The facial feedback hypothesis: what Strack's pen test showed, and what the replication changed

The facial feedback hypothesis says that movements of the face feed back into the brain and shape what you feel. The weak form says that a smile makes a pleasant moment a little more pleasant. The strong form says that the smile creates the feeling on its own. Only the weak form has support.

• Charles Darwin proposed in 1872 that free expression of an emotion intensifies it, and repression softens it.
• William James proposed in 1884 that the feeling of bodily changes is the emotion, with the sentence we feel sorry because we cry.
• Strack, Martin and Stepper tested the face-only form in 1988 with a pen held in the teeth or in the lips, and the teeth group rated cartoons as funnier.
• A registered replication in 17 labs in 2016 found no systematic effect, and later work points to camera awareness as one moderator.
• A 2019 meta-analysis of 138 studies found a small robust effect, with d equal to 0.2 overall.
• Botox paralysis of frown muscles slows reading of angry and sad sentences and lowers amygdala response, but people with facial paralysis still report emotions, so feedback modulates rather than creates.

What the hypothesis claims

The hypothesis concerns the face only. It says that activation of facial muscles sends feedback to the brain, and that feedback changes emotional experience. Inhibition of the face removes that input and dampens the experience.

Two forms matter for exams, and they predict different outcomes.

Only the weak form has support. People with facial paralysis still report emotions, and Botox changes experience only in some conditions. For the broader picture of what emotions are, see work on basic emotions.

The idea comes from Darwin and James. Darwin wrote that the free expression by outward signs of an emotion intensifies it, and that even the simulation of an emotion tends to arouse it. James wrote that our feeling of the same bodily changes as they occur is the emotion. Later writers narrowed the claim to the face, and the term facial feedback hypothesis entered use around 1980.

Facial feedback differs from the James-Lange theory. James-Lange concerns the whole body: heart, breathing, gut, muscles, and skin. Facial feedback concerns the face alone. It also differs from the Cannon-Bard claim that body and feeling run at the same time through separate paths, and from the two-factor claim that arousal plus a label creates the emotion. The exam sequence for facial feedback is expression then change in feeling.

Strack's pen study in 1988

Strack, Martin and Stepper wanted a test without asking people to pose an emotion. They used a cover story about difficulty of tasks without use of the hands. Participants held a pen in the mouth in one of three ways while they rated cartoons.

1. 1.Hold the pen with the teeth, without the lips touching it, which activates the smile muscles.
2. 2.Hold the pen with the lips, without the teeth touching it, which contracts the lip muscles and prevents a smile.
3. 3.Hold the pen in the non-dominant hand, as the control.

The last task was the real measure. Each group rated how funny the same cartoons were. The teeth group gave higher funniness ratings than the lips group. The authors read this as support for facial feedback, because no emotional state was asked for, and the muscle positions were set by the pen, not by an instruction to smile or frown.

The muscle logic uses four names that exams test. The teeth position uses the zygomaticus major and the risorius, the smile muscles. The lips position uses the orbicularis oris, the muscle around the mouth, and it blocks the smile. The corrugator supercilii furrows the brow in frowning, and the orbicularis oculi acts around the eyes in a full smile. A pen study works because it sets these muscles without naming an emotion.

The 17-lab replication and what followed

In 2016 a large registered replication report ran the pen procedure again in 17 labs in different countries and cultures. It found no systematic effect of pen position on funniness ratings. The result did not show that the first study was faked. It showed that the effect, if present, is small and sensitive to method.

One identified moderator is observation. Noah, Schul and Mayo reported in 2018 that a camera directed at participants removed the effect, and that the original result returned under the no-camera condition. When a camera watches participants, they hold the pen and rate in a different way, which matters for a cover story that depends on not knowing the aim.

The best summary now is the 2019 meta-analysis by Coles, Larsen and Lench. It combined 138 studies and found a small but robust effect, with no sign of publication bias. The size was d equal to 0.2 overall. Without an emotional stimulus the size was 0.32, and with one it was 0.13. In plain terms, the face has more room to start a feeling when little else drives it, and less room when a strong film or cartoon already sets the feeling.

For study, remember the order as pen position, then rating. If a question asks for the sequence tested by Strack, the answer is facial posture first, amusement rating second. If it asks what the 2016 result means, the answer is failure to replicate a small effect under stricter control, not proof that the face never matters.

What the Botox studies add

Botox offers a cleaner causal test than instruction to smile, because it paralyzes muscle without asking for control. Botulinum toxin type A blocks release of acetylcholine at the neuromuscular junction, so the motor command leaves the brain but the muscle stays still, and no feedback returns. Restylane, a filler with no muscle effect, serves as a control for getting a cosmetic procedure.

Havas and colleagues tested 41 first-time patients receiving Botox in the corrugator supercilii, the frown muscle, in 2010. Each person read angry, sad, and happy sentences before treatment and again two weeks after. Reading times for angry sentences rose after Botox, with t of 2.33, p of 0.025, and d of 0.22. Reading times for sad sentences rose as well, with t of 2.35, p of 0.024, and d of 0.23. Happy sentences did not change. The result is selective, which argues against a general slowing account.

Davis and colleagues compared 33 Botox patients with 35 Restylane patients in 2010. They used positive, negative, and mildly positive video clips before and after treatment. Within the Botox group alone, there was no drop for the most positive and most negative clips. Between groups, the Botox group showed a relative decrease, driven by a drop for mildly positive clips in the Botox group and a rise for negative clips in the control group. The authors read this as mixed support, consistent with a weak form in which feedback matters more when the emotional impulse is mild.

Brain data point the same way. Hennenlotter and colleagues found that Botox in the corrugator reduced amygdala and brainstem response during imitation of angry faces. The face input reaches circuits that process emotion, but it does not switch them on alone.

Two limits keep the claim modest. First, injection site differs by design, because Botox goes in upper-face muscles and Restylane goes in folds near the mouth, so site and drug stay linked. Second, only women took part in these patient samples, because few men sought the procedures at the time, so generalization across gender rests on theory rather than data.

Where feedback is not needed

Feedback helps in some cases, but it is not required. Keillor and colleagues reported in 2002 that a patient with total facial paralysis from Guillain-Barre syndrome still reported normal emotional experience. Davis and colleagues note the limit of one case, because pre-paralysis status was unknown and the disorder may have wider effects, yet the direction is clear. Later work on facial nerve paralysis found no large difference in experience either. A strong form that says no face means no feeling fails against these cases.

This is also where students mix terms. James-Lange says bodily feedback is the emotion. Facial feedback says face feedback shapes the emotion a little. Cannon-Bard says body change and feeling run together through separate paths. Schachter and Singer say arousal plus a label sets the emotion. If a question gives full-body arousal, choose James-Lange. If it gives smile or frown only, choose facial feedback. If it gives simultaneous body and feeling, choose Cannon-Bard.

To retain the set, learn the members against each other rather than one by one. Our guide to memorise psychology terms shows how to build one list and retrieve each pair from memory instead of rereading it.

How to use this for exams

Examiners test three things: the definition, the pen procedure, and the replication. Keep each answer short and tied to a study.

First, define the weak form in one sentence. The face feeds back and modulates experience, but it does not create it alone. If the question asks for strong versus weak, name both and say only the weak form has support.

Second, describe the pen positions with muscles. Teeth means smile muscles active. Lips means smile blocked. Hand means control. Outcome means funniness rating. Do not add mood scales that the study did not use.

Third, state the replication plainly. Seventeen labs found no systematic effect in 2016. Camera presence removed the effect in 2018 work. A 2019 meta-analysis found d of 0.2 overall. Botox work shows selective slowing and lower amygdala response, not a full block of feeling.

A sorting question helps. If the stem mentions holding a pen, Botox in the corrugator, or inability to mimic, the tested idea is facial feedback. If it mentions heart rate, sweating, or gut response as the cause of fear, it is James-Lange. If it mentions the thalamus sending separate signals to body and cortex, it is Cannon-Bard. If it mentions the same arousal labeled as joy in one context and anger in another, it is two-factor.