Humor Effect: Why the Funny Line Is the One You Remember

Funny material gets more of the two things that build a memory: attention while it is in front of you, and rehearsal after it has gone. A joke makes you stop, resolve the incongruity, and often repeat the line to yourself or to someone else. A plain sentence of the same length asks for none of that. Schmidt's 1994 paper tested attention, arousal, rehearsal, retrieval and surprise as explanations for the advantage of humorous sentences and concluded that increased attention and rehearsal fit the results best.

The advantage depends on contrast. In the sentence experiments it appeared when humorous and plain sentences shared a list and disappeared when separate groups studied lists of only one kind. The same pattern held for cartoons in 2002: original humorous cartoons were recalled better than literal and weird versions in within-subject comparisons, and the difference was not found between groups. Schmidt read this as differential processing: when a funny item sits among plain ones, it draws processing away from them, so the plain items in a mixed list are recalled worse than they would be in a list of their own. The joke is not free. It is paid for by whatever sits next to it.

Humor is not just strangeness. In the 2001 cartoon study Schmidt and Williams compared three versions of each drawing: the original joke, a literal translation with the incongruity removed, and a weird version in which incongruous material was inserted without being funny. The weird and literal versions were remembered equally well; only the funny original had the advantage. Detecting an incongruity was not enough. Something about the material being funny did the work, and Carlson found in 2011 that participants' own ratings of how funny an item was predicted recall even after semantic elaboration and incongruity resolution were controlled.

Two limits are worth holding onto. Humor is subjective, so a line that lands for one reader is a plain sentence for another, and the effect follows the rating rather than the writer's intention. And the effect is about the funny item itself. Material presented alongside a joke, such as the product in a funny advertisement or the fact a teacher attaches to a quip, does not reliably benefit; in the mixed-list experiments it did the opposite.

Stephen Schmidt published "Effects of humor on sentence memory" in the Journal of Experimental Psychology: Learning, Memory, and Cognition in 1994. His materials were humorous sayings and a plain rewrite of each one that kept the meaning and the length. One pair he reports: "The only way to keep your good health is to eat what you don't want, drink what you don't like, and do what you'd rather not," against "The only way to keep your good health is to eat good food, drink healthy drinks, and do healthy activities." Participants read lists in which half the sentences were humorous and half were plain, with each base sentence appearing in only one version for a given person, and were then tested by free recall and by cued recall.

Humorous sentences were remembered better on both tests, whether scoring counted whole sentences or the words within them. The advantage survived a warning that a humorous sentence was about to appear, so surprise alone did not explain it. It was smaller under incidental learning, when participants did not expect a memory test, and it was limited to within-subjects designs: when different groups read an all-humorous or an all-plain list, the groups did not differ. In the incidental conditions Schmidt also scored recall against each participant's own humor ratings, and rated humor predicted memory in both the mixed and the separate designs.

Schmidt and Williams extended the result to cartoons in 2001, with the three-version design described above, and Schmidt's 2002 paper in Memory added heart rate as a measure: the humorous cartoons produced a larger slowing of the heart, a marker of attention, than the literal and weird versions, and good recall of the humorous cartoons came at the expense of the non-humorous ones in the same lists. Later work has asked what in the joke does the work. Summerfelt, Lippman and Hyman found in 2010 that knock-knock jokes were recalled and recognized better with their puns intact than with the pun removed, that rehearsal did not explain it, and that the constraints a pun places on the final line may make the line easier to reconstruct.

The rows below are illustrations of where the pattern from the list experiments is likely to show up, not findings from the studies cited on this page.

The effect is useful when the funny thing is the thing you need to remember and a cost when it is not, so the counter is to decide which case you are in.

• When you want something remembered, make it the joke. A humorous line that carries the point is recalled; a humorous line beside the point competes with it.
• When you are the listener, write the plain point down. The experiments found plain material in a mixed list recalled worse than it would be alone, and notes do not have that problem.
• Separate humor from the material it is meant to serve in time. A joke at the end of a section competes less with the content than a joke in the middle of it.
• Check what you actually remember from a funny presentation, advertisement or chapter before you judge it effective. The feeling that it was memorable is often the memory of the joke.
• Be cautious about the claim that humor improves learning in general. The evidence is for the funny item itself, in contrast with plainer items around it; it is not evidence that everything near a joke sticks.

Several entries describe one item standing out from a set. They differ in what does the standing out and in what the cost is.

• It is not the bizarreness effect, the better recall of strange or impossible material. The two are easy to confuse because jokes are often strange, but in the 2001 cartoon study weird versions with the incongruity left in and the humor taken out were recalled no better than literal ones. Humor added something that oddness alone did not.
• It is not the Von Restorff effect, where an item is isolated by its form: a word in red among black ones, a number among words. Isolation needs only a visible difference; the humor effect runs on meaning, and the isolated item can be perfectly dull.
• It is not the picture superiority effect, the better memory for images than for their names. That is a difference of format, and it does not depend on a mixed list the way the humor advantage does.