Facial Feedback Hypothesis

What it is

The hypothesis is old and simple. If the body takes part in producing emotion, then arranging the body into the shape of an emotion should push the emotion a little way into being. The face is the obvious place to test it, because people can hold an expression without knowing they are doing so.

The demonstration that made it famous was reported by Strack, Martin and Stepper in 1988. Participants held a pen between their teeth, which forces the mouth into a smile-like shape, or between their lips, which prevents one, and then rated how funny cartoons were. The teeth group rated the cartoons as funnier.

The idea matters well beyond cartoons because other theories lean on it. The three-stage account of emotional contagion proposed by Elaine Hatfield and colleagues has a feedback stage in the middle: a person copies another's expression, the copied expression changes how they feel, and the feeling has then spread. If facial feedback is weak, that middle stage is weak, and the standard explanation of how moods pass between people loses a support while the observation that moods converge stays where it was.

The record since 1988 has three parts. A registered replication report led by Eric-Jan Wagenmakers and colleagues, published in 2016 in Perspectives on Psychological Science, ran 17 direct replications of the first study of the 1988 paper with one vetted protocol. The pooled difference between the teeth and lips conditions was 0.03 units on a ten-point scale, with a confidence interval from minus 0.11 to 0.16, against a difference of 0.82 units in the original. A meta-analysis by Nicholas Coles, Jeff Larsen and Heather Lench, published in 2019 in Psychological Bulletin, pooled 138 studies and found a small but reliable effect of facial manipulations on emotional experience. A preregistered adversarial collaboration between proponents and sceptics, the Many Smiles Collaboration, reported in 2022 in Nature Human Behaviour with 3,878 participants in 19 countries. Its authors concluded that a facial mimicry task and a voluntary facial action task could both amplify and initiate feelings of happiness, and that the evidence was less conclusive when the expression was manipulated unobtrusively with a pen in the mouth.

In effect

In the laboratory

The pen task is the case to remember, because it shows how a result can be famous and fragile at once. The original difference was 0.82 units. Seventeen laboratories using one protocol found 0.03. A 2018 paper by Noah, Schul and Mayo reported that the original result reappeared when participants were not being filmed, and that the cameras used in the replication laboratories may have changed what the task did. That is a single study offering an explanation, and it should be read as a hypothesis about the failure and not as a rescue of the pen task.

In everyday life

The version that has held up is the plain one. Asked to pose a smile deliberately, or to copy a smiling face, people report feeling slightly happier, and the Many Smiles Collaboration found that effect in its preregistered tasks. The effect is small. A person who forces a smile on a bad day should expect a nudge and not a change of mood, and the evidence says nothing about whether a nudge of that size lasts.

In commerce and service

Service training often rests on two separate beliefs. One is that smiling changes the person who smiles. The other is that it changes the person who is smiled at. Facial feedback bears only on the first. The second belongs to emotional contagion, and the evidence there is split. In a staged service encounter reported by Thorsten Hennig-Thurau and colleagues in 2006, how much actors smiled had no effect on customers' emotions, while the genuineness of the display did. A smile can therefore change the smiler a little and still leave the customer where they were, which is why this page and the one on emotional contagion keep the two questions apart.

What it does not say

It does not say that holding a pen in the teeth makes cartoons funnier. The 2016 registered replication found no systematic effect.

It does not say that the hypothesis is false. Meta-analytic and adversarial evidence points to a small effect of deliberate and imitative expressions on self-reported feeling.

It does not say that a forced smile changes how another person feels. That is a separate claim about contagion, and the best-known service-encounter test of it found that the amount of smiling did nothing.

It does not say that the effect is large, lasting or useful as a way of managing mood. The studies measure reported feelings shortly after a manipulation.

---