1

The Many Smiles Collaboration: A Multi-Lab Foundational Test of the Facial Feedback Hypothesis

Stage 1 Registered Report

Submitting to Nature Human Behaviour

Corresponding author: Nicholas A. Coles (colesn@vols.utk.edu), Department of Psychology, University

of Tennessee, Knoxville, USA

*Nicholas A. Coles1, David S. March1, Fernando Marmolejo Ramos2, Hassan Banaruee3, Natalie Butcher4,

Mikael Cavallet5, Nikolay Dagaev6, Daniel Eaves7, Francesco Foroni8, Elena Gorbunova6, Pascal Gygax9,

José Antonio Hinojosa Poveda10, Ayumi Ikeda11, Omid Kathin-Zadeh3, Asil Ali Özdoğru12, Michal

Parzuchowski13, Susana Ruiz-Fernandez14, 15, Bidisha Som16, Isabel Suarez17, Natalia Trujillo18, Sandra

Trujillo18, Tim van der Zee19, Cristina Villalba-García10, Megan Willis8, Yuki Yamada20, Phoebe Ellsworth21,

Lowell Gaertner1, Fritz Strack22, Marco Tullio Liuzza23, Marco Marozzi24

1 Department of Psychology, University of Tennessee, Knoxville, United States of America

2 School of Psychology, University of Adelaide, Australia

3 Language Department, Chabahar Maritime University, Iran

4 School of Social Sciences, Humanities and Law, Teesside University, United Kingdom

5 Department of Psychology, Integrated Colleges of Ribeira Valley, Brazil

6 Laboratory for Cognitive Psychology of Digital Interfaces User, National Research University Higher

School of Economics, Russia

7 School of Health and Social Care, Teesside University, United Kingdom

8 School of Psychology, Australian Catholic University, Strathfield, Australia

9 Department of Psychology, University of Fribourg, Switzerland

10 Departamento de Psicología Experimental, Procesos Cognitivos y Logopedia, Universidad

Complutense, Madrid, Spain

11 Graduate School of Human-Environment Studies, Kyushu University, Japan

12 Department of Psychology, Üsküdar University, Turkey

2

13 Department of Psychology, Center of Research on Cognition and Behavior, SWPS University of Social

Sciences and Humanities, Poland

14 FOM Hochschule für Oekonomie & Management, Germany

15 Leibniz-Institut für Wissensmedien, Tübingen, Germany

16 Department of Humanities and Social Sciences, Indian Institute of Technology Guwahati, India

17 Department of Psychology, Universidad del Norte, Barranquilla, Colombia.

18 Mental Health Group, National Faculty of Public Health, University of Antioquia, Colombia

19 Human-Technology Interaction Group, Eindhoven University of Technology, Netherlands

20 Faculty of Arts and Science, Kyushu University, Japan

21 University of Michigan, United States of America

22 Department of Psychology, University of Würzburg, Germany

23 Department of Medical and Surgical Sciences, University Magna Graecia of Catanzaro, Italy

24 Department of Environmental Sciences, Informatics and Statistics, Ca' Foscari University of Venice,

Italy

3

Abstract

The facial feedback hypothesis suggests that an individual’s subjective experience of emotion is

influenced by their own facial expressions. However, researchers currently face conflicting narratives

about whether this hypothesis is valid. A large collaborative effort consistently failed to replicate a seminal

demonstration of the facial feedback hypothesis, but meta-analysis suggests the effects are real.

Consequently, we conducted a foundational test of the facial feedback hypothesis, wherein a large group

of researchers specified the best way(s) to test the hypothesis and used this information to design and

execute an international multi-lab experiment. Two pilot studies suggested that smiling could both magnify

ongoing feelings of happiness and initiate feelings of happiness in otherwise non-emotional scenarios.

Next, 18 labs from 17 countries will examine whether these findings can be replicated.

4

The Many Smiles Collaboration: A Multi-Lab Foundational Test of the Facial Feedback Hypothesis

The facial feedback hypothesis suggests that an individual’s subjective experience of emotion is

influenced by their facial expressions. For example, smiling should typically make an individual feel

happier and frowning should make them feel sadder. Researchers suggest that these effects emerge

because facial expressions provide sensorimotor motor feedback that (a) contributes to the sensation of

an emotion1–6, (b) primes emotion-related concepts, facilitating emotion reports7,8, and/or (c) serves as a

cue that individuals use to make sense of ongoing emotional feelings9–12. This facial feedback hypothesis

is notable because it supports broader theories that contend emotional experience is influenced by bodily

feedback from the peripheral nervous system13–15, as opposed to experience and bodily sensations being

independent manifestations of an emotional reaction16–18. Furthermore, this hypothesis supports claims

that facial feedback interventions—for example, simply smiling more frequently—can help manage

distress19,20, improve well-being21,22, and reduce depression23-43.

Researchers, however, currently face conflicting narratives about the validity of the facial

feedback hypothesis. On one hand, a large collaborative effort consistently failed to replicate a seminal

demonstration of facial feedback effects44. On the other hand, meta-analytic evidence supports the facial

feedback hypothesis45. In this paper, we argue that neither of these observations adequately test whether

the facial feedback hypothesis is fundamentally valid. To provide a foundational test of the facial feedback

hypothesis, a large and diverse group of researchers came together to (1) specify their beliefs regarding

when facial feedback effects should most reliably emerge, (2) determine the best way(s) to test those

beliefs, and (3) use this information to design and execute an international multi-lab experiment.

Failure to Replicate the Pen-in-Mouth Effect

In one seminal facial feedback study, participants viewed humorous cartoons while holding a pen

in their mouth in a manner that either produced a smile (pen held in teeth) or prevented smiling (pen held

by lips)46. Consistent with the facial feedback hypothesis, participants manipulated to smile reported

feeling more amused by cartoons than those prevented from smiling. These findings were influential

because previous facial feedback experiments often explicitly instructed participants to pose a facial

expression, raising concerns about demand characteristics47–49. Furthermore, facial feedback theorists

disagreed about whether these effects could occur outside of awareness5,9,50. Since participants in this

5

study were presumably unaware they were smiling, the authors concluded that facial feedback effects

were not driven by demand characteristics or awareness of the emotional nature of their facial poses.

More recently, a collaborative effort involving 17 labs and 1,894 participants consistently failed to

replicate the pen-in-mouth effect44. However, the implications of this failure-to-replicate for the facial

feedback hypothesis are unclear. One obvious possibility is that the facial feedback hypothesis may be

false. However, this conclusion is beyond the scope of the direct replication because it was limited to a

specific test of the facial feedback hypothesis. Indeed, the replicators were careful to point out that their

findings “do not invalidate the more general facial feedback hypothesis”44. Similarly, while arguing that the

pen-in-mouth effect is unreliable, some researchers conceded that “other paradigms may produce

replicable results”51.

A second possibility is that both the facial feedback hypothesis and the pen-in-mouth effect are

true. If this is the case, researchers must determine why others were unable to replicate these real

effects. One suggestion is that the replicators did not perform a true direct replication because they

deviated from the original study by overtly recording participants52. According to this explanation,

awareness of video recording induces a self-focus that interferes with participants’ internal experiences

and emotional behavior52. Evidence for this claim is mixed. Meta-analysis did not reveal that facial

feedback studies using overt video recording had smaller effects than studies that did not45. However, the

only experimental manipulation of video recording presence reported marginal evidence for this claim53.

A third possibility is that the facial feedback hypothesis is true, but not under the context

examined in the pen-in-mouth study. Perhaps facial feedback effects only occur when participants are

aware they are posing an emotional facial expression, an explanation that the pen-in-mouth study was

designed to eliminate. Alternatively, perhaps the pen-in-mouth task is not a reliable facial feedback

manipulation. Indeed, some theorists have predicted that facial feedback effects will only emerge when

the patterns of facial movements resemble a prototypical emotional facial expression2,54–58. Although the

pen-in-mouth task is designed to make participants express happiness, previous research indicates that

this task does not reliably produce a prototypical expression of happiness59—a smile that is accompanied

by the contraction of the muscles surrounding the eyes60. Consistent with the prototypicality explanation,

6

these researchers only detected a significant facial feedback effect when participants performed a variant

of the pen-in-mouth procedure that produced a more prototypical smile1.

To summarize, the failure-to-replicate does not provide a test of the fundamental validity of the

facial feedback hypothesis because it is limited to a narrow operationalization of the facial feedback

hypothesis. Furthermore, researchers can offer several explanations for why the facial feedback

hypothesis is valid despite difficulty replicating the pen-in-mouth effect. Consequently, to examine the

facial feedback hypothesis more comprehensively, researchers turned their attention to the cumulative

evidence for the facial feedback hypothesis.

Cumulative Evidence for the Facial Feedback Hypothesis

Amid the uncertainty created by the failure-to-replicate, a meta-analysis was performed on 286

effect sizes from 137 studies testing the effects of various facial feedback manipulations on emotional

experience45. Results indicated that facial feedback manipulations have a small, but highly varied, effect

on emotional experience. Notably, this effect could not be explained by publication bias. Published (s =

117) and unpublished studies (s = 20) reported effects of a similar magnitude, a variety of publication bias

analyses failed to uncover significant evidence of publication bias, and bias-corrected overall effect

estimates were significant. This meta-analysis revealed that facial feedback effects tend to be larger in

some circumstances (e.g., in the absence vs. presence of emotional stimuli), but it did not identify any

clear conditions where facial feedback effects do not emerge. Consequently, this meta-analysis did not

provide an explanation for why facial feedback effects were not observed in the replications of the pen-in-

mouth study.

Although the meta-analysis seems to suggest that the facial feedback hypothesis is

fundamentally valid, there are at least three limitations that undermine this conclusion. First, although

significant evidence of bias was not detected, this cannot be taken as evidence of a lack of bias in the

facial feedback literature given that publication bias analyses often have poor Type II error rates61–63.

Consequently, it is possible that a seemingly robust facial feedback effect is driven by studies that used

1 Although some argue that the pen-in-mouth task does not reliably produce facial feedback effects because the posed expressions of happiness are not prototypical, participants who pose these happy expressions are often compared to participants who hold a pen in their mouth in a manner that suppresses smiling. Although not at all facial feedback theories predict that suppressing smiling will lead to an attenuated experience of happiness, the lack of an unaffected control group often makes it difficult to form conclusions about these effects.

7

undetected questionable research practices64. Second, it is possible that the overall effect size estimates

in this literature are driven by low-quality studies, such as studies that did not address demand

characteristics or used improper randomization procedures65. Third, even relatively similar subsets of

facial feedback studies varied beyond what would be expected from sampling error alone, meaning that

moderator analyses had lower power and potentially contained unidentified confounds. Consequently, the

meta-analysis could not reliable identify moderators that may help explain why some researchers fail to

observe facial feedback effects.

Foundational Tests and the Many Smiles Collaboration

Both the failure-to-replicate the pen-in-mouth effect and the meta-analysis have a unique set of

limitations that make it difficult to assess whether the facial feedback hypothesis is fundamentally valid. In

scenarios like these—which are not unique to the facial feedback hypothesis or psychology—we suggest

that researchers perform a foundational test. We define foundational test as the identification and testing

of researchers’ core beliefs regarding the conditions in which a hypothetical effect should most reliably

emerge. Importantly, these beliefs are specified a priori. Therefore, whether this predicted effect emerges

can provide insight into the degree that these core beliefs are correct. Failing to find the predicted effect,

of course, does not necessarily mean that the hypothesis or theory is invalid. However, it can provide

evidence that researchers’ core beliefs regarding this hypothesis or theory are invalid.

In some scenarios, researchers may illuminate that a foundational test of their predictions has

already been performed, meaning that this foundational test will be a direct replication of this previous

study. However, in other instances—for example, in the project described here—researchers may be led

to a novel methodology that is a more appropriate test of their core beliefs. With the epistemological goal

of a foundational tests in mind, the Many Smiles Collaboration—henceforth referred to as the MSC—was

formed.

The Many Smiles Collaboration.

The MSC is an international group of researchers—some advocates of the facial feedback

hypothesis, some critics, and some without strong beliefs—who came together to: (1) specify their beliefs

regarding when facial feedback effects should reliably emerge, (2) determine the best way(s) to test those

beliefs, and (3) use this information to design and execute an international multi-lab experiment.

8

Participating labs were first given two months to specify the details of this foundational replication in an

online document. Three project coordinators then used this information to develop a research protocol.

This research protocol was subsequently reviewed and vetted by three external consultants, one who

served as an advocate of the effect, one who served as a critic, and one who served as a neutral

reviewer. This agreed-upon methodology was then tested in two pre-registered pilot studies. After

concluding the pilot studies, labs were given the opportunity to suggest changes to the experimental

protocol. Next, eighteen participating labs from 17 countries will conduct the experiment following the

vetted protocol. At the end of the project, participating labs will be given the opportunity to upload

dissenting opinions or commentaries to the Open Science Framework.

The MSC agreed that one of the simplest necessary conditions for facial feedback effects to

emerge is that participants adopt a facial posture resembling an emotional expression and subsequently

provide self-reports of the associated emotional state. However, the MSC disagreed about (a) the degree

to which a posed expression must resemble a prototypical emotional expression, and (b) whether facial

feedback can initiate emotional experiences or only modulate ongoing emotional experiences. These

disagreements ultimately informed the final experimental design: a 2 (type of pose: happy or neutral) x 2

(facial feedback task: facial mimicry or voluntary facial action technique) x 2 (stimuli presence: present or

absent) design, with type of pose manipulated within-participants and facial feedback task and stimuli

presence manipulated between-participants (Figure 1). Notably, this specific combination of scenarios

has never been tested in a single facial feedback experiment, further demonstrating the distinction

between a foundational tests and direct replications.

9

Figure 1.

Experimental design of the MSC and two pilot studies.

Prototypicality of posed expression.

Although not all facial feedback researchers have been concerned about the degree to which

facial feedback patterns must resemble prototypical emotional expressions, attention to this

methodological detail was heightened in the 1980’s after a set of findings consistently failed to

demonstrate significant facial feedback effects66. In reply to these contradictory findings, several

researchers emphasized that facial feedback effects will only emerge when the facial feedback postures

resemble a prototypical emotional expression2,54–58. To provide an easy-to-follow task that would produce

more prototypical facial expressions, the MSC elected to use a facial mimicry paradigm, wherein

participants were shown images of actors displaying prototypical expressions of happiness and later

asked to mimic the expressions. However, this manipulation could not eliminate the possibility of demand

characteristics. Consequently, the MSC elected to also use a second manipulation, the voluntary facial

action technique, a subtler muscle-movement task that better limits demand characteristics but produces

what MSC members believed would be less prototypical expressions of happiness67.

Initiation vs. modulation of emotional experience.

The MSC also elected to manipulate whether participants were exposed to emotional stimuli

while engaging in the facial feedback tasks. This decision was informed by debates regarding whether

facial feedback can only modulate ongoing emotional experiences (i.e., in the presence of emotional

10

stimuli) or also initiate emotional experiences in otherwise non-emotional scenarios. Many theories predict

that facial feedback can only modulate ongoing emotional experience6,11,12,68,69. Indeed, one of the first

theories of facial feedback suggested that autonomic activity produced undifferentiated feelings of

positivity and negativity that were subsequently categorized into emotional categories based on patterns

of facial feedback11,12. For example, ongoing feelings of negativity would be categorized as anger if the

individual was scowling, but scowling could not initiate the experience of anger in the absence of ongoing

feelings of negativity. However, other facial feedback theories contend that facial feedback can also

initiate emotional experience5,7,8. For example, one theory posited that facial expressions can activate

innate affect programs, producing a set of coordinated emotional responses that lead to emotional

experience57.

Meta-analysis indicated that facial feedback effects were present both in studies that tested

initiation effects and studies that tested modulating effects45. Furthermore, this meta-analysis revealed

that facial feedback effects were larger in studies that examined initiating facial feedback effects.

However, the only two facial feedback studies to experimentally manipulate initiation vs. modulation

uncovered a conflicting pattern of results. In the first experiment, fearful and sad facial expression poses

neither modulated nor initiated experiences of fear and sadness66. In a second experiment, surprise

poses neither modulated nor initiated experiences of surprise70. Given pre-existing debates and these

conflicting sets of findings, the MSC manipulated some participants to view a series of positive images

while engaging in the facial poses.

Method

Participants were run in-person, and the experiment was presented through an online research

platform, Qualtrics.

Participants were told that the experiment investigated how physical movements and cognitive

distractors influenced mathematical speed and accuracy and that the computer would randomly assign

them to complete five movement tasks. The first, second, and last tasks were filler trials included to

ensure the cover story was believable (e.g., “place your left hand behind your head and blink your eyes

once per second for 10 seconds”). In the two critical tasks, participants were randomly assigned to pose

happy and neutral facial expressions (in randomized order) through either a facial mimicry procedure or

11

the voluntary facial action technique. While posing these expressions, some participants were randomly

assigned to view positive images. To reinforce the cover story, participants were provided with an on-

screen timer during all tasks. After each task, participants completed two simple filler arithmetic problems

and answered four items measuring happiness, four items measuring anger, and two filler items

measuring anxiety using a subset of the Discrete Emotions Questionnaire71. The orders of emotion self-

report items were randomized. Afterwards, participants answered questions regarding the difficulty of the

task and arithmetic problems to reinforce the cover story.

In the facial mimicry condition, participants were shown a 2x2 matrix of actors posing happy

expressions. Participants were then instructed to either mimic these expressions (happy condition) or to

maintain a blank expression (neutral condition). Having participants view the happy expression matrix

before both the happy and neutral trials ensured that any potentially confounding effects that images of

smiling people have on emotional experience were constant across mimicry trials. The photo matrix was

displayed for a minimum of 5 seconds, and participants indicated when they were ready to perform the

mimicry task. In the voluntary facial action technique condition, participants were instructed to either move

the corner of their lips up towards their ear, elevating their cheeks (happy condition) or maintain a blank

facial posture (neutral condition). In both the mimicry and voluntary facial action technique conditions,

participants were instructed to maintain the poses for 5 seconds, the approximate duration of

spontaneous happiness expressions72. After completing the five movement tasks, participants reported

their self-identified gender, age, and ethnicity.

To assess awareness of the purpose of the experiment, participants were asked a variety of

questions about their beliefs regarding the purpose of the experiment via open-ended response prompts.

These questions were part of a funneled debriefing, wherein participants were gradually informed of the

true nature of the study. Using two coders, each lab examined these responses to rate the degree each

participant was suspicious of the true purpose of the experiment (1 = “not at all aware” to 5 = “completely

aware”). Ideally, participants would be debriefed by an experimenter, but this would have prohibited labs

from running multiple participants at once. However, in-person debriefings were used in two pilot studies,

which indicated that participants were largely unaware of the purpose of the experiment (see Pilot

Studies).

12

As an indicator of the quality of the posed expressions, participants were asked at the end of the

experiment how closely they believed their expressions during the critical posing tasks resembled images

of prototypically happy or neutral facial expressions. Ideally, the quality of participants’ expression would

be assessed by covertly video recording participants and coding their expressions. However, this would

have been prohibitively costly given the large number of labs and participants in this study. Furthermore,

some members of the MSC expressed doubts about receiving ethical approval to covertly record

participants, and researchers are still debating whether awareness of video recording interferes with facial

feedback effects45,52,53,73. Consequently, participants were covertly recorded in one of the two pilot studies,

which indicated that most participants successfully posed the target facial expressions (see Pilot Studies).

Materials

Photos for the facial mimicry tasks were obtained from the Extended Cohn-Kanade Dataset74

(Figure 2). For the purposes of the project, it was necessary to use images that were (1) confirmed to

display prototypical emotional facial expressions by coders trained in the Facial Action Coding System75,

and (2) given permission by the photographed participants to openly share. Using images that met these

criteria, the 2x2 matrix of happy poses were created.

13

Figure 2.

Image matrix of actors posing happy expressions shown during the mimicry condition.

Participants who were manipulated to view positive photos during the two critical tasks viewed

images chosen from an database used in previous research76. This image database included 100 images

from the Internet and the International Affective Picture System77 that were rated on how good they were

(50 coders) and how bad they were (51 coders) on a 7-point scale (1 = “not at all” to 7 = “extremely”).

From these images, two positive image matrixes that were matched in their contents were created (Figure

3).

14

Figure 3.

Images shown to participants in the condition testing the initiation facial feedback hypothesis. One set of

images (either a or b) was shown during the happy and neutral trials (in randomized order).

For labs outside of the United States, experiment materials were translated into the following

languages using a back-translation process: Dutch, French, German, Hindi, Italian, Japanese, Persian,

Polish, Portuguese, Russian, Spanish, and Turkish.

Pilot Studies

Two pilot studies used a 2 (type of pose: happy or neutral) x 2 (facial feedback task: facial

mimicry or voluntary facial action technique) design. However, the two studies differed in whether

participants viewed emotional stimuli while engaging in the critical poses, with the first pilot testing

initiating effects (i.e., no stimuli present) and the second pilot testing modulating effects (i.e., stimuli

present; Figure 1). Together, data were collected from 206 participants (67% female; age M = 18.52 SD =

0.96). Patterns of data and inferences were identical across the two studies. Therefore, pooled analyses

are reported unless otherwise noted.

Test of the Facial Feedback Hypothesis

To examine whether facial feedback impacted self-reported happiness, a 2(type of pose: happy

or neutral) x 2(facial feedback task: facial mimicry or voluntary facial action technique) x 2(stimuli

15

presence: present or absent) mixed-effect ANOVA was fitted, with type of pose included as a within-

participant factor. Consistent with the facial feedback hypothesis, participants reported experiencing more

happiness after posing happy (M = 2.47, SD = 1.48) vs. neutral expressions (M = 1.93, SD = 1.18), F(1,

202) = 43.65, p < .001. No main effect for facial feedback task was detected, F(1, 202) = .38, p = .54.

There was a main effect for stimuli presence, wherein participants reported more happiness when positive

stimuli were present (M = 2.65, SD = 1.54) vs. absent (M = 1.72, SD = 0.94), F(1, 202) = 36.06, p < .001.

Results also revealed an interaction between facial feedback task and stimuli presence, wherein the

difference in self-reported happiness between the happy and neutral trials was larger in the presence vs.

absence of emotional stimuli, F(1, 202) = 5.79, p = .02.

The results from the facial feedback task x stimuli presence interaction suggests that modulating

facial feedback effects are larger than initiating effects. To further examine the modulation and initiation

hypotheses, 2 (type of pose: happy or neutral) x 2 (facial feedback task: facial mimicry or voluntary facial

action technique) mixed ANOVAs were separately fitted for each study. Results confirmed that facial

feedback both initiated (F(1, 98) = 15.56, p < .001) and modulated (F(1, 104) = 29.40, p < .001) emotional

experience. However, some may argue that the inclusion of the mimicry condition prevents a true test of

the initiation hypothesis since images of smiling actors may cause participants to experience happiness.

This seems unlikely given that participants did not report more happiness in the mimicry vs. voluntary

facial action technique condition. Nevertheless, follow-up analyses excluding the mimicry condition

confirmed that facial feedback can initiate emotional experience (F(1, 50) = 6.76, p = .01).

Participant Awareness

Although pilot results provide evidence of facial feedback effects, it is possible that these effects

are driven by demand characteristics. To assess demand characteristics, experimenters rated the degree

to which participants were aware of the purpose of the experiment based on their in-person funnel

debriefing responses (1 = “not at all aware” to 5 = “completely aware”). Results indicated that participants

generally exhibited low awareness of the purpose of the experiment (M = 1.54, SD = 0.96), with 85% of

participants characterized as not at all or slightly aware of the purpose of the experiment. Nevertheless,

all analyses reported above were re-run excluding participants who exhibited any degree of awareness

(i.e., had an awareness score higher than 1). All results were robust except for the interaction between

16

facial feedback task and stimuli presence. Consequently, there is not reliable evidence that modulating

facial feedback effects are larger than initiating effects.

To examine whether participant awareness varied across conditions, awareness ratings were

modeled using a 2 (facial feedback task: facial mimicry or voluntary facial action technique) x 2 (stimuli

presence: present or absent) ANOVA. Contrary to the MSC’s prediction, results did not indicate that

participants were more aware of the purpose of the experiment in the mimicry (M = 1.47, SD = 0.85) vs.

voluntary facial action technique (M = 1.61, SD = 1.06) conditions, F(1, 202) = 1.04, p = .31.

Unexpectedly, participants exhibited more awareness of the experiment’s purpose when emotional stimuli

were present (M = 1.68, SD = 1.03) as opposed to absent (M = 1.39, SD = 0.86), F(1, 202) = 4.78, p

= .03. No interaction between these two factors was detected, p = .49.

Quality of Posed Expressions

In pilot study 1, participants were covertly recorded to assess the quality of their posed

expressions. For participants who consented for their videos to be analyzed (n = 80), videos recordings of

their neutral and happy posing trials were processed through Noldus’ FaceReader 7.0, which provided

continuous ratings of expressed happiness (0 to 1) during the trials78. FaceReader failed to code videos

from two participants, leaving a final sample of 78 pairs of videos.

Ratings of expressed happiness were modeled using a 2 (type of pose: happy or neutral) x 2

(facial feedback task: facial mimicry or voluntary facial action technique) mixed-effect ANOVA, with type

of pose included as a within-participant factor. As expected, participants expressed more happiness

during the happy (M = .65, SD = .27) vs. neutral (M = .03, SD = .06) trails, F(1, 75) = 445.83, p < .001.

Participants also expressed more happiness in the mimicry (M = .39, SD = .39) vs. voluntary facial action

technique (M = .29, SD = .34) conditions, F(1, 75) = 15.19, p = .001. These main effects were qualified by

a significant interaction, wherein the difference in expressed happiness between the happy and neutral

trials was larger in the mimicry condition, F(1, 75) = 15.32, p < .001. These patterns of results are

consistent with the MSC’s prediction that the mimicry condition would produce more prototypical

expressions of happiness, although our results so far provide no evidence that high vs. low quality poses

influences the magnitude of facial feedback effects.

17

Main Study

For the main study, all labs will use a 2 (type of pose: happy or neutral) x 2 (facial feedback task:

facial mimicry or voluntary facial action technique) x 2 (stimuli presence: present or absent) design, with

type of pose manipulated within-participants and facial feedback task and stimuli presence manipulated

between-participants (see Figure 1). We hypothesize that participants will report more happiness when

(a) posing happy vs. neutral expressions, and (b) when presented with emotional stimuli (i.e., main effects

for type of pose and stimuli presence). Of secondary interest, we hypothesize an interaction between type

of pose and stimuli presence, wherein facial feedback effects will be larger in the presence vs. absence of

emotional stimuli.

Each lab agreed to collect 100 participants for the main study, with an estimated total sample of

1,800 participants. Power analyses estimated that this sample provides over 95% power to detect the

hypothesized main effect of pose, over 90% power to detect the hypothesized main effect of stimuli

presence, and over 90% power to detect the hypothesized interaction between type of pose and stimuli

presence. Details regarding the power analysis are available in the pre-registration plan.

Data Analysis

Primary analyses.

Due to the nested nature of the experimental design (ratings nested within individuals, nested

within labs), data for the main study will be analyzed using linear multilevel modeling. More specifically,

happiness reports will be modeled with (a) type of pose, facial feedback task, and stimuli presence

entered as factors, (b) random intercepts for each laboratory and each participant, and (c) random slopes

for each laboratory. To examine possible interactions between the three factors, higher-order interactions

will be included in the model specified above while keeping the random structure maximal79.

Although not of primary interest, we will also examine whether facial feedback effects are

moderated by self-reported pose-quality. To test whether participants with higher self-reported pose

quality exhibit larger facial feedback effects, self-reported pose-quality will be entered as a factor with

random slopes. Last, to examine whether results are robust to participant awareness, all primary analyses

will be rerun using only participants who are completely unaware of the purpose of the experiment (as

confirmed by two coders).

18

References

1. Tomkins, S. S. Affect Imagery Consciousness: The Positive Affects. (Springer, 1962).

2. Tomkins, S. S. The role of facial response in the experience of emotion: A reply to Tourangeau

and Ellsworth. J. Pers. Soc. Psychol. 37, 1519–1531 (1981).

3. Izard, C. E. Facial expressions and the regulation of emotions. J. Pers. Soc. Psychol. 58, 487–498

(1990).

4. Izard, C. E. The Face of Emotion. (Appleton-Century-Crofts, 1971).

5. Ekman, P. Biological and cultural contributions to body and facial movement. in Anthropology of

the Body (ed. Blacking, J.) 34–38 (Routledge, 1979).

6. Scherer, K. R. The dynamic architecture of emotion: Evidence for the component process model.

Cogn. Emot. 23, 1307–1351 (2009).

7. Berkowitz, L. On the formation and regulation of anger and aggression: A cognitive-

neoassociationistic analysis. Am. Psychol. 45, 494–503 (1990).

8. Guenther, K. Mood and Memory. Am. Psychol. 36, 129–148 (1981).

9. Laird, J. D. & Bresler, C. The process of emotional experience: A self-perception theory. in Review

of Personality and Social Psychology: Emotion (ed. Clark, M. S.) 213–234 (1992).

10. Laird, J. D. & Crosby, M. Individual differences in self-attribution of emotion. in Thought and

Feeling: Cognitive Alteration of Feeling States (eds. London, H. & Nisbett, R. E.) 44–59

(Routledge, 1974).

11. Allport, F. H. Feeling and emotion. in Social Psychology (ed. Allport, F. H.) 84–98 (Houghton

Mifflin Company, 1924).

12. Allport, F. H. A physiological-genetic theory of feeling and emotion. Psychol. Rev. 29, 132–139

(1922).

19

13. Friedman, B. H. Feelings and the body: The Jamesian perspective on autonomic specificity of

emotion. Biol. Psychol. 84, 383–393 (2010).

14. James, W. Discussion: The physical basis of emotion. Psychol. Rev. 1, 516–529 (1894).

15. Lange, C. G. Om sindsbevaegelser; et psyko-fysiologisk studie. (Lund, 1885).

16. Cannon, W. The James-Lange theory of emotions : A critical examination and an alternative

theory. Am. J. Psychol. 39, 106–124 (1927).

17. Cannon, W. Bodily Changes in Pain, Hunger, Fear and Rage. (D Appleton and Company, 1915).

18. Sherrington, C. S. Experiments on the value of vascular and visceral factors for the genesis of

emotion. Proc. R. Soc. London 66, 390–403 (1900).

19. Ansfield, M. E. Smiling when distressed: when a smile is a frown turned upside down. Personal.

Soc. Psychol. Bull. 33, 763–775 (2007).

20. Kraft, T. L. & Pressman, S. D. Grin and Bear It: The Influence of Manipulated Facial Expression on

the Stress Response. Psychol. Sci. 23, 1372–8 (2012).

21. Schmitz, B. Art-of-Living. 63, (2016).

22. Lyubomirsky, S. The how of happiness: A scientific approach to getting the life you want.

(Penguin, 2008).

23. Alam, M., Barrett, K. C., Hodapp, R. M. & Arndt, K. A. Botulinum toxin and the facial feedback

hypothesis: can looking better make you feel happier? J. Am. Acad. Dermatol. 58, 1061–1072

(2008).

24. Alves, M. C., Sobreira, G., Aleixo, M. A. & Oliveira, J. M. Facing depression with botulinum toxin:

Literature review. Eur. Psychiatry 335, 5290–5643 (2016).

25. Chugh, S., Chhabria, A., Jung, S., Kruger, T. H. C. & Wollmer, M. A. Botulinum toxin as a

treatment for depression in a real-world setting. J. Psychiatr. Pract. 24, 15–20 (2018).

20

26. Finzi, E. Botulinum Toxin for Depression More Than Skin Deep. Dermatologic Surg. 44, 1363–

1365 (2018).

27. Finzi, E. & Rosenthal, N. E. Emotional proprioception: Treatment of depression with afferent facial

feedback. J. Psychiatr. Res. 80, 93–96 (2016).

28. Finzi, E. & Rosenthal, N. E. Treatment of depression with onabotulinumtoxinA: A randomized,

double-blind, placebo controlled trial. J. Psychiatr. Res. 52, 1–6 (2014).

29. Finzi, E. & Wasserman, E. Treatment of depression with botulinum toxin A: A case series.

Dermatologic Surg. 32, 645–649 (2006).

30. Fromage, G. Exploring the effects of botulinum toxin type A injections on depression. Aesthetic

Nurs. 7, 315–317 (2018).

31. Hexsel, D. et al. Evaluation of self-esteem and depression symptoms in depressed and

nondepressed subjects treated with onabotulinumtoxinA for glabellar lines. Dermatologic Surg. 39,

1088–1096 (2013).

32. Krüger, T. H. C., Jung, S. & Wollmer, M. A. Botulinumtoxin - Ein neuer wirkstoff in der

psychopharmakotherapie? Psychopharmakotherapie 23, 2–7 (2016).

33. Lewis, M. B. & Bowler, P. J. Botulinum toxin cosmetic therapy correlates with a more positive

mood. J. Cosmet. Dermatol. 8, 24–26 (2009).

34. Magid, M. et al. Treating depression with botulinum toxin: A pooled analysis of randomized

controlled trials. Pharmacopsychiatry 48, 205–210 (2015).

35. Magid, M. & Reichenberg, J. S. Botulinum toxin for depression? An idea that’s raising some

eyebrows. Curr. Psychiatr. 14, 43–56 (2015).

36. Magid, M. et al. Treatment of major depressive disorder using botulinum toxin A: a 24-week

randomized, double-blind, placebo-controlled study. J. Clin. Psychiatry 75, 837–844 (2014).

21

37. Parsaik, A. K. et al. Role of botulinum toxin in depression. J. Psychiatr. Pract. 22, 99–110 (2016).

38. Reichenberg, J. S. et al. Botulinum toxin for depression: Does patient appearance matter? J. Am.

Acad. Dermatol. 74, 171–173 (2016).

39. Wollmer, M. A., Magid, M. & Kruger, T. H. C. Botulinum toxin treatment in depression. in Practical

Psychodermatology (eds. Bewley, A., Taylor, R. E., Reichenberg, J. S. & Magid, M.) 216–219

(John Wiley & Sons, 2014).

40. Wollmer, M. A. et al. Agitation predicts response of depression to botulinum toxin treatment in a

randomized controlled trial. Front. psychiatry 5, 36 (2014).

41. Wollmer, M. A. et al. Facing depression with botulinum toxin: A randomized controlled trial. J.

Psychiatr. Res. 46, 574–581 (2012).

42. Zamanian, A., Jolfaei, A. G., Mehran, G. & Azizian, Z. Efficacy of botox versus placebo for

treatment of patients with major depression. Iran J. Public Heal. 46, 982–984 (2017).

43. Finzi, E. The face of emotion: How botox affects our moods and relationships. (St. Martin’s Press,

2013).

44. Wagenmakers, E. J. et al. Registered replication report: Strack, Martin, & Stepper (1988).

Perspect. Psychol. Sci. 11, 917–928 (2016).

45. Coles, N. A., Larsen, J. T. & Lench, H. C. A meta-analysis of the facial feedback hypothesis

literature. (2017).

46. Strack, F., Martin, L. L. & Stepper, S. Inhibiting and facilitating conditions of the human smile: A

nonobtrusive test of the facial feedback hypothesis. J. Pers. Soc. Psychol. 54, 768–777 (1988).

47. Buck, R. Nonverbal behavior and the theory of emotion: The facial feedback hypothesis. J. Pers.

Soc. Psychol. 38, 811–824 (1980).

48. Zuckerman, M., Klorman, R., Larrance, D. T. & Spiegel, N. H. Facial, autonomic, and subjective

22

components of emotion: the facial feedback hypothesis versus externalizer-internalizer distinction.

J. Pers. Soc. Psychol. 41, 929–944 (1981).

49. Ekman, P. & Oster, H. Facial Expressions of Emotion. Annu. Rev. Psychol. 30, 527–554 (1979).

50. Laird, J. D. Self-attribution of emotion: The effects of expressive behavior on the quality of

emotional experience. J. Pers. Soc. Psychol. 29, 475–486 (1974).

51. Schimmack, U. & Chen, Y. The Power of the Pen Paradigm: A Replicability Analysis. (2017).

52. Strack, F. Reflection on the smiling registered replication report. Perspect. Psychol. Sci. 11, 929–

930 (2016).

53. Noah, T., Schul, Y. & Mayo, R. When both the original study and its failed replication are correct:

Feeling observed eliminates the facial-feedback effect. J. Pers. Soc. Psychol. 114, 657–664

(2018).

54. Hager, J. C. & Ekman, P. Methodological problems in Tourangeau and Ellsworth’s study of facial

expression and experience of emotion. Journal of Personality and Social Psychology 40, 358–362

(1981).

55. Matsumoto, D. The role of facial response in the experience of emotion: more methodological

problems and a meta-analysis. J. Pers. Soc. Psychol. 52, 769–74 (1987).

56. Levenson, R. W., Heider, K., Ekman, P., Friesen, W. V & Levenson, E. W. Emotion and

Autonomic Nervous System Activity in the Minangkabau of West Sumatra. I J. Personal. Soc.

Psychol. 62, 972–988 (1992).

57. Levenson, R. W., Ekman, P. & Friesen, W. V. Voluntary facial action generates emotion-specific

autonomic nervous system activity. Psychophysiology 27, 363–384 (1990).

58. Ekman, P. Facial Expression and Emotion. Am. Psychol. 48, 376–379 (1993).

59. Soussignan, R. Duchenne smile, emotional experience, and autonomic reactivity: A test of the

23

facial feedback hypothesis. Emotion 2, 52–74 (2002).

60. Ekman, P., Davidson, R. J. & Friesen, W. V. The duchenne smile: Emotional expression and brain

physiology II. J. Pers. Soc. Psychol. 58, 342–353 (1990).

61. Carter, E. C., Schönbrodt, F. D., Hilgard, J. & Gervais, W. M. Correcting for bias in psychology: A

comparison of meta-analytic methods. (2017).

62. Macaskill, P., Walter, S. D. & Irwig, L. A comparison of methods to detect publication bias in meta-

analysis. Stat. Med. 20, 641–654 (2001).

63. Stanley, T. D. Limitations of PET-PEESE and other meta-analysis methods. Soc. Psychol.

Personal. Sci. 8, 581–591 (2017).

64. John, L. K., Loewenstein, G. & Prelec, D. Measuring the prevalence of questionable research

practices with incentives for truth telling. Psychol. Sci. 23, 524–532 (2012).

65. Eysenck, H. J. An exercise in mega-silliness. Am. Psychol. 33, 517 (1978).

66. Tourangeau, R. & Ellsworth, P. C. The role of facial response in the experience of emotion. J.

Pers. Soc. Psychol. 37, 1519–1531 (1979).

67. Dimberg, U. & Söderkvist, S. The Voluntary Facial Action Technique: A Method to Test the Facial

Feedback Hypothesis. J. Nonverbal Behav. 35, 17–33 (2011).

68. Gellhorn, E. Motion and emotion: The role of proprioception in the physiology and pathology of the

emotions. Psychol. Rev. 71, 457–472 (1964).

69. Gellhorn, E. The physiological basis of neuromuscular relaxation. AMA Arch. Intern. Med. 102,

392–399 (1958).

70. Reisenzein, R. & Studtmann, M. On the expression and experience of surprise: No evidence for

facial feedback, but evidence for a reverse self-inference effect. Emotion 7, 612–627 (2007).

71. Harmon-Jones, C., Bastian, B. & Harmon-Jones, E. The discrete emotions questionnaire: A new

24

tool for measuring state self-reported emotions. PLoS One 11, e0159915 (2016).

72. Ekman, P. Darwin, Deception, and Facial Expression. in Annals of the New York Academy of

Sciences 1000, 205–221 (2003).

73. Marsh, A. A., Rhoads, S. A. & Ryan, R. M. A multi-semester classroom demonstration yields

evidence in support of the facial feedback effect. Emotion (2018). doi:10.1037/emo0000532

74. Lucey, P. et al. The extended cohn-kanade dataset (ck+): A complete dataset for action unit and

emotion-specified expression. in Computer Vision and Pattern Recognition Workshops (CVPRW),

2010 IEEE Computer Society Conference on 94–101 (IEEE, 2010).

75. Ekman, P. & Rosenberg, E. L. What the face reveals: Basic and applied studies of spontaneous

expression using the Facial Action Coding System (FACS). (Oxford University Press, USA, 1997).

76. March, D. S., Gaertner, L. & Olson, M. A. In harm’s way: On preferential response to threatening

stimuli. Personal. Soc. Psychol. Bull. 43, 1519–1529 (2017).

77. Lang, P. & Bradley, M. M. The International Affective Picture System (IAPS) in the study of

emotion and attention. Handb. Emot. elicitation Assess. 29, (2007).

78. Lewinski, P., den Uyl, T. M. & Butler, C. Automated facial coding: Validation of basic emotions and

FACS AUs in FaceReader. J. Neurosci. Psychol. Econ. 7, 227 (2014).

79. Barr, D. J., Levy, R., Scheepers, C. & Tily, H. J. Random effects structure for confirmatory

hypothesis testing: Keep it maximal. J. Mem. Lang. 68, 255–278 (2013).

25

Reprints and permissions: Reprints and permissions information is available at

www.nature.com/reprints

Competing interests. The authors declare no competing interests.

Correspondence: Correspondence and requests for materials should be addressed to

colesn@vols.utk.edu.

Author contributions. NC, DM, and FM proposed and coordinated the project, helped design the project

and analysis plan, drafted the Stage 1 submission, and will carry out data collection. PE, LG, and FS

provided feedback on the project. ML, MM, and LG helped design the analysis plan. All other authors

helped design the project, provided feedback on the Stage 1 submission, and will carry out data

collection.

Acknowledgments. Nicholas A Coles is supported by the National Science Foundation Graduate

Research Fellowship #R010138018. Natalia Trujillo is supported by Colciencias Grant

111577757638. Yuki Yamada is supported by JSPS KAKENHI (15H05709, 16H01866, 17H00875

18H04199, and 18K12015).