first, and nothing in the task tells you that it is there.

Recent experience makes the function heavier. Duncker's preutilization condition worked by having the box perform its ordinary job immediately before the problem, and the general finding is that priming the usual function lowers the solution rate relative to conditions where it is not primed. Sam Glucksberg and Robert Weisberg added a result in 1966 that narrowed the mechanism: people who failed to solve the problem were afterwards less likely to produce the word box in a free-association task, which they read as the concept not being available when it was needed.

Michael Frank and Michael Ramscar tested that reading in 2003 and found it too simple. In their written version of the problem, every participant saw the word box printed in front of them, so the concept was plainly available, and the solution rate still more than doubled when the object phrases were underlined. What changes, on their account, is not whether the word is accessible but how fully the object is represented: a partial representation carries the function, a fuller one carries the material and the shape as well.

This is why the counters work where they work. Tony McCaffrey's generic parts technique asks solvers to break each object into parts and describe each part without naming a function, so that a box becomes rigid cardboard rectangles. Subjects trained in that technique solved on average 67 percent more problems than a control group.

Two cautions. Failing one insight problem is not a measure of a person's flexibility, and functional fixedness is a property of a situation rather than a trait: the same person who cannot see the box as a shelf sees it at once in a room where the tacks are already out.

Karl Duncker's monograph "On problem-solving" appeared in Psychological Monographs in 1945, translated from work he had completed in Germany in the 1930s, and it introduced both the term functional fixedness and the box problem the demo above uses. Duncker's method was to give people practical problems, ask them to think aloud, and record the paths they took to a solution or to a dead end. The box problem was one of a family: others required a pair of pliers to serve as a weight, or a paper clip and a gimlet to do something other than their own jobs.

His manipulation was preutilization. In one arrangement the materials needed for the solution had just been used for their ordinary purpose, so the tacks were in the box and the box was a container. In the other the same objects were present with the solution object free of that role, the tacks lying loose and the box empty. People in the first arrangement solved the problem less often and more slowly. Duncker's explanation was that the object was "fixed" in its established function, and the solver had to perform a restructuring of the situation to see it any other way.

Robert Adamson repeated three of Duncker's experiments in 1952 in the Journal of Experimental Psychology, reporting that 57 participants were divided between a preutilization group and a control group and given the box, gimlet and paper clip problems in that order. The results confirmed Duncker's direction with a larger sample and more carefully specified conditions, and the 1952 paper is the citation usually given for the replication rather than the original effect.

The written version used on this page comes from Michael Frank and Michael Ramscar's 2003 report, which put the problem to 68 Stanford undergraduates in three conditions as a printed description and asked them to write any solution they could find. In the plain-text control condition the solution rate was 23 percent. Underlining the phrases candle, book of matches and box of tacks raised it to 55 percent, and underlining every individual noun raised it to 47 percent.

The pattern appears wherever a job needs an object to do something other than its job. The rows below are illustrations, not findings from the studies cited on this page.

Most of the counters that have been tested work by describing the thing without naming what it is for.

• Break the object into parts and describe each part without a function word. A candle becomes wax and a wick; a box becomes flat rigid panels of cardboard. This is McCaffrey's generic parts technique, and in his study it raised the number of problems solved by 67 percent on average.
• State the requirement physically. "Something that will hold 200 grams, 10 centimeters off the wall" admits answers that "a shelf" rules out.
• Empty the container. Duncker's result says the object in use is the hardest to repurpose, so clearing it, putting it down or drawing it empty can change what you see.
• Change who is looking. Someone who has never met the object in its usual role has nothing to set aside, which is one reason an outsider's first suggestion is sometimes the one that works.
• Count the objects and the surfaces in front of you, out loud. Frank and Ramscar's underlining shows how small a nudge to attention can be and still change the outcome.
• Leave the problem and come back. Restructuring is more likely once the first framing has faded, and nothing is lost by returning to a puzzle you were stuck on.

Functional fixedness is specifically about objects and their uses, which separates it from several neighboring kinds of stuckness.

• It is not the Einstellung effect that Abraham Luchins described in 1942, in which a method that worked on earlier problems is applied to a later one that has a simpler solution. That is fixation on a procedure rather than on what a thing is for.
• It is not the curse of knowledge, the difficulty of imagining what it is like not to know something you know. The curse of knowledge is about other minds; functional fixedness operates entirely within your own view of an object.
• It is not selective perception, which filters what gets noticed according to expectations and interests. In the candle problem the box is noticed by everyone; what fails is the use it is put to.
• It is not congruence bias, the habit of testing only the hypothesis you already hold. That one is about how experiments are chosen, not about how objects are represented.