Abstract
The relationship between facial expressions and facial shape is surprisingly understudied. We use facial videos and photographs, along with others’ judgements, to investigate the interplay among facial morphology, dynamics, and perception. Facial expressions were quantified through Action Units (AUs) using the Facial Action Coding System. Facial shape was analysed using geometric morphometrics. Analysis of the relationships between 14 AUs and 13 morphological parameters revealed that women display more intense facial movements, but men exhibit a tighter link between facial movements and facial structures. Next, we used Principal Component Analysis for a broader categorisation. The principal components (PCs) of the AU data corresponded to groups of AUs, revealing specific facial expressions. Three of the expressions thus identified were the same in both sexes, while the fourth was sex specific. The PCs of the morphometric data were more mosaic in composition; however, one male and one female PC largely agreed, reflecting a similar facial-shape phenotype. In both sexes, this facial shape dimension was associated with the tendency to produce a Duchenne smile. In turn, the expression of a Duchenne smile systematically affected how people in the videos were judged by an independent panel on various attributes (e.g., generosity, trustworthiness, happiness). In men, a male-specific PC of the morphometric data corresponded to facial masculinity, as determined by Linear Discriminant Analysis. Men with more masculine facial shapes were more likely to produce lower-face anger expression. We conclude that natural facial expressions are to a certain extent related to facial morphology, especially in men.
Similar content being viewed by others
Introduction
Since Darwin’s (1872) seminal work, human facial expressions have been extensively studied across various scientific disciplines. A vast body of knowledge has been accumulated concerning the evolutionary origin of facial expressions and their functions, the genetics behind facial expressions and their ontogenetic development, the underlying neurophysiology and anatomical mechanisms, individual and sex differences in facial expressions, and the interplay between universality and cultural specificity (Van Hooff, 1962, 1967; Andrew, 1963; Rinn, 1984; Schmidt & Cohn, 2001; Ekman, 2006; Waller & Micheletta, 2013; Fridlund, 2014; Fridlund et al., 2014; Dols & Russell, 2017; Wilkins, 2017; Ross et al., 2019; Kavanagh et al., 2022; Waller et al., 2024: Butovskaya & Rostovtseva, 2025).
The relationship between facial displays and facial morphology is of considerable interest from both proximate and ultimate evolutionary perspectives (Ramsey & Aaby, 2022). Facial displays are produced through a complex system of facial muscles, which is well studied both in humans (Ekman et al., 2002; Schmidt & Cohn, 2001; Norton & Netter, 2012; Moore et al., 2014; Standring, 2016; Netter, 2019; Sinelnikov et al., 2024), and other animals, including non-human primates (Burrows et al., 2006, 2011; Vick et al., 2007; Waller et al., 2008, 2012; Diogo et al., 2009; Wilkins, 2017; Kavanagh et al., 2022). At the same time, the development of bones and muscles is a mutually adaptive process (Matic et al., 2007; Krenn & Buehler, 2019; Sinelnikov et al., 2024). Patterns of facial growth are associated with facial muscle activity and facial expressions (Alabdullah et al., 2015; Grover et al., 2015). The same muscles, stretched across different morphologies, may produce variation in expressiveness, as seen in human and primate facial displays (Preuschoft, 2000; Schmidt & Cohn, 2001). It is therefore conceivable that specific facial expressiveness correlates with broader morphological phenotypes, such as particular facial shape patterns in humans. From an evolutionary perspective, this implies that the adaptive value of facial expressions may vary with an individual’s facial morphology, which, to a certain extent, has also been shaped by selective evolutionary processes (Butovskaya & Rostovtseva, 2025).
Yet, the relationship between facial expressions and facial morphology is surprisingly understudied. Researchers exploring the anatomical component of facial expressions, particularly the variation, phylogeny and functions of mimetic muscles (Schmidt & Cohn, 2001), acknowledge the important role of facial hard- and soft-tissue structures in the production and perception of expressions. Yet, direct investigations into the connection between facial morphology and facial expressiveness are limited. The topic is somewhat addressed in medical studies (neurophysiology of movement deficit, orthodontics; Weeden et al., 2001; Grover et al., 2015), and some scholars investigate the perceptual effects produced by the overlap between facial morphological types and emotional facial expressions using computer-generated stimuli (Hess et al., 2009; Gill et al., 2014; Jack & Schyns, 2015). Quite some studies focus on the resemblance between the geometry of some static neutral facial features and certain emotional facial expressions (Todorov et al., 2008; Adams et al., 2015; Eggleston et al., 2022; Rostovtseva et al., 2024), or how morphological facial features may bias perception of emotional facial expressions (Deska et al., 2018). However, there is hardly any study investigating associations between facial shape and natural facial displays regarding the proneness of individuals with different morphological phenotypes to produce specific facial expressions.
Here, we address this knowledge gap by exploring the relationship between dynamic facial expressions and facial morphology in 142 subjects of European origin (71 males and 71 females). To this end, we utilise facial photographs and videos collected during an earlier experiment (Rostovtseva et al. 2023a). Moreover, judgments on these photos and videos by an independent panel regarding certain attributes of the person shown (such as attractiveness, generosity, or trustworthiness) were available. Based on these data, we address four research questions: (a) To what extent are facial muscle movements associated with morphological parameters of the face or specific facial areas? (b) Can specific facial expressions be distilled from the facial muscle movements, and to what extent are these expressions associated with principal characteristics of facial shape? (c) To what extent are these associations sex-specific? (d) How are the faces of individuals who are prone to produce certain facial expressions perceived by independent observers of the same sex?
In our study, we quantify facial expressions in terms of Action Units (AUs), either single or correlated in groups, using the Facial Action Coding System (FACS) (Ekman et al., 2002). To this end, we use an automated version of FACS – FaceReader – which analyses facial expressions through artificial neural networks (Lewinski et al., 2014; Loijens & Krips, 2025; Nomiya et al., 2025). This technology enabled us to record even the subtlest facial movements that are otherwise difficult to detect. FACS is an anatomically grounded system of detection and classification of facial movements based on underlying muscle action. The method has the advantage of describing the facial movements without tying them to specific emotions or other potential sources of action. Hence, there is no need to ascribe a meaning to facial movements. The interpretation of detected facial expressions can be further developed in any conceptual framework. The FaceReader analysis was based on facial video recordings of the subjects, and more complex facial expressions were then quantified as consistent combinations of single AUs.
Static facial morphology was traced using two different approaches. Discrete morphological features of the face – craniofacial, mandible, and soft-tissue parameters – were approximated from the frontal anthropological photographs of the subjects. The overall facial shape was captured by applying geometric morphometrics (Bookstein, 1997; Zelditch et al., 2012) to the same set of facial photographs. Next, we investigate the relationship between discrete morphological parameters of the face and the activation of single AUs. This part of the analysis provides insight into which morphological areas of the face may affect elementary facial movements. Finally, we explore the relationship between the overall facial shapes of individuals and their propensity to display certain facial expressions. This allows us to assess the presence and nature of more complex associations between facial morphology – i.e., its overall shape – and the features of facial expressions. Such connections might arise from sexual and social selection (Crespi et al., 2022).
Materials and Methods
Participants
The facial expression recording and photography phase of the study involved 142 individuals (71 men, 71 women), aged 18 to 29 years (mean age 22 ± 2.5 years). All participants were students of European descent (Dutch, Belgian, and German). Participants had no congenital or acquired facial deformations or dysfunctions, scars, or skin diseases. The Ethical Committee Psychology (ECP) of the University of Groningen approved the study (Research Code: 16250-O). All participants provided informed consent prior to the experiment.
Procedure
Each participant was invited to a separate room where a short facial video (20 s of neutral talk to the camera) and face photographs were taken. For the video, participants were asked to look into the camera and give a 20-second self-introduction, explaining how they spent their morning. The video was recorded in daylight using a Logitech C920 Pro HD webcam at maximum resolution (1080p) and capture rate of 30 frames per second. The camera was positioned at eye level at a standard distance to the subject (0.5 m). Participants were given no instructions regarding the content of their story. The experimenter was present in the room during recording. In addition, anthropological photography was conducted, with each participant photographed in frontal view with a neutral facial expression, sitting on a fixed chair with a straight back and a natural head position. The camera was levelled at eye height with a fixed distance of 1.7 m. Participants were asked to use a hair hoop to remove hair from their faces when needed. Each participant also completed a demographic questionnaire.
Videos as a Source of Facial Expressions
Facial expressions in our study were assessed through the above-mentioned self-presentation videos. These expressions occurred in an emotionally neutral context, but participants were informed that their videos would afterwards be watched by other individuals. Self-presentation is an important source of facial expressions (DePaulo, 1992; Zang & Yang, 2025). It may involve both involuntary signalling of individual qualities and contain intended social signals directed to the public (Biel et al., 2012). Since we used a format that included a verbal component, we need to elaborate a little more on the concept of conversational facial displays, which could also be a part of the expressive profiles in our study. Conversational facial expressions, defined as expressions produced in social interactions alongside speech (for instance, in dialogues), are well described in the literature (Chovil, 1991; Ekman, 2004; Cunningham et al., 2004). They are considered a specific type of facial display that, along with emotional signalling, includes so-called emblems, illustrators, regulators, and self-adaptors, which are highly dependent on the interactive structure of communication, for example, in a dialogue (Ekman & Friesen, 1969). It is important to note that in our study, we used a different approach. Under the conditions of social self-presentation, participants received no interactive responses to their talks, got no feedback on the content of their speech, and had no immediate interlocutor. Instead, the camera served as the receiver of their signals, delaying the communication in time and broadening the potential audience to the general public. This setup removed most elements of a typical conversation, reducing facial displays to a combination of intentional social signals and involuntary signs of emotional states.
Facial Expression Analysis
The video recordings were processed using the emotion recognition system, FaceReader 9 (Noldus Information Technology, The Netherlands), which analyses facial expressions in images/videos using artificial neural networks (Lewinski et al., 2014; Loijens & Krips, 2025; Nomiya et al., 2025) trained on large datasets based on the Facial Action Coding System (FACS) technique (Ekman et al., 2002). FACS is an anatomically grounded system that classifies facial movements into Action Units (AUs), each having its own muscular basis. Basically, the degree of activation of each AU (its evidence) is assessed by intensity scores (A-B-C-D-E scoring scale) (Ekman et al., 2002). FaceReader can output AU intensities on a continuous scale ranging from 0 (fully relaxed) to 1 (maximum expression). In our study, we focused on 15 of the 20 AUs available in FaceReader, excluding AUs related to speech articulation and eyelid closure, as they were not relevant for our purposes. The list of 15 AUs included in the study, with action descriptions, is presented in Table 1 (Supplementary Table S1 provides additional information on the facial muscles involved in each AU, as well as a list of muscle attachment sites to the skull and other tissues).
For the analysis, we used a general face model of FaceReader, which targets representatives of European populations. The quality of analysis was tracked and remained above 0.75. The 20-second videos were analysed frame by frame, yielding the intensity of each AU for each frame. With a frame rate of 30 per second, we obtained 600 intensity values per AU per individual. Subsequently, for each participant, we calculated a generalised intensity value for each AU, corresponding to the third quartile (Q3) of the frame-by-frame intensity distribution. We chose the Q3 value because it avoids underestimating the generalised intensity score (as with the mean or median) and, at the same time, eliminates high-intensity outliers that do not reflect typical background facial expressions. These Q3 intensity values were further used in the analysis.
Morphological Facial Parameters
To explore potential morphological correlates of facial expressions, we selected 13 facial parameters to represent craniofacial and mandibular features, as well as facial soft-tissue shapes (Rostovtseva et al., 2021). These parameters include the skull areas underlying specific facial muscles, the soft tissues directly involved in specific expressions, and the bones where facial muscles attach (originate) (see Supplementary Table S1): forehead height (tri-gl), eyebrows region height (gl-n), distance between eyes (en-en), interpupillary distance (pu-pu), eye height (ps-pi), nose height (n-sn), maxillary ridge height (sn-ls), bizygomatic width (zy-zy), nose width (al-al), mouth width (ch-ch), lips thickness (ls-li), mandible height (sto-gn), and bigonial width (go-go). Figure 1 illustrates the locations of the reference morphometric landmarks.
All facial parameters were calculated based on the coordinates of digitised facial landmarks. The calculations were made after Generalised Procrustes superimposition (see next subsection), which standardises all photographs (configurations of landmarks) for their position, orientation, and scale (Zeldich et al., 2012). Thus, all morphological parameters represent relative, but not absolute, facial size values. For obtaining these discrete parameters, no bilateral symmetrisation of facial configurations was applied.
Facial Shape Analysis
Geometric morphometrics (Bookstein, 1997; Windhager et al., 2011) was used to analyse facial shape. For each photograph, seventy landmarks and semi-landmarks were manually digitised using tpsDig2 version 2.17 (Rohlf, 2015). This set included 36 landmarks, representing established anthropometric points related to craniofacial and soft-tissue facial morphology, and 34 semi-landmarks, which traced facial outlines and eyebrow contours (Rostovtseva et al. 2023b).
To assess the reliability of the manual digitisation, two independent observers placed all 70 points on randomly selected photographs of 20 males and 20 females. Inter-observer agreement was quantified using geometric morphometrics as the ratio of the among-individual variance component to the sum of the among-individual and measurement-error components (Zelditch et al., 2012), using the “vegan” package in R (Oksanen et al., 2025). The resulting inter-observer agreement was 0.96, deemed sufficient to proceed with landmark coordinates obtained by a single observer. Individual facial configurations were standardised for position, orientation, and scale via Generalised Procrustes superimposition across the entire sample (N = 142). Superimposition was implemented together with sliding semi-landmarks using the minimum bending energy criterion in the “geomorph” package for R (Adams et al., 2025). Since estimating facial asymmetry was not a goal of this study, we symmetrised all facial configurations (Mitteroecker & Gunz, 2009) to minimise potential distortions caused by head positioning in the 2D images. This symmetrisation was carried out in R using both basic functions and those developed by Claude (2008). The visualisation of facial shapes was realised through (a) thin-plate spline deformation grids, generated in R using functions adapted from Claude (2008), and (b) through geometric morphometric morphs, which were created by unwarping and averaging individual photographs in tpsSuper 2.04 (Rohlf, 2015).
Sex differences in facial shape were assessed using permutational multivariate analysis of variance, implemented with the “vegan” package of R (Oksanen et al., 2025). Statistical significance was evaluated using 10,000 permutations (Good, 2000). Principal Components Analysis of the shape data was conducted using the “geomorph” package (Adams et al., 2025) in R. Individual femininity/masculinity shape scores were obtained using linear discriminant analysis for two groups (males and females), conducted on the Procrustes coordinates of the facial landmarks and semi-landmarks using the “MASS” package of R (Venables & Ripley, 2002).
Ratings of Photographs and Videos
As mentioned earlier, the photographs and videos were taken for a study on human prosocial behaviour (Rostovtseva et al. 2023a). In that study, the pictures and videos were also rated according to eight criteria: How rational, risk-taking, trustworthy, greedy, attractive, generous, angry, and happy would you rate the person shown on a five-point scale (1 = not at all, 5 = very much)? The ratings were conducted by an independent panel of 87 females and 60 males of European descent, aged 18 to 30 years. Each rater was shown a series of facial photographs and videos featuring different participants, ensuring that each person judged 5 unique videos and 5 unique photographs, with no overlap between them. Each female photograph and video was evaluated by 14–15 female raters, while 9–11 male raters assessed each male photograph and video. For the purpose of rating, all videos were muted to prevent observers from reacting to the content of the speech or the speaker’s voice characteristics.
Results
Single Action Units and Morphological Face Parameters
To start the analysis, we ensured that each Action Unit (AU) generally had a non-zero activation level and that the intensity values for each AU varied across participants. For this purpose, we plotted the distributions of Q3 intensities for each AU for men and women. These extensive sets of plots are presented in Supplementary Figure S1. The plotted data demonstrate considerable individual variation in the activation of almost all AUs. The majority of AUs were most commonly expressed at trace or slight intensities (corresponding to A and B scores, according to Ekman et al., 2002). However, a number of AUs, such as Inner Brow Raiser (AU01), Outer Brow Raiser (AU02), Cheek Raiser (AU06), Lid Tightener (AU07), and especially Lip Corner Puller (AU12), were expressed by a considerable fraction of participants with severe and even extreme intensities (corresponding to D and E scores in FACS). The least individual variation was observed in both men and women for Nose Wrinkler (AU09) (generally, the least involved unit), so it was excluded from further analysis. Lip Pressor (AU24) had low individual variation among males. Based on this understanding, we proceeded with the analysis, keeping in mind that – except for AUs associated with eyebrow movements and smiling – facial displays under the given condition were generally at slight and trace levels of intensity.
The data in Fig. 2a reveal significant sex differences in intensities for seven of the 14 AUs. Females displayed more intense activation of Lip Corner Puller (AU12), Lid Tightener (AU07), Cheek Raiser (AU06), Lip Stretcher (AU20), Upper Lip Raiser (AU10), Chin Raiser (AU17), and Lip Corner Depressor (AU15). Most of these facial movements are located in the mouth area, except for Cheek Raiser and Lid Tightener. These results indicate that females generally produced more intense movements, especially in the mouth area, and especially displayed more intense and/or frequent smiles (AU12, AU06) than males.
Significant sex differences were revealed for nine of the 13 morphological parameters (Fig. 2b). Specifically, females had relatively higher eyebrows, more rounded eyes, more elongated noses with narrower nasal wings, fuller lips, relatively wider faces at the cheekbones, relatively wider set eyes (relatively higher interpupillary distance), and shorter maxillary ridges and mandibles. These results (Fig. 2) indicate that males and females differed in both facial morphology and facial movements. Therefore, further analysis of relationships between facial structures and facial expressions was conducted separately for males and females.
Relationships Between Action Units and Morphological Parameters
In a first step, we investigated the association between the 14 Action Units and the 13 morphological facial parameters through a regression analysis. The association patterns are shown, separately for males and females, in Supplementary Figure S2. For both men and women, we found consistent relationships between the Lip Corner Puller (AU12) and mouth width, nasal wing width, and maxillary ridge height (the latter did not reach statistical significance in women). This suggests that both men and women with wider mouths, wider nasal wings, and lower maxillary ridges tended to produce more intense smile-like facial displays. However, this was the only cluster of associations that was consistent across the sexes. Other patterns differed considerably between men and women. Men had twice as many significant associations (N = 36) between AUs and morphological parameters as women (N = 18). Men also demonstrated greater interconnectedness among facial regions in their involvement in facial movements. For instance, there were clear relationships between mandible height and mouth/chin movements: men with higher mandibles showed a more intense activation of the Lip Corner Depressor (AU15), Chin Raiser (AU17), and Lip Stretcher (AU20). These links were absent in women. Men also exhibited clusters of associations that could not be explained by the spatial relations between facial muscles involved in the movement and underlying facial morphology. For example, raising the eyebrows (AU01, AU02) in men was positively linked to eye and eyebrow heights, which is intuitive; however, it was also negatively related to several lower face structures: men with lower mandibles, narrower mouths, narrower nasal wings, and thinner lips were predisposed to raise their brows more actively. Such spatially distant connections could arise from correlations between the shapes of different parts of the face.
Taking all this into consideration, we conclude that inspecting the relationship between specific morphological parameters and specific Action Units provides limited insight. Therefore, we will henceforth focus on broader determinants of facial shapes and facial expressions, as revealed by Principal Component Analyses.
Identifying Facial Shape Dimensions
To identify broader determinants of facial shapes, we conducted Principal Component Analyses (PCAs) of the facial shape data for men and women. To distinguish the outcome from the later PCAs for facial expression, the principal components of the facial shape analysis will be denoted as s-PCs. Figure 3 displays the results of the analysis for men (a, b) and women (c, d). For men, the first four s-PCs together explain 67.6% of the variance in the male facial shape data. For women, the first four s-PCs together explain 69.4% of the variance. In both sexes, the first two s-PCs account for approximately 30% and 20% of the variance, while the third and fourth s-PCs each explain around 10%. The visual assessment of the deformation grids of the male and female s-PCs (see Supplementary Figure S3) suggests that s-PC1 in females describes a similar aspect of facial shape as s-PC2 in males (except for the face outline): For both female s-PC1 and male s-PC2, a facial shape change in the positive direction (+ 1 SD visualisation) produces a shorter face, shorter mandibles, thinner lips, a more elongated nose, and more V-shaped eyebrows. While we cannot propose specific underlying biological or environmental determinants for the extracted s-PCs, we investigated whether any of the s-PCs correspond to the facial shape femininity-masculinity vectors in men and women. To explore this, we (1) analysed sex differences in facial shapes, (2) assigned a masculinity score to each individual, and (3) created visualisations of typical feminine female and masculine male facial shapes.
According to the geometric morphometrics analysis, there are significant sex differences in facial shape. Sex accounts for approximately 7% of the total facial shape variance (p < 0.001). A visualisation of sex differences in facial shape, using thin-plate spline deformation grids, is presented in Supplementary Figure S4. To quantify the degree of facial femininity or masculinity for each individual, we used linear discriminant analysis (LDA), which separates men and women based on their facial features (Fig. 4a). The individual scores generated by the LDA were then interpreted as a measure of facial masculinity, meaning that women, who generally have more feminine faces, would have negative masculinity scores corresponding to positive facial femininity.
Visual comparison of the deformation grids (Supplementary Figures S3, S4) for the s-PCs and feminine/masculine facial shapes reveals that male facial shapes with high scores on male s-PC3 and high scores on male facial masculinity share many features. Among those are a relatively low eyebrow region, a relatively short mid-face, wider-set eyes and eyebrows. Visualisations in the form of geometrically morphed portraits show that these shapes almost fully converge, being a variant of masculine male facial appearance (see the +1SD morphed male portraits in Fig. 4a and b). Therefore, a positive direction in the male s-PC3 may be interpreted as a change in facial shape associated with facial masculinisation.
Identifying Facial Expressions
To explore whether specific shape dimensions of male and female faces are associated with dynamic facial expressions, we examined which groups of correlated AUs were produced by men and women during their self-presentation to the camera. For this purpose, we performed Principal Component Analyses on the AU data of men and women. To distinguish the principal components of the facial expression analysis from those of the earlier facial shape analysis, they will be indicated as e-PCs. Figure 5 provides an interpretation of the first four e-PCs for men and women.
In men, the first four e-PCs together explain 58.4% of the variance in facial movements, whereas in women, the first four e-PCs account for 56.7% of the variance. Typically, the principal components identified by a PCA do not have a clear-cut interpretation. However, in this case, the clear association of specific AUs with only one (and no more than one) PC suggests that each e-PC represents a distinct facial expression. These expressions are characterised by the combination of AUs with high positive loadings on that e-PC. Take, for example, the first e-PC in men and women, which is, in both sexes, associated with the combination of Inner Brow Raiser (AU01), Outer Brow Raiser (AU02), and Upper Lid Raiser (AU05). This pattern corresponds to a well-known facial expression, commonly referred to as “eyebrow flash” (Eibl-Eibesfeldt, 1972; Grammer et al., 1988). Therefore, we labelled it as E-flash. This expression is frequently observed in humans and is thought to convey the intention to communicate (e.g., greetings, attracting a partner’s attention, or sexual flirting). It is reassuring that the primary e-PC in both sexes corresponds to a commonly observed facial expression, well documented in the literature.
Another well-established expression identified by the PCA is a smile, characterised by a combination of Lip Corner Puller (AU12) and Cheek Raiser (AU06). This pattern appeared as the third principal component in both men (male e-PC3) and women (female e-PC3). It corresponds to a Duchenne smile (D-smile) (Ekman et al., 1990; Frank et al., 1993), which involves not only pulling the mouth corners laterally and upward but also lifting the cheeks and wrinkling the skin around the eyes. The Duchenne smile is widely regarded as a genuine, spontaneous smile, in contrast to a “social” smile. However, there is evidence that a significant portion of people can pose it voluntarily (Krumhuber & Manstead, 2009; Gosselin et al., 2010; Gunnery et al., 2013).
Another expression shared between the sexes – represented by the second component in women (female e-PC2) and the fourth component in men (male e-PC4) – is a combination of Upper Lip Raiser (AU10), Lip Corner Depressor (AU15), and Chin Raiser (AU17), which is known to be a relatively stable combination of AUs (Ekman et al., 2002). However, this combination does not have a specific name or clear emotional interpretation in FACS. The closest description might be a display of disbelief or scepticism, but only for the lower face (the full scepticism pattern includes AU01, AU02, AU10, AU15, AU17, and AU41). Since AU41 was not included in our analysis, we tentatively labelled this principal component as LF-scepticism (lower face scepticism).
The remaining two components were sex-specific. The second component in men (male e-PC2) combined Chin Raiser (AU17), Lip Tightener (AU23), and Lip Pressor (AU24). This combination is part of the prototypical anger expression in FACS, but involves only the lower face (the full anger pattern includes AU04, AU05, and AU07, related to corresponding brow and eye movements). To facilitate communication, we labelled this component as LF-anger (lower face anger). This facial expression was not observed in females.
Finally, a female-specific expression – represented by the fourth female component (female e-PC4) – comprised Brow Lowerer (AU04), Lid Tightener (AU07), and Lip Stretcher (AU20). The FACS manual does not offer a specific interpretation for this combination, even as a partial pattern. Therefore, we assigned a label based on our own interpretation. Considering the study context (verbal self-presentation to the camera), such a combination of facial movements could potentially occur when a person is trying to recall something or facing some mental challenge. We labelled this expression as Perplexity.
Except for E-flash and D-smile, the labels assigned to the expressions in our study are arbitrary. We do not claim that any of the facial expressions are associated with the emotional loading that the names might suggest; rather the names are used solely to help distinguish between the e-PCs throughout the text.
Relationships Between Facial Expressions and Facial Shape
Figure 6 displays the results of an analysis that explores the relationships between facial expressions (represented by e-PC scores) and facial shape (represented by s-PC scores and femininity/masculinity scores). Separate analyses were conducted for men (Fig. 6a) and women (Fig. 6b). To examine these relationships, we used a series of linear regression models. For each model, one of the facial expression PCs (e-PCs) was the response variable, while one of the facial shape PCs (s-PCs), the male facial masculinity score (fMSC), or the female facial femininity score (fFMM) was entered as the sole predictor.
Our analysis reveals specific connections between facial expressions and facial shapes. The LF-scepticism expression in men is significantly negatively related to the male facial shape variation along s-PC1, and positively related to shape variation along s-PC4. The D-smile expression in men is significantly positively related to facial shape variation along s-PC2. Interestingly, high male s-PC3 scores and high male facial masculinity (fMSC) scores both had a significant positive effect on LF-anger displays in men. This is consistent with our earlier conclusion that high male s-PC3 scores and high male facial masculinity represent very similar facial shape patterns (Fig. 4). In women, the only significant relationship between facial shape and facial displays is a positive link between s-PC1 and D-smile expression. Hence, the D-smile expression is related to s-PC1 in females and s-PC2 in males. As discussed above, female s-PC1 and male s-PC2 both describe a similar aspect of facial shape: men and women scoring high on these s-PCs tend to have shorter faces, shorter mandibles, thinner lips, more elongated noses, and more V-shaped eyebrows. Hence, the D-smile expression (which indicates “smiling proneness” in the context of social self-presentation) is related to the same facial shape characteristics in men and women.
The Effect of Facial Expressions on Facial Judgements
In this section, we examine whether the combination of facial shape and related facial expressions produces a distinct perceptual effect. First, we investigated whether static neutral faces that score high (or low) on specific s-PCs induce similarly high (or low) ratings from independent observers on various social traits. To this end, we analysed judgements of neutral facial photographs of participants, provided by independent observers of the same sex as the judged individual. The regression analysis for men revealed that higher values of male s-PC1 are associated with higher attractiveness ratings of photographs (R² = 0.072, B = 4.956, p = 0.025). High male s-PC4 scores positively affected generosity (R² = 0.078, B = 7.231, p = 0.019) and happiness (R² = 0.094, B = 8.746, p = 0.010) ratings. In women, high scores for s-PC3 induced negative attractiveness ratings (R² = 0.073, B = -8.203, p = 0.048), while a high value of female s-PC4 induced high rationality (R² = 0.113, B = 5.752, p = 0.013) and high happiness (R² = 0.077, B = 7.477, p = 0.043) ratings. These results demonstrate that some aspects of the facial shape influence social perception even before any facial movement occurs.
Next, we investigated the effect of the facial expressions identified above on the judgment of the facial videos. The linear regression analysis revealed that only the D-smile expression had a significant effect on the rating of facial videos: A high value of the facial expression D-smile tends to induce a significantly higher rating regarding the attributes generosity, trustworthiness, and happiness, and a lower rating regarding anger, which was evident in both men and women. For statistical details and sex differences, see Supplementary Figure S5.
In the preceding analyses, we examined the effects of expressions on facial judgments without considering the influence of underlying facial shapes. In this final step, we turn our attention to whether combinations of specific facial shapes and expressions – those we have already identified as related – produce a specific perceptual effect on independent observers. In this analysis, we focus exclusively on female s-PC1 and male s-PC2, as they represent a common aspect of facial structure shared by both sexes and affect the D-smile expression in both men and women. In other words, this was the only shape dimension we found that was consistent across men and women with respect to both its underlying facial shape features and its influence on a specific facial expression.
We tested how female s-PC1 and male s-PC2 affected all eight judgment criteria of the facial videos (see Methods). The only significant relationship was revealed for female trustworthiness ratings, while for male judgements the effect, although noticeable, was weaker and did not reach significance (p = 0.093) (Fig. 7).
Figure 7a shows visualisations of the typical facial shapes of women who are prone to display D-smile (right side; +1 SD of the female s-PC1 shape scores), and those who are not prone to D-smile (left side; -1 SD of the female s-PC1 shape scores). A significant positive relationship between female facial shape as measured by s-PC1 and the propensity to produce D-smiles has already been described in the previous section (see Fig. 6). For convenience, Fig. 7b illustrates this relationship again, but this time as a scatterplot, showing that D-smile (female e-PC3 scores) is positively related to female s-PC1 scores. Figure 7c visualises how female s-PC1 scores affect trustworthiness ratings of the neutral facial photographs. The results indicate that variation of female facial shape along s-PC1 does not have any systematic effect on judgements of their trustworthiness based on neutral photographs. In turn, Fig. 7d illustrates how female s-PC1 scores affect trustworthiness ratings of the self-presentation videos. Here we observe a clear positive relationship: higher female s-PC1 shape scores are associated with higher trustworthiness ratings of their videos. A similar situation is observed in men. D-smile intensities increased with higher male scores on s-PC2 (Fig. 7f); however, trustworthiness ratings of the male photographs were not affected by facial shape variation along male s-PC2 (Fig. 7g). This means that, as in women, men with “D-smile prone” facial shapes do not appear more trustworthy in their neutral photographs. Similarly, higher male s-PC2 scores tended to yield higher trustworthiness ratings of the self-presentation videos (Fig. 7h). The D-smile was the most prominent facial expression among men and women scoring high on male s-PC2 and female s-PC1, respectively (Fig. 6). This suggests that D-smiles enhanced perceived trustworthiness in individuals with this particular facial shape. Importantly, although D-smile had a positive effect on the judgments of almost all “positive” traits in females, when linked to this specific facial shape, it significantly increased only trustworthiness ratings.
Discussion
The results of our study indicate that natural facial displays produced during self-presentation to the camera are associated with facial morphology, especially in men. That is, individuals with different facial morphological phenotypes are prone to produce specific facial expressions. These associations suggest that correlated morpho-dynamic compounds play a role in human facial communication.
The analysis of individual Action Units (AUs) and individual morphometric parameters showed that women displayed the AUs more pronouncedly, but these displays were less strongly linked to underlying facial morphology. Conversely, men showed a tighter association between facial movements and facial structures. A potential explanation – that less intense facial movements in men highlight more evident links with morphology – is unlikely, as morphology was generally associated with the most intense movements (e.g., smiling), not the least intense, in both sexes.
The most intriguing results emerged from the analysis of the full facial shapes of participants and their relationships with facial expressions, as quantified by Principal Component Analyses (PCA) of both facial morphology and facial displays. The principal components of facial expressions (e-PCs) all corresponded to unique combinations of Action Units, allowing us to identify them with specific facial displays. Three of these unique facial expressions were common for men and women (E-flash, D-smile, and LF-scepticism), while LF-anger and Perplexity were specific to males and females, respectively. The principal components of the full facial shapes (s-PCs) were more mosaic in their trait composition. However, one pair of facial shapes (one female [positive s-PC1] and one male [positive s-PC2]) shared many facial traits. Besides, one male facial shape [positive s-PC3] corresponded well with the male-masculine facial shape, as revealed by separate shape analysis. These results allowed us to work with relationships between shapes and distinct complex facial expressions.
Many details of our findings are already discussed in the Results section; here, we focus on the most insightful issues. As in the case of low-level associations (single AUs and morphometric parameters), males also exhibited tighter relationships between facial expressions and higher-level aspects of facial shape (associations between e-PCs and s-PCs). However, in both sexes, the Duchenne smile (D-smile) was positively associated with a specific dimension of facial shape (positive s-PC1 in women; positive s-PC2 in men). These male and female shapes converged at the general geometry of the mouth, nose, eyes, and eyebrows. This result, in our view, warrants special attention, as it revealed a common facial shape dimension associated with Duchenne smiles in both men and women. As shown in Fig. 7dh, the judgement of trustworthiness of the facial videos is positively related to this component of facial shape (female s-PC1, male s-PC2). From this, one might conclude that the corresponding aspect of the facial shape causes a favourable judgement of trustworthiness. This, however, is unlikely, as the relationship between facial shape components and trustworthiness judgements disappeared when judgements were based on facial photographs rather than videos (Fig. 7cg). Apparently, the perceptual effect is related to facial dynamics, suggesting it is not induced by shape characteristics but by aspects of facial expressions. Since this component of facial shape is positively related to the tendency to produce a Duchenne smile, and the Duchenne smile, in turn, is positively related to the judgment of trustworthiness, it is reasonable to assume that the positive effect of female s-PC1 and male s-PC2 shapes on perceived trustworthiness is indirect (mediated by the effect of the D-smile on trustworthiness). We conclude that the association between components of facial shape and components of facial expressions may make it difficult to ascribe a causal role to any of these components concerning facial perception.
However, in our data, the pattern described for trustworthiness is not representative of the other social traits. The expression of Duchenne smile had a positive effect on the judgement of almost all “positive” social traits, especially in females (positive relationships with perceived generosity, trustworthiness, attractiveness, happiness, risk-taking, a negative relationship with perceived anger; see Suppl. Fig. S5), which is in line with the existing literature (Surakka & Hietanen, 1998; Gunnery & Ruben, 2016). Some traits (generosity, happiness, anger) had an even tighter relationship with D-smile than trustworthiness (as quantified by the R2 values; see Suppl. Fig. S5). Yet, in contrast to trustworthiness, none of these judged traits revealed any significant or pronounced associations with the “prone to smile” facial shape dimension. Of course, this lack of association could have arisen from the stochastic nature of the data. Alternatively, it could indicate a perceptual effect arising from the interplay between this specific facial shape and the Duchenne smile appearing on it. This would mean that particular combinations of facial shape and expressions may elicit specific social responses. The fact that other judged traits, which are strongly related to Duchenne smile, did not leave any considerable trace on the facial judgements of individuals with the “prone to smile” facial shape (s-PC1/s-PC2) could be explained by differential individual contributions to the linear relationships between D-smile and judged traits. In other words, the linear relations presented in Suppl. Fig. S5 for such perceived traits as generosity, attractiveness, happiness, and anger could be predominantly produced by individuals with the other “types” of facial shapes.
In men, the male-specific facial expression – lower-face anger (LF-anger) – was also related to specific facial shape characteristics. Men with more masculine facial shapes were more prone to produce LF-anger expression. This result is consistent with the literature, providing evidence that men are more likely than women to exhibit angry or threatening facial expressions (Hess et al., 2009; McDuff et al., 2017). However, we did not find perceptual reactions to LF-anger displays (in terms of facial judgements on rationality, risk-taking, trustworthiness, greediness, attractiveness, generosity, anger, and happiness). But one should remember that in our study we did not obtain facial judgements of traits such as masculinity, dominance, and aggressiveness, which could be relevant in this case.
Could our results regarding shape-expression relationships be a by-product of detection bias? For example, might FaceReader tend to record certain AU activations when faces had just a specific geometric configuration? Though possible, we consider this explanation unlikely. FaceReader analysed dynamic videos in which participants moved and talked, constantly changing their facial positions. In other words, the proportions (geometry) of the facial shape changed instantly during the videos. Yet, individual facial shapes were assessed from an independent set of photographs taken with fixed perspective and head positioning – not from the expression videos. This almost precludes any direct geometric bias in the shape-expression associations.
Despite the rather consistent relationships uncovered in our study, one should bear in mind that most of the revealed significant associations do not survive correction for multiple testing. However, the effects are largely systematic. For example, the same facial shape pattern (s-PC2 in males and s-PC1 in females) is associated with the same facial expression (D-smile) in both males and females. The masculine male shape pattern (fMSC) is visually repeated in the male s-PC3 shape obtained via a different approach to morphometric analysis. Still, both male shapes are associated with the same male facial expression (LF-anger). The same applies to facial perception analysis, which reveals similar relations in both men and women, which are independent samples of subjects (Fig. 7). Such systematic associations could hardly emerge by chance.
Our results remain open to considerable speculation, both in terms of interpretation and functional significance. Individual variation in facial shape arises from a complex interplay of numerous factors, including genetic inheritance (Claes et al., 2018; Zhang et al., 2022; Murillo-Rincón et al., 2025), developmental and environmental influences (Hujoel et al., 2017; Larson et al., 2018), and physiological processes, such as hormone levels and their changes (Schaefer et al., 2005; Law Smith et al., 2006; Marečková et al., 2011; Mitteroecker et al., 2017; Roosenboom et al., 2018). Consequently, when a specific pattern of facial expressiveness is linked to a particular facial shape, it is also linked to the factors that underlie that shape. Thus, it remains unclear whether the relationships we observe between facial shape and expression reflect direct morphological or functional associations or have a more indirect and complex origin. For example, facial expressiveness is closely linked to individual features of the neurohumoral system (Kraaijenvanger et al., 2017; Bress & Cascio, 2024). At the same time, hormones also influence the development of facial shape, which may lead to correlations between shape and expression at both the individual and the sex level. It is therefore conceivable that our finding that male facial expressions are more tightly linked to facial morphology than female expressions reflects the effects of testosterone, which affects both facial shape development and expressiveness in males.
However, it remains unknown whether our findings extend to representatives of other cultures and populations. The observed relationships between facial shape and expression – and their sex specificity – may also be shaped by social norms and cultural prescriptions and thus could vary across human societies. Cross‑cultural studies using the same methodology would help clarify the roles of universal physiological processes versus more specific environmental (including social and cultural) influences.
An alternative perspective is that the link between facial shape and the display of certain expressions, which in turn influence social perception, may stem from deliberate human behaviour that evolved as an adaptation for successful social communication. This interpretation is supported by our facial perception results (Fig. 7). According to this view, the shape-expression link may arise because not every expression fits every face. This could be due to interactions between facial morphology and muscle dynamics, which affect the intensity and form of expressions, as well as the overall geometry of the face. For example, a smile may appear more trustworthy on some faces than on others. In fact, certain facial morphologies may even cause a smile to negatively affect overall social perception of the face (e.g., Havens et al., 2010). Adults may learn through personal experience which expressions best suit their faces and in which social contexts, allowing them to use these dynamic signals more effectively to achieve social goals. This idea is supported by research in plastic surgery and orthodontics, where facial morphology is considered when evaluating the aesthetics of facial expressions (Sugahara et al., 2008; Grover et al., 2015; Hussain et al., 2016; Melo et al., 2020; Romeo, 2021). Another explanation for our finding that Duchenne smiles are associated with specific facial shapes in both men and women is that these shapes may simply make it easier to produce a Duchenne smile – due to interactions between facial morphology, muscle dynamics, and even muscle innervation. This also raises the possibility of a compensatory effect: any negative perceptions of such facial shapes could be offset by an individual’s greater tendency to smile. While our study highlights promising directions for future research, it certainly has various limitations. The most important limitation is that our data (videos, facial photographs and facial judgements) were collected in the context of a different project (Rostovtseva et al. 2023a), for different purposes. This specifically affected the part of the study involving facial judgements. The judgments were made only within the same sex, preventing us from examining inter-sex perceptual effects, which are likely to have an impact. The list of judged traits was also limited by the purposes of the initial study. As already mentioned, including judgements on characteristics such as masculinity, dominance, and aggressiveness could reveal additional effects, including those related to male LF-anger expression. Another limitation is that we conducted facial shape analysis from 2D images, whereas studying the relationship between facial morphology and dynamics would benefit from 3D analysis. It would also be interesting to extend our findings by examining shape-expression relationships under high emotional tension – conditions that typically elicit more intense facial expressions. Facial movements detection in our study was based on FaceReader analysis, which relies on visual signals rather than direct measurements. Although earlier versions of FaceReader received some criticism regarding detection accuracy and lack of independent validation studies, the later versions, FaceReader 7–9, demonstrated sufficient detection accuracy levels and high convergence with human FACS coders, with FaceReader 9 reaching the level of top reference AI facial action coding systems (Skiendziel et al., 2019; Landmann, 2023; Hsu & Sato, 2023; Nomiya et al., 2025). Still, it is sensitive to lighting conditions and less accurate than electromyography (EMG). However, the unavoidable psychological effects associated with EMG preclude the use of this method in studies of natural facial expressions. Finally, our study was conducted among students with a WEIRD (Western Educated, Industrialised, Rich, Democratic) background (Henrich et al., 2010), which limits the extrapolation of the results to other cultural groups and age cohorts. To our knowledge, this is the first study to explore the relationship between natural facial expressions and facial shape in humans. Therefore, it needs replication and retesting for the robustness of the findings. Whether similar effects will be observed in other populations, shaped by different social and environmental conditions, remains an open question for future research.
Data Availability
The data supporting our findings are available at https://doi.org/10.34894/0S4P4M
References
AdamsJr, R. B., Hess, U., & Kleck, R. E. (2015). The intersection of gender-related facial appearance and facial displays of emotion. Emotion Review, 7(1), 5–13.
Adams, D., Collyer, M., Kaliontzopoulou, A., & Baken, E. (2025). Geomorph: Software for geometric morphometric analyses. R package version 4.0.10. https://cran.r-project.org/package=geomorph
Alabdullah, M., Saltaji, H., Abou-Hamed, H., & Youssef, M. (2015). Association between facial growth pattern and facial muscle activity: a prospective cross-sectional study. International Orthodontics, 13(2), 181–194.
Andrew, R. J. (1963). Evolution of Facial Expression: Many human expressions can be traced back to reflex responses of primitive primates and insectivores. Science, 142(3595), 1034–1041.
Biel, J. I., Teijeiro-Mosquera, L., & Gatica-Perez, D. (2012). Facetube: predicting personality from facial expressions of emotion in online conversational video. In: Proceedings of the 14th ACM International Conference on Multimodal Interaction (pp. 53–56). NY: Association for Computing Machinery.
Bookstein, F. L. (1997). Morphometric Tools for Landmark Data: Geometry and Biology. Cambridge University Press.
Bress, K. S., & Cascio, C. J. (2024). Sensorimotor regulation of facial expression–an untouched frontier. Neuroscience & Biobehavioral Reviews, 162, 105684.
Burrows, A. M., Waller, B. M., Parr, L. A., & Bonar, C. J. (2006). Muscles of facial expression in the chimpanzee (Pan troglodytes): descriptive, comparative and phylogenetic contexts. Journal of Anatomy, 208(2), 153–167.
Burrows, A. M., Diogo, R., Waller, B. M., Bonar, C. J., & Liebal, K. (2011). Evolution of the muscles of facial expression in a monogamous ape: evaluating the relative influences of ecological and phylogenetic factors in hylobatids. The Anatomical Record: Advances in Integrative Anatomy and Evolutionary Biology, 294(4), 645–663.
Butovskaya, M., & Rostovtseva, V. (2025). Human face as a biosocial marker in human evolution. BioSystems, 250, 105427.
Chovil, N. (1991). Discourse-oriented facial displays in conversation. Research on Language & Social Interaction, 25(1–4), 163–194.
Claes, P., Roosenboom, J., White, J. D., Swigut, T., Sero, D., Li, J., & Weinberg, S. M. (2018). Genome-wide mapping of global-to-local genetic effects on human facial shape. Nature Genetics, 50(3), 414–423.
Claude, J. (2008). Morphometrics with R. Springer Science & Business Media.
Crespi, B. J., Flinn, M. V., & Summers, K. (2022). Runaway social selection in human evolution. Frontiers in Ecology and Evolution, 10, 894506.
Cunningham, D. W., Nusseck, M., Wallraven, C., & Bülthoff, H. H. (2004). The role of image size in the recognition of conversational facial expressions. Computer Animation and Virtual Worlds, 15(3-4), 305–310.
Darwin, C. R. (1872). The Expression of the Emotions in Man and Animals. John Murray.
DePaulo, B. M. (1992). Nonverbal behavior and self-presentation. Psychological Bulletin, 111(2), 203.
Deska, J. C., Lloyd, E. P., & Hugenberg, K. (2018). The face of fear and anger: Facial width-to-height ratio biases recognition of angry and fearful expressions. Emotion, 18(3), 453.
Diogo, R., Wood, B. A., Aziz, M. A., & Burrows, A. (2009). On the origin, homologies and evolution of primate facial muscles, with a particular focus on hominoids and a suggested unifying nomenclature for the facial muscles of the Mammalia. Journal of Anatomy, 215(3), 300–319.
Dols, J. M. F., & Russell, J. A. (Eds.). (2017). The Science of Facial Expression. Oxford University Press.
Eggleston, A., Tsantani, M., Over, H., & Cook, R. (2022). Preferential looking studies of trustworthiness detection confound structural and expressive cues to facial trustworthiness. Scientific Reports, 12(1), 17709.
Eibl-Eibesfeldt, I. (1972). Similarities and differences between cultures in expressive movements. In R. A. Hinde (Ed.), Non-verbal Communication (p. 297). Cambridge University Press.
Ekman, P. (2004). Emotional and conversational nonverbal signals. In: Language, Knowledge, and Representation: Proceedings of the Sixth International Colloquium on Cognitive Science, 39–50. Dordrecht: Springer Netherlands.
Ekman, P. (Ed.). (2006). Darwin and Facial Expression: A century of Research in Review. Academic.
Ekman, P., & Friesen, W. V. (1969). The repertoire of nonverbal behavior: Categories, origins, usage, and coding. Semiotica, 1(1), 49–98.
Ekman, P., Davidson, R. J., & Friesen, W. V. (1990). The Duchenne smile: Emotional expression and brain physiology: II. Journal of Personality and Social Psychology, 58(2), 342.
Ekman, P., Friesen, W. V., & Hager, J. C. (2002). Facial action coding system. The manual. Research Nexus division of Network Research Corporation.
Frank, M. G., Ekman, P., & Friesen, W. V. (1993). Behavioral markers and recognizability of the smile of enjoyment. Journal of Personality and Social Psychology, 64(1), 83.
Fridlund, A. J. (2014). Human Facial Expression: An Evolutionary View. Academic.
Fridlund, A. J., Ekman, P., & Oster, H. (2014). Facial expressions of emotion: Review of literature, 1970–1983. In A. W. Siegman, & S. Feldstein (Eds.), Nonverbal Behavior and Communication (pp. 143–224). NY, Psychology Press.
Gill, D., Garrod, O. G., Jack, R. E., & Schyns, P. G. (2014). Facial movements strategically camouflage involuntary social signals of face morphology. Psychological Science, 25(5), 1079–1086.
Good, P. (2000). Permutation Tests: A Practical Guide to Resampling Methods for Testing Hypotheses. Springer.
Gosselin, P., Perron, M., & Beaupré, M. (2010). The voluntary control of facial action units in adults. Emotion, 10(2), 266.
Grammer, K., Schiefenhövel, W., Schleidt, M., Lorenz, B., & Eibl-Eibesfeldt, I. (1988). Patterns on the face: the eyebrow flash in crosscultural comparison. Ethology, 77(4), 279–299.
Grover, N., Kapoor, D. N., Verma, S., & Bharadwaj, P. (2015). Smile analysis in different facial patterns and its correlation with underlying hard tissues. Progress in Orthodontics, 16, 1–13.
Gunnery, S. D., & Ruben, M. A. (2016). Perceptions of Duchenne and non-Duchenne smiles: A meta-analysis. Cognition and Emotion, 30(3), 501–515.
Gunnery, S. D., Hall, J. A., & Ruben, M. A. (2013). The deliberate Duchenne smile: Individual differences in expressive control. Journal of Nonverbal Behavior, 37, 29–41.
Havens, D. C., McNamara Jr, J. A., Sigler, L. M., & Baccetti, T. (2010). The role of the posed smile in overall facial esthetics. The Angle Orthodontist, 80(2), 322–328.
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33(2–3), 61–83.
Hess, U., Adams, R. B., Grammer, K., & Kleck, R. E. (2009). Face gender and emotion expression: Are angry women more like men? Journal of Vision, 9(12), 19.
Hsu, C. T., & Sato, W. (2023). Electromyographic validation of spontaneous facial mimicry detection using automated facial action coding. Sensors (Basel, Switzerland), 23(22), 9076.
Hujoel, P. P., Masterson, E. E., & Bollen, A. M. (2017). Lower face asymmetry as a marker for developmental instability. American Journal of Human Biology, 29(5), e23005.
Hussain, A., Louca, C., Leung, A., & Sharma, P. (2016). The influence of varying maxillary incisor shape on perceived smile aesthetics. Journal of Dentistry, 50, 12–20.
Jack, R. E., & Schyns, P. G. (2015). The human face as a dynamic tool for social communication. Current Biology, 25(14), R621–R634.
Kavanagh, E., Kimock, C., Whitehouse, J., Micheletta, J., & Waller, B. M. (2022). Revisiting Darwin’s comparisons between human and non-human primate facial signals. Evolutionary Human Sciences, 4, e27.
Kraaijenvanger, E. J., Hofman, D., & Bos, P. A. (2017). A neuroendocrine account of facial mimicry and its dynamic modulation. Neuroscience & Biobehavioral Reviews, 77, 98–112.
Krenn, B., & Buehler, C. (2019). Facial features and unethical behavior–Doped athletes show higher facial width-to-height ratios than non-doping sanctioned athletes. PLoS One, 14(10), e0224472.
Krumhuber, E. G., & Manstead, A. S. (2009). Can Duchenne smiles be feigned? New evidence on felt and false smiles. Emotion, 9(6), 807.
Landmann, E. (2023). I can see how you feel—Methodological considerations and handling of Noldus’s FaceReader software for emotion measurement. Technological Forecasting and Social Change, 197, 122889.
Larson, J. R., Manyama, M. F., Cole, J. B., Gonzalez, P. N., Percival, C. J., Liberton, D. K., & Hallgrimsson, B. (2018). Body size and allometric variation in facial shape in children. American Journal of Physical Anthropology, 165(2), 327–342.
Law Smith, M. J., Perrett, D. I., Jones, B. C., Cornwell, R. E., Moore, F. R., Feinberg, D. R., & Hillier, S. G. (2006). Facial appearance is a cue to oestrogen levels in women. Proceedings of the Royal Society B: Biological Sciences, 273(1583), 135–140.
Lewinski, P., Uyl, D., T. M., & Butler, C. (2014). Automated facial coding: validation of basic emotions and FACS AUs in FaceReader. Journal of Neuroscience Psychology and Economics, 7(4), 227.
Loijens, L., & Krips, O. (2025). FaceReader Methodology Note. Noldus Information Technology.
Marečková, K., Weinbrand, Z., Chakravarty, M. M., Lawrence, C., Aleong, R., Leonard, G., & Paus, T. (2011). Testosterone-mediated sex differences in the face shape during adolescence: subjective impressions and objective features. Hormones and Behavior, 60(5), 681–690.
Matic, D. B., Yazdani, A., Wells, R. G., Lee, T. Y., & Gan, B. S. (2007). The effects of masseter muscle paralysis on facial bone growth. Journal of Surgical Research, 139(2), 243–252.
McDuff, D., Kodra, E., Kaliouby, R. E., & LaFrance, M. (2017). A large-scale analysis of sex differences in facial expressions. PLoS One, 12(4), e0173942.
Melo, M., Ata-Ali, J., Ata-Ali, F., Bulsei, M., Grella, P., Cobo, T., & Martínez-González, J. M. (2020). Evaluation of the maxillary midline, curve of the upper lip, smile line and tooth shape: a prospective study of 140 Caucasian patients. Bmc Oral Health, 20, 1–9.
Mitteroecker, P., & Gunz, P. (2009). Advances in geometric morphometrics. Evolutionary Biology, 36, 235–247.
Mitteroecker, P., Windhager, S., Müller, G. B., & Schaefer, K. (2017). Patterns of correlation of facial shape with physiological measurements are more integrated than patterns of correlation with ratings. Scientific Reports, 7, 45340.
Moore, K. L., Dalley, A. F., & Agur, A. M. R. (2014). Clinically Oriented Anatomy (7th ed.). Lippincott Williams & Wilkins.
Murillo-Rincón, A. P., Seton, L. W., Escamilla-Vega, E., Damatac, A., Fuß, J., Fortmann-Grote, C., & Kaucká, M. (2025). Positional programs in early murine facial development and their role in human facial shape variability. Nature Communications, 16(1), 10112.
Netter, F. H. (2019). Atlas of Human Anatomy (7th ed.). Saunders.
Nomiya, H., Shimokawa, K., Namba, S., Osumi, M., & Sato, W. (2025). An artificial intelligence model for sensing affective valence and arousal from facial images. Sensors (Basel, Switzerland), 25(4), 1188.
Norton, N. S., & Netter, F. H. (2012). Netter’s Head and Neck Anatomy for Dentistry (2nd ed.). Elsevier/Saunders.
Oksanen, J., Simpson, G., Blanchet, F. (2025). vegan: Community Ecology Package. R package version 2.7-0. https://vegandevs.github.io/vegan/
Preuschoft, S. (2000). Primate faces and facial expressions (pp. 245–271). Social Research.
Ramsey, G., & Aaby, B. H. (2022). The proximate-ultimate distinction and the active role of the organism in evolution. Biology & Philosophy, 37(4), 31.
Rinn, W. E. (1984). The neuropsychology of facial expression: a review of the neurological and psychological mechanisms for producing facial expressions. Psychological Bulletin, 95(1), 52.
Rohlf, F. J. (2015). The tps series of software. Hystrix, 26, 1.
Romeo, G. (2021). Smile makeover and the oral facial harmony concept in a new era: relationship between tooth shape and face configuration. International Journal of Esthetic Dentistry, 16, 2–8.
Roosenboom, J., Indencleef, K., Lee, M. K., Hoskens, H., White, J. D., Liu, D., & Weinberg, S. M. (2018). SNPs associated with testosterone levels influence human facial morphology. Frontiers in Genetics, 9, 497.
Ross, E. D., Gupta, S. S., Adnan, A. M., Holden, T. L., Havlicek, J., & Radhakrishnan, S. (2019). Neurophysiology of spontaneous facial expressions: II. Motor control of the right and left face is partially independent in adults. Cortex; A Journal Devoted To The Study Of The Nervous System And Behavior, 111, 164–182.
Rostovtseva, V. V., Mezentseva, A. A., Windhager, S., & Butovskaya, M. L. (2021). Sexual dimorphism in facial shape of modern Buryats of southern Siberia. American Journal of Human Biology, 33, e23458.
Rostovtseva, V. V., Puurtinen, M., Méndez Salinas, E., Cox, R. F., Groothuis, A. G., Butovskaya, M. L., & Weissing, F. J. (2023a). Unravelling the many facets of human cooperation in an experimental study. Scientific Reports, 13(1), 19573.
Rostovtseva, V. V., Butovskaya, M. L., Mezentseva, A. A., & Weissing, F. J. (2023b). Effects of sex and sex-related facial traits on trust and trustworthiness: An experimental study. Frontiers in Psychology, 13, 925601.
Rostovtseva, V. V., Butovskaya, M. L., Mezentseva, A. A., Dashieva, N. B., Korotkova, A. A., Kavina, A., & Singh, M. (2024). Cross-cultural differences in perception of facial trustworthiness based on geometric morphometric morphs. Journal of Cross-Cultural Psychology, 55(2), 216–235.
Schaefer, K., Fink, B., Mitteroecker, P., Neave, N., & Bookstein, F. L. (2005). Visualizing facial shape regression upon 2nd to 4th digit ratio and testosterone. Collegium Antropologicum, 29(2), 415–419.
Schmidt, K. L., & Cohn, J. F. (2001). Human facial expressions as adaptations: Evolutionary questions in facial expression research. American Journal of Physical Anthropology: The Official Publication of the American Association of Physical Anthropologists, 116(S33), 3–24.
Sinelnikov, R. D., Sinelnikov, Y. A. R., & Sinel’nikov, A. Y. A. (2024). Atlas anatomii cheloveka: Uchebnoe posobie: V 3 T. T.1. 8-e izd. [Atlas of Human Anatomy: Study Guide: In 3 Volumes (8th ed., Vol. 1)]. Moscow: RIA Novaya Volna, Izdatel Umerenkov.
Skiendziel, T., Rösch, A. G., & Schultheiss, O. C. (2019). Assessing the convergent validity between the automated emotion recognition software Noldus FaceReader 7 and Facial Action Coding System Scoring. PLoS One, 14(10), e0223905.
Standring, S. (2016). Gray’s Anatomy (41st ed.). Elsevier Churchill Livingstone.
Sugahara, T., Yamada, N., Sadoyama, T., Kamijo, M., Iguchi, T., & Nakamura, T. (2008). Analysis of the relationship between an attractive smile and eye shape. Kansei Engineering International, 7(2), 155–161.
Surakka, V., & Hietanen, J. K. (1998). Facial and emotional reactions to Duchenne and non-Duchenne smiles. International Journal of Psychophysiology, 29(1), 23–33.
Todorov, A., Said, C. P., Engell, A. D., & Oosterhof, N. N. (2008). Understanding evaluation of faces on social dimensions. Trends in Cognitive Sciences, 12(12), 455–460.
Van Hooff J. (1962). Facial expressions in higher primates. Symposia of the Zoological Society of London, 8, 97–125.
Van Hooff J. (1967). The facial displays of Catarrhine monkeys and apes. The facial displays of the Catarrhine monkeys and apes. In D. Morris (Ed.), Primate Ethology (pp. 7–68). Chicago, IL: Aldine.
Venables, W. N., & Ripley, B. D. (2002). Modern Applied Statistics With S. (4th ed.). New York, NY: Springer.
Vick, S. J., et al. (2007). A Cross-species comparison of facial morphology and movement in humans and chimpanzees using the Facial Action Coding System (FACS). Journal of Nonverbal Behavior, 31, 1–20.
Waller, B. M., & Micheletta, J. (2013). Facial expression in nonhuman animals. Emotion Review, 5(1), 54–59.
Waller, B. M., Parr, L. A., Gothard, K. M., Burrows, A. M., & Fuglevand, A. J. (2008). Mapping the contribution of single muscles to facial movements in the Rhesus Macaque (Vol. 95, pp. 93–100). Physiology & Behavior.
Waller, B. M., Lembeck, M., Kuchenbuch, P., Burrows, A. M., & Liebal, K. (2012). GibbonFACS: a muscle-based facial movement coding system for hylobatids. International Journal of Primatology, 33, 809–821.
Waller, B. M., Kavanagh, E., Micheletta, J., Clark, P. R., & Whitehouse, J. (2024). The face is central to primate multicomponent signals. International Journal of Primatology, 45(3), 526–542.
Weeden, J. C., Trotman, C. A., & Faraway, J. J. (2001). Three dimensional analysis of facial movement in normal adults: influence of sex and facial shape. The Angle Orthodontist, 71(2), 132–140.
Wilkins, A. S. (2017). Making Faces. Cambridge. Belknap Press of Harvard University.
Windhager, S., Schaefer, K., & Fink, B. (2011). Geometric morphometrics of male facial shape in relation to physical strength and perceived attractiveness, dominance, and masculinity. American Journal of Human Biology, 23, 805–814.
Zang, X., & Yang, J. (2025). Dynamic facial emotional expressions in self-presentation predicted self-esteem. Behavioral Sciences, 15(5), 709.
Zelditch, M. L., Swiderski, D. L., & Sheets, H. (2012). Geometric Morphometrics for Biologists: A Primer. London; Waltham.
Zhang, M., Wu, S., Du, S., Qian, W., Chen, J., Qiao, L., & Wang, S. (2022). Genetic variants underlying differences in facial morphology in East Asian and European populations. Nature Genetics, 54(4), 403–411.
Acknowledgements
We want to thank our colleagues – Mikael Puurtinen, Emiliano Mendez Salinas, and Antonius Groothuis – for their help and contribution to the organisation of the experimental study in the University of Groningen. We are also grateful to Anna Mezentseva for her work in digitalising facial landmarks from the participants’ photographs.
Funding
The experiment was conducted with the support of the Dr. J.L. Dobberke Foundation, and the Erasmus Mundus Action 2 programme EMA2 Aurora II (2013–2021); data analysis and manuscript preparation were supported by the grant 24-18-00457 of the Russian Science Foundation (on behalf of VVR, and MLB).
Author information
Authors and Affiliations
Contributions
V.V.R. and M.L.B. contributed to the conception of the study. V.V.R. and F.J.W. conducted the experiment, analysed the data and wrote the main manuscript text. All authors reviewed the manuscript.
Corresponding author
Ethics declarations
Ethics Statement
The study received approval by the Ethical Committee Psychology (ECP) of the University of Groningen (Research Code: 16250-0). Before the experiment, all subjects signed informed consents.
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Below is the link to the electronic supplementary material.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Rostovtseva, V.V., Butovskaya, M.L. & Weissing, F.J. Relationships Between Facial Expressions and Facial Shape. Hum Nat (2026). https://doi.org/10.1007/s12110-026-09528-x
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1007/s12110-026-09528-x