Do you figure out how you feel by watching what you do?

Sometimes, yes, according to self-perception theory, but only when you don't already know how you feel. We usually assume feelings come first and behavior follows. The psychologist Daryl Bem proposed that when you have no prior attitude about something, or your emotional response is ambiguous, you work out what you feel the way you'd work out a stranger's feelings: by watching the behavior and inferring the attitude behind it, except that the behavior is your own.

Bem tested this with a clever twist on a classic experiment. Outside observers listened to a recording of a man enthusiastically praising an extremely boring task, and were told he'd been paid either $20 or just $1 to say so. Observers who heard he'd only gotten $1 assumed he must have genuinely enjoyed the task, since $1 wasn't enough reason to fake it; observers who heard about the $20 assumed he was just saying it for the money. These outside observers had zero access to the man's actual inner feelings, yet they guessed his attitude correctly just by watching his behavior and the situation around it. If strangers can infer an attitude from behavior alone, Bem reasoned, people may do the same with themselves.

Other researchers went further and manipulated behavior directly. In James Laird's 1974 experiments, people were secretly guided into smiling or frowning by moving specific facial muscles, without being told what expression they were making. They then reported feeling happier during the smiling trials and angrier during the frowning trials, and even rated cartoons as funnier while smiling than while frowning. The expression came first; the feeling followed.

A later study pushed this into more serious territory. Researchers had people secretly smile, by holding a pencil in their mouth, while looking at photos of unfamiliar black or white men, then measured implicit racial bias with a standard test. People who'd been nudged into smiling at photos of black men showed less implicit prejudice afterward than those nudged into smiling at photos of white men. Treat such effects with some caution, though: in 2016 a 17-lab replication failed to reproduce a classic pen-in-mouth smiling study, and facial feedback is now thought to have only a small effect on emotion.