Emotion · about 5 minutes
Backwards Feelings
You smile because you are happy. Obviously. The feeling comes first and the face reports it, the way a thermometer reports a room.
In 1884 William James proposed the reverse -- that you feel afraid because you notice yourself trembling, and that the emotion just is the perception of the body's change. This room tests the smallest, most testable version of that, on you, and then tells you what happened when seventeen laboratories tried it at once.
You will need a pen or a pencil. Go and get one -- the room does not work without it and takes about three minutes with it. If that is not practical, everything below the interactive stands on its own and the story is arguably better than the demonstration.
Two rounds of four short jokes. Before each round we tell you how to hold the pen; rate each joke honestly, and nobody is scoring your sense of humour.
Ready when you are
Your two averages
The pen was doing something to your face
Held in your teeth, a pen forces the zygomaticus major -- the muscle that pulls your mouth into a smile -- into roughly the position it takes when you are amused. Held in your lips, it engages orbicularis oris and makes smiling mechanically difficult.
Neither instruction mentions smiling, which is the elegance of it. Fritz Strack, Leonard Martin and Sabine Stepper published the design in 1988, told participants it was a study about how disabled people write, and found that people rated cartoons as funnier in the teeth condition. The face had been put into a shape, and the feeling had followed.
It became one of the most famous demonstrations in social psychology. Textbooks, TED talks, business books about smiling your way to happiness.
Then 2016 happened
Seventeen laboratories, in eight countries, ran the pen-in-mouth study again at the same time, on nearly two thousand participants, using a protocol Strack himself reviewed in advance. This is a registered replication report: everything preregistered, all results published whichever way they come out.
The effect did not appear. Pooled across all seventeen labs, the difference between teeth and lips was indistinguishable from zero.
That result did roughly as much to change psychology as the original did. It arrived in the middle of the replication crisis, it concerned a beloved finding, and it was done fairly and in public with the original author's cooperation. Strack, for his part, pointed out a genuine confound: the replications filmed participants, and being on camera may suppress exactly the spontaneous responses the task depends on.
Which is where it gets properly interesting
Because it did not end there, and this is the part that gets left out of the telling. A later multi-laboratory project went back to facial feedback with several methods -- including simply asking people to smile deliberately, and having them mimic photographs -- rather than only the pen. Those methods did produce small but consistent effects on reported emotion.
So the honest state of play is not "facial feedback is dead" and it is certainly not "smiling makes you happy". It is narrower and more careful: deliberately arranging your face appears to nudge reported feeling a little, and the specific pen trick -- the one that made the textbooks -- does not reliably do it.
Which is why this room's needle sits where it does, and why your own two averages up there are worth almost nothing as evidence. Eight jokes on one person, who knew what was being tested, is not a study. It is a way of finding out what the study felt like from the inside -- and that is a genuinely useful thing to have felt before reading about effect sizes.
What survives, and it is not small
Strip away the pen and James's actual claim is in far better health than his most famous demonstration.
The body is not a bystander to emotion. You met the machinery two rooms ago: the sense of your own insides from Exhibit № 34 feeds directly into how you feel, and people who read their bodies differently report their emotions differently. Give someone adrenaline without telling them and they will look around for something to be upset about. The arousal dial you are about to meet in Exhibit № 37 is, in the end, a reading taken from your heart and your gut.
What has not survived is the tidy causal arrow. Emotion does not run body → feeling, and it does not run feeling → body. It appears to be a loop with the brain's predictions in it -- your brain generating a best guess about what is happening and what state your body ought to be in, and updating both from whatever comes back. Which makes the next room, about a sugar pill that treats real pain, less of a change of subject than it looks.
Take it further
The original is Strack, Martin & Stepper (1988). The replication is Wagenmakers and colleagues (2016), and it is worth reading for the format alone -- seventeen labs, one protocol, published regardless of outcome, which is what the field looks like when it is behaving. The later multi-method project is the Many Smiles Collaboration (Coles et al., 2022). William James's original essay, What is an Emotion? (1884), is short, strange, and much better written than most of what followed it.
One honest note on your own numbers: you knew what was being manipulated, which the real participants did not, and expectation is a powerful thing -- as the very next room is about to demonstrate at some length. If your teeth round scored higher, the most likely explanation is that you knew it was supposed to.