George W. Bush was not known for his cunning intellect, but he did have a good sense of humor.  In a commencement address at Southern Methodist University, he famously told the graduates, “For those of you graduating with high honors and distinctions, I say well done.  And as I like to tell the “C” students, you too can be president.”  Like Bush, many of us use humor to diffuse difficult situations, mask nervousness, soften criticism, and cope with failure.  Humor also serves the role of locksmith in both platonic and romantic social interactions, as it helps us break the ice, gain social acceptance, and initiate romantic overtures.  Both men and women tend to seek mates who have a good sense of humor, and we perceive funny people as smarter, more attractive, and more personable.

Given that humor is such a powerful tool for social success, it’s not surprising that scientists have sought to determine the perfect formula for funny.  Although there are many competing theories (and no definitive answers) about how humor functions, new research by Chris Westbury, Cyrus Shaoul, Gail Moroschan, and Michael Ranscar suggests that at least one key ingredient can be found in a 200 year-old theory proposed by philosopher Arthur Schopenhauer.

In a nutshell, Schopenhauer suggested that humor derives from an incongruous outcome of an event for which there is a very specific expectation.  It is the violation of the specific expectation that creates humor.  Consider this pun: “When the clock is hungry it goes back four seconds.”  The notion of a clock eating is incongruous with our knowledge of the world, but that alone is insufficient to create humor.  The statement, “When the clock is hungry it eats a cheeseburger,” is also incongruous, but “eating a cheeseburger” does not violate any specific expectation about a clock and so the statement is far less amusing.  It is the ending, “goes back four seconds,” that elicits a humorous response (albeit an extremely mild one), and it does so because of our understanding of the dual meanings of the words “four” and “seconds,” and our expectation about which of those meanings apply to a clock.

In three experiments, Westbury and colleagues tested the idea that greater incongruity between expectations and outcomes produces a stronger feeling of humor.  They did so by examining the humor in non-words, which are strings of letters (e.g., digifin) that form a pronounceable but meaningless unit.  Non-words offer a unique advantage in the analysis of the role of expectation-violation in humor, as they are relatively devoid of meaning and thus allow a more pure assessment of influence of incongruency on funniness. 

In their first study, Westbury et al. assessed whether there is any consistency in the funniness ratings of non-words.  Although specific instances of humor are not always considered universally funny (consider for example popular skits from Saturday Night Live, which tickle many people but also offend others), something is objectively humorous only if there is some consensus about its hilarity. Thus Westbury and colleagues asked nearly 1000 students to rate a total of almost 6000 randomly-generated non-word strings (e.g., artorts) on funniness.  The results indicated that these random non-words had reliably consistent humor ratings.  If one participant found a given non-word funny, it was likely that others found that same item funny as well, and vice versa.

Westbury et al. next tried to understand what made certain non-words funny (and others not so much).  In two additional studies, they directly examined funniness ratings for non-words that varied with respect to entropy, that is, the extent to which the combination of letter strings was incongruous or unexpected.  To understand incongruity in non-words, it is important to know that some letters are more likely than others in the English language (e.g., “E” is more frequent than “Q”), and furthermore that some letter combinations are more likely than others.  Thus the entropy of a non-word is essentially a measure of the summed probabilities of the individual letters in each string.  Non-words with unusual letters and/or combinations have low entropy and offer more surprise.  In line with Schopenhauer’s theory, Westbury et al. predicted that items with low entropy would receive the highest humor ratings, as these items were most likely to violate expectations about letters and words.

In one study, participants saw two non-word strings (e.g., quarban, mestead) that appeared simultaneously on a computer screen, and on each trial had to select the non-word that they perceived to be more humorous.   Each participant made judgments for 50 pairs.  In another study, participants saw non-word strings that appeared one at a time on a computer screen, and had to rate the humor of each item on a scale from “least humorous” to “most humorous.”  Participants each rated the humor of 100 non-words.  The findings from both of these studies supported the hypothesis that non-word strings with low entropy are perceived as more humorous.  Strings with low entropy (e.g., himumma) were reliably chosen as more humorous than paired strings with higher entropy (e.g., tessina), and strings with lower entropy were judged to be funnier than strings with higher entropy. When we expect one thing, even something as simple as letter combinations, and that expectation is violated, we chuckle. 

It is important to note, however, that at this point we cannot pinpoint low entropy as the definitive source of humor. While these studies demonstrate that expectation violation increases perceived humor, only one type of entropy (i.e., the probability associated with letter strings) was studied here, and with more complex stimuli other types of expectation violation may contribute to amusement.  Even for non-words, many other layers of expectation violation are possible (e.g., how many double letters, such as “zz,” are included in a string, how unusual is the string’s phonology). Indeed, although Westbury et al. intentionally used non-word stimuli because non-words are fairly meaningless, they still found that a handful of the non-word items that were rated most humorous were not necessarily those with lowest entropy, bur rather those that were similar to or contained parts of dirty words (e.g., whong, nip, poo).  Of course one could argue that this finding demonstrates a different kind of expectation violation, as taboo words are arguably unexpected in a serious scientific study, and so are likely to be perceived as funny in that context. 

Unfortunately, understanding that outcomes that violate expectations tend to be perceived as funny doesn’t necessarily make it easier to say or write something humorous.  If creating humor involved a simple scientific calculation, more of us nerdy researchers would be out of the classroom and into the late night comedy circuit (or perhaps we too could be president).  Instead, we’ll likely go back to the lab and tweak our non-word generators. Himumma!