The majority of humor experts today subscribe to some variation of the incongruity theory, the idea that humor arises when there’s an inconsistency between what people expect to happen and what actually happens. Incongruity has a lot going for it—jokes with punch lines, for example, fit well. But scientists have found that in comedy, unexpectedness is overrated. In 1974, two University of Tennessee professors had undergraduates listen to a variety of Bill Cosby and Phyllis Diller routines. Before each punch line, the researchers stopped the tape and asked the students to predict what was coming next, as a measure of the jokes’ predictability. Then another group of students was asked to rate the funniness of each of the comedians’ jokes. The predictable punch lines turned out to be rated considerably funnier than those that were unexpected—the opposite of what you’d expect to happen according to incongruity theory.
There’s another problem with these theories. While they all have their strengths, they share a major malfunction: They can’t explain why some things are not funny. Accidentally killing your mother-in-law would be incongruous, assert superiority, and release pent-up tensions, but it’s hardly a gut buster if you have to explain the catastrophe to your wife. ...
Working with his collaborator Caleb Warren and building from a 1998 HUMOR article published by a linguist named Thomas Veatch, [Peter McGraw] hit upon the benign violation theory, the idea that humor arises when something seems wrong or threatening, but is simultaneously OK or safe.
--Peter McGraw and Joel Warner, Slate, on the mystery of humor