The Complicated Search For The Origins Of Humor

Click here to read an original op-ed from the TED speaker who inspired this post and watch the TEDTalk below.

Two posts in this edition of TEDWeekends have presented differing objections about our work; both hinge upon falsifiability. Peter McGraw and Joel Warner have promoted the idea that while their theory of humor has run the rigors of a few tests in McGraw's lab, our theory should be taken with a heftier grain of salt simply because it has not. Christian Hempelmann, for his part, has alleged, more severely, that our model is categorically unfalsifiable. We'll evaluate these two objections, in that order, ultimately assessing both to be methodologically conservative or, worse, unimaginative. Before we get started, we should note: Our intent here is primarily to discuss scientific methods, not to defend the details of our theory. Because of this, we think a one- or two-paragraph long description of the theory, which is what it takes to make the details clear, would take us too far afield. The interested reader can look in our book, Inside Jokes: Using Humor to Reverse-Engineer the Mind or find summaries in the detailed reviews of it online.

According to McGraw and Warner, "the Hurley model of humor is based solely on argumentation, intuition and thought experiments." That is false, and it also expresses an overly narrow view of science. Theoretical physicists don't themselves do experiments, but they are genuine scientists and their arguments and discoveries are constrained by all the experiments that have been done and observations that have been previously reported by others. Similarly, we, as theoretical cognitive scientists, have built our model of humor on all the (good) work, experimental and theoretical, that has been done before us.

Some puns have clear violations of social or ethical norms: "A hole has been found in the nudist camp wall. The police are looking into it." This fits with McGraw and Warren's theory. But here's another: "Two goldfish are in their tank. One says to the other, "you man the guns, I'll drive." There is no clear violation here -- the humor depends simply upon an initial misinterpretation of the word "tank." -- Matthew Hurley

When there is a well-defined, unambiguous hypothesis to test, an experiment can be very simple: find a clear counterexample. In the case of a theory of humor, find something that is funny but is not predicted to be funny by the hypothesis, or something the hypothesis says should be funny that isn't. To be sure, philosophers sometimes construct fantastical examples that, while perhaps logically possible in some imaginable universe, are not grounded in observations of our universe and so may not be relevant to theories about this world. Still, real-world counterexamples are often easy to find and cannot be ignored. Counterexamples are observational data. The fact that they occur in the world and not the laboratory makes them all the more relevant for a theory that purports to account for a phenomenon in the world.

Both our model and the Benign Violation Theory (the "BVT" which, as the name implies, says that humor occurs when we 1) feel there is a violation of some kind but 2) see that the violation is not harmful), are expressed in terms that need close attention to definitions. What might seem to be a "slam dunk" counterexample may instead just trade on a misconstrual of the terms involved. We have acknowledged the need to extend and complicate our view somewhat in order to account for some possible counterexamples, and McGraw should consider taking the same stance with regard to his theory.

If puns, for instance, are "benign violations", what kind of rules does McGraw have in mind? There is no rule against making puns (though we sometimes wish there was!) and, usually, puns are composed with legitimate grammatical structure (not to mention, benign that syntax violations are not funny). Perhaps the violated rules are semantic ones then? But are there rules for semantics? People bend the meanings of words all the time. Even the word "bend", in the previous sentence makes us think of words as physically pliable objects (a conceptual "violation"). If bent meanings like this were the kind of benign violations that counted as humor then everyday language, riddled with analogies of this sort, would be constantly mirthful. It doesn't seem as if the violation of linguistic rules can be what McGraw and the coauthor of his theory, Caleb Warren, intend.

Some puns have clear violations of social or ethical norms: "A hole has been found in the nudist camp wall. The police are looking into it." This fits with McGraw and Warren's theory. But here's another: "Two goldfish are in their tank. One says to the other, "you man the guns, I'll drive." There is no clear violation here--the humor depends simply upon an initial misinterpretation of the word "tank" (and we've just determined that semantic violations aren't the kind of violation that could make the BVT work). If McGraw and Warren want to say that the mere idea of a goldfish driving a military tank or shooting a gun is a violation, then they would be committed to any outlandish fictional story being humorous: Johnny, the land crab, crawled up the drainpipe to meet his girlfriend for dinner. Ha ha ha ha ha! No, not really.

These counterexamples aside, we could give McGraw and Warren (and Warner) the benefit of the doubt and assume that, though the very vague term "violation" is hard to pin down, there may be a violation of some kind in every pun; Still, if a linguistic violation that causes no harm were funny, then standard, everyday typos also would be, riht? And so would any misinterpretation of meaning: "I've got a cold." "Oh, here's a blanket." "No - I said I've GOT a cold, not I'm feeling cold."


We could go on with the counterexamples, but we plan to belabor that point elsewhere. For now, as we said, our point is methodological: examples and counterexamples are evidence that can be marshaled to support or disconfirm theories. On one hand, an accumulation of good examples over time, joined with a failure to find good counterexamples is strong evidence that we should provisionally accept a theory given no better alternatives ... On the other hand, one solid counterexample is evidence enough that a theory must either be discarded or, at the least, amended.

Now we will briefly address Hempelmann's objection. Hempelmann suggests that our model is unfalsifiable just because it is an evolutionarily inspired theory. The motivation for this common criticism of evolutionary thinking is a good one, as the challenge of falsifiability is a general concern that rings especially true for some biological traits that, like humor and other psychological traits, don't leave obvious fossil remains. However, our model is not only an evolutionary explanation. It is a theory, also, of the modern day trait of humor, a trait to which we have as much empirical access as we do to any other modern psychological trait. The theory in Inside Jokes is presented in cognitive and emotional terms -- some of which are new (such as epistemic commitment), but others of which are commonly understood modern psychological theoretical terms, or only slight modifications thereof (e.g. just-in-time spreading activation, misattribution of arousal, and our use of the notions of working memory and mental spaces). The fact that our theory is evolutionarily inspired only gives us a perspective from which to reason about the trait and to help make our hypothesis convincing. The same perspective -- what is it good for? -- has helped many even pre-Darwinian thinkers correctly deduce the functions of many biological traits. William Harvey, famously, was able to explain the working of the heart and the circulatory system in 1628 by wondering what it was for. Nonetheless, despite the fact that we cannot look into history and determine with certainty the trajectory along which the trait of humor evolved, the theory of how the modern-day trait works is certainly falsifiable by observation and experimentation in modern populations. And, if it turns out that the trait works roughly how we propose it might, then it will be good for what we think it is good for... and if that is the case, then we will have good evidence -- knowing what it actually is good for -- that our evolutionary explanation (it was refined and preserved in our species by being good for that) is also likely.

Ideas are not set in stone. When exposed to thoughtful people, they morph and adapt into their most potent form. TEDWeekends will highlight some of today's most intriguing ideas and allow them to develop in real time through your voice! Tweet #TEDWeekends to share your perspective or email to learn about future weekend's ideas to contribute as a writer.