Receiving human and android facial mimicry induces empathic experiences and oxytocin release

ABSTRACT The ability to generate facial expressions is essential for humanoid social robots to engage in natural, human‐like interactions. This capability significantly enhances the fluidity of human‐robot communication and the precision of emotional expression. However, current methods rely heavily on pre‐programmed behavioral patterns, which are manually implemented at considerable cost in both time and human labor. To enable humanoid robots to autonomously acquire generalized expressive capabilities, they must learn human‐like expressions via self‐supervised training. To address this challenge, we present a highly biomimetic robotic face equipped with physically actuated electronic facial units, alongside an end‐to‐end learning framework that integrates Kolmogorov‐Arnold Networks (KAN) with attention mechanisms. In contrast to previous approaches, we have also developed an automated data collection system guided by expert‐designed facial motion primitives, enabling the construction of a high‐quality dataset. Notably, to the best of our knowledge, this constitutes the first facial expression dataset specifically designed for humanoid social robots. Extensive evaluations demonstrate that our method enables accurate and diverse facial mimicry across a range of test subjects.

Artificial facial mimicry (AFM) is increasingly used to enhance social interaction with virtual agents in immersive virtual reality. However, its psychological and ethical implications remain insufficiently explored. This article conceptualizes AFM as an effective and embodied intervention, examining the role of emotional congruence, individual differences, and clinical vulnerability in shaping user responses. We further outline methodological directions involving physiological measures and embodied coordination. By framing AFM within affective computing and embodied cognition, this work contributes to the responsible design of emotionally adaptive virtual agents.

SEX in bonobos: The intensity of sexual stimulation sharply drops after facial mimicry

Bodily illusion research has demonstrated that altered bodily-self-perception in eating disorders (EDs) may be linked to abnormalities in the integration of sensory bodily signals. Experiencing bodily illusions can also temporarily reduce body image disturbance. Whether similar alterations in multisensory integration processes extend to self-face representation and whether face-based multisensory illusions can reduce face or body image disturbance remains unclear. This study investigated whether susceptibility to the enfacement illusion differs based on ED risk status and whether experiencing the illusion reduces face and body image disturbance. The sample included 226 women classified as high (n = 102) or low (n = 124) ED risk, who underwent an enfacement illusion induction procedure involving synchronous (ilillusion-inducing) versus asynchronous (control) visuo-motor stimulation (via facial mimicry) between their own face and an unfamiliar person's face. Illusion strength was assessed subjectively (via self-report) and objectively (via a self-face recognition task), alongside pre- and post-illusion face and body image outcomes. Synchronous interpersonal visuo-motor stimulation led to modest changes in self-face recognition (i.e., the other person's face came to be perceived as more similar to one's own); however, these changes were not modulated by ED risk status (high versus low). Cognitive-affective responses to the illusion diverged in unexpected ways. Low ED-risk participants reported reduced body dissatisfaction and dysmorphic concern following synchronous interpersonal visuo-motor stimulation, whilst high ED-risk participants reported increased head and body dissatisfaction following both synchronous and asynchronous stimulation. These findings suggest that the multisensory processes underlying self-face representation, and ultimately supporting self-recognition and the integrity of self-other boundaries, may not be disrupted in individuals with elevated ED symptomatology. This observation may challenge the notion of a globally disrupted sense of bodily self in EDs, at least with respect to self-face processing. Instead, current results suggest that ED-related body image disturbance may reflect altered higher-order evaluative or affective processing of self-related social information rather than a fundamental deficit in multisensory integration.

Recent advances in the understanding of facial anatomy have contributed significantly to the refinement of injection techniques for the treatment of dynamic forehead lines. A comprehensive assessment of eyebrow shape, position, and the aging process is essential, as the latter are closely linked to the functional balance between the frontalis muscle and the upper facial depressors. Optimal outcomes also depend on the accurate determination of dosage per injection point, injection depth, and strategic distribution of injection sites within the frontalis, which should be carefully considered and tailored to the individual’s anatomical characteristics and therapeutic goals—whether the aim is neuromodulation for muscle activity reduction or intradermal application for skin quality enhancement. A round table discussion session among three experienced international dermatology experts in aesthetic botulinum toxin type A was performed during a MERZ LATAM-sponsored medical education session. Recent insights in facial anatomy, including the precise location and distribution of motor endplates, as well as the direction of muscular force vectors during contraction; aging processes; and interindividual variability in facial musculature and mimicry patterns are discussed, and the results are described herein. These factors play a critical role in customizing personalized injection strategies and improving aesthetic outcomes in the treatment of forehead lines.

The human capacity for facial mimicry plays a vital role in social bonding and emotional comprehension. This study investigates the role of facial mimicry in preference formation. To do so, One-hundred and six participants, comprising of fifty-three dyads engaged in a preference-based paradigm where in each trial participants had to listen and then choose between two movie synopses, while their facial muscle activity was recorded using wearable facial electromyography (EMG) electrodes. In the first part of the experiment, one participant read aloud the two synopses, and in the second part both participants listened to a pre-recorded actress. Our findings revealed that speaker-listener facial mimicry of muscles associated with positive emotions was linked with increased likelihood of selecting that synopsis. In addition, listener's choice was better predicted by speaker-listener's facial mimicry than by individuals' facial expressions alone. Notably, these results were maintained even when only smiling mimicry was measured with the pre-recorded actress, highlighting the robustness of this effect. Our findings reveal the central role facial mimicry plays in preference formation during social interaction.

To introduce the KANET-connectome matrix (KANET-Con) as a conceptual framework linking fetal behaviors observed on four-dimensional (4D) ultrasound to underlying neural network maturation, and to evaluate optimal gestational timing for functional neurodevelopmental screening. A narrative review was conducted using a PRISMA-guided literature identification and screening process. PubMed, Scopus, and Web of Science were searched (January 2000-March 2025) for studies addressing fetal connectomics, fetal neurobehavior, KANET scoring, and developmental neuroimaging. Forty-two peer-reviewed studies met inclusion criteria. Observed fetal behaviors-including facial mimicry, eye blinking, limb movement, and overall gestalt coordination-were aligned with their most plausible neural substrates using evidence from developmental neuroscience and imaging. Findings demonstrated clear temporal associations between specific fetal behaviors and neural circuit development. Eye blinking and facial expressions (24-26 weeks) correlated with brainstem-cortical integration; hand-to-face gestures (26-30 weeks) reflected emerging interhemispheric pathways; and complex limb coordination (28-32 weeks) was linked to corticospinal and basal ganglia maturation. Collectively, these data indicate that 24-32 weeks of gestation represents an optimal window for KANET-based neurobehavioral screening. Additionally, emerging artificial intelligence applications show potential to enhance scoring objectivity by detecting subtle movement features such as behavioral entropy, asymmetry, and latency. KANET, interpreted through a fetal connectomic lens, provides a functional window into early neural integration. The KANET-Con offers a clinically relevant, globally accessible conceptual model to support early detection of neurodevelopmental deviations and inform risk stratification, particularly in resource-limited settings.

Facial mimicry plays a crucial role in human communication. Previous studies have reported that observing human facial expressions induced spontaneous facial mimicry using facial electromyography (EMG) and video data. A previous study reported that the expressions of an android also elicited such mimicry, as reflected in EMG recordings. However, whether such android-induced mimicry could be visually detectable remains unknown. To investigate this issue, we showed dynamic facial expressions of anger and happiness from the android Nikola via a prompter system to participants. We recorded the EMG of their corrugator supercilii and zygomatic major and videotaped their faces, which were automatically coded for facial actions to evaluate brow lowering and lip-corner pulling actions. Nikola’s angry expressions induced stronger corrugator supercilii activity in the EMG and brow lowering in the video data, while happy expressions induced stronger zygomatic major activity and lip corner pulling. These results imply that the facial expressions of humanlike androids can induce visually detectable spontaneous facial mimicry in humans, which could be further utilized to coordinate human–robot interactions.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-25394-6.

The role of the body in altered facial emotion perception in autism and social anxiety.

Numerous studies have shown that the processes underlying trait judgments can be influenced by concurrent affect processing. The present project explores the role of emotional mimicry in trait attribution. Across three experiments, we asked participants to assess social characteristics of faces expressing happiness, sadness, and anger. In Experiments 1 and 3, we used facial electromyography to predict participants' inferences about trustworthiness, confidence, and attractiveness (Experiment 1) or their behaviorally assessed trust by asking participants to share virtual points in a "trust/investment game" (Experiment 3). In Experiment 2, we tested the causal relationship between facial activity and trait judgments. Participants were asked to assess trustworthiness while performing facial movements that either enhanced or inhibited muscle activity during mimicry of given emotional expressions. The results indicate that mimicry of happiness not only predicts but is causally linked to perceptions of trustworthiness-the stronger the imitation, the more positive the assessments. The results of Experiments 1 and 3 show that increased sadness mimicry is associated with lower trust ratings, although the results of Experiment 2 do not support a causal relationship. Additionally, we confirmed previous observations that people are more likely to mimic affiliative displays (i.e., happiness and sadness) than antagonistic ones (i.e., anger), with happiness being the most likely to be mimicked. In summary, these studies provide evidence that facial mimicry modulates social trait inferences and underscores the functional role of mimicry in social interactions. (PsycInfo Database Record (c) 2026 APA, all rights reserved).

Evaluating indicators of intentionality, flexibility, and goal-association in domestic cat (Felis silvestris catus) facial signals.

Theories of embodied cognition suggest that after an initial visual processing stage, emotional faces elicit spontaneous facial mimicry and that the accompanying change in proprioceptive facial feedback contributes to facial emotion recognition. However, this temporal sequence has not yet been properly tested, given the lack of methods allowing to manipulate or interfere with facial muscle activity at specific time points. The current study (N = 52, 28 women) investigated this key question using EEG and facial neuromuscular electrical stimulation (fNMES)—a technique offering superior control over which facial muscles are activated and when. Participants categorised neutral, happy and sad avatar faces as either happy or sad and received fNMES (except in the control condition) to bilateral zygomaticus major muscles during early visual processing (−250 to +250 ms of face onset), or later visual processing, when mimicry typically arises (500–1000 ms after face onset). Both early and late fNMES resulted in a happiness bias specific to neutral faces, which was mediated by a reduced N170 in the early window. In contrast, a modulation of the beta-band (13–22 Hz) coherence between somatomotor and occipital cortices was found in the late fNMES, although this did not predict categorisation choice. We propose that facial feedback biases emotion recognition at different visual processing stages by reducing visual processing load.

Mutual gaze and facial mimicry as drivers of shared engagement in macaque play fighting

Background/Objectives: Individuals with eating disorders (EDs) experience stronger body illusions than control participants, suggesting that abnormalities in multisensory integration may underlie distorted body perception in these conditions. These illusions can also temporarily reduce body image disturbance. Given the centrality of the face to identity and social functioning-and emerging evidence of face image disturbance in EDs-this study examined, for the first time, whether individuals with EDs exhibit heightened susceptibility to a facial illusion (the enfacement illusion) and whether experiencing this illusion improves face and/or body image. Methods: White Australian female participants (19 with an ED and 24 controls) completed synchronous and asynchronous facial mimicry tasks to induce the enfacement illusion. Susceptibility was assessed via self-report and an objective self-face recognition task, alongside pre- and post-task measures of perceived facial attractiveness, facial adiposity estimation, and head/body dissatisfaction. Results: The illusion was successfully induced across both groups. Contrary to predictions, ED and control participants demonstrated comparable susceptibility, and neither group experienced improvements in face or body image. Notably, participants with EDs experienced increased head dissatisfaction following the illusion. Conclusions: These findings indicate that the multisensory integration processes underlying self-face perception, unlike those underlying body perception, may remain intact in EDs. Participant reflections suggested that the limited therapeutic benefit of the enfacement illusion for EDs may reflect the influence of maladaptive social-evaluative processing biases inadvertently triggered during the illusion. A novel dual-process model is proposed in which distorted self-face perception in EDs may arise from biased social-cognitive processing rather than sensory dysfunction alone.

ABSTRACT Facial mimicry is known to play an important role in emotional communication. While major models of facial mimicry assume a relationship between facial mimicry and emotion recognition, the empirical evidence for such a relationship is mixed. In this study, the facial movements of the same participants were compared when they recognised the same uncertain faces as “happiness” or “anger.” When faces expressed anger in the upper half (upper anger) and happiness in the lower half (lower happiness), participants responded that it was “happiness” in some trials and “anger” in others. The electromyography data showed that participants frowned more in trials in which they recognised anger in the upper anger and lower happiness faces than in trials in which they recognised happiness in the same faces (Experiment 1). By presenting the upper and lower half of expressions individually, we confirmed that upper anger and lower happiness were both salient in the expression intensity and that both the upper and lower half of the faces individually induced emotionally congruent facial movements (Experiment 2). Despite limitations, this study supported the relationship between facial mimicry and emotional recognition. These findings aid in further understanding social interaction at an unconscious and physical level.

Empathic stress, the spontaneous reproduction of psychosocial stress by mere observation, has been shown to occur between strangers, romantic partners and in mother-child dyads. However, the mechanisms by which stress is transmitted have yet to be understood. We investigated whether facial mimicry modulates the transmission of psychosocial stress. Adolescents (13-16 years old) observed their mothers or fathers (N = 77) undergo a standardized laboratory stressor. Parents’ and adolescents’ faces were videotaped during the stress task and dyads simultaneously provided multiple samples of subjective stress, heart rate, heart rate variability (HRV), and salivary cortisol. The degree to which adolescents mimicked their parents’ facial expressions was calculated in a multi-step procedure based on windowed-cross-lagged-regressions. To integrate the correlational structure of mimicry across different facial action units (AU), an exploratory factor analysis was employed. The solution revealed a two-factor model, constructed of a positive latent factor subsuming mimicked action units associated with the act of smiling and a negative latent factor, subsuming mimicked action units used for various negative emotions. None of the stress markers were significantly associated with the extracted latent factors indexing mimicry between parents and adolescents, providing no statistically significant evidence for an association between facial mimicry and stress-transmission in the parent-adolescent dyad. Bayes Factors generally indicated moderate evidence for a lack of association with the positive and anecdotal evidence for a lack of association with negative latent mimicry factors. In conclusion, our approach to video-based mimicry calculation showed promising results in that mimicry of positive and negative emotions could be detected, albeit no evidence for a link to actual empathic stress transmission in the laboratory was found.

The growing availability of affordable Virtual Reality (VR) hardware and the increasing interest in the Metaverse are driving the expansion of Social VR (SVR) platforms. These platforms allow users to embody avatars in immersive social virtual environments, enabling real-time interactions using consumer devices. Beyond merely replicating real-life social dynamics, SVR platforms offer opportunities to surpass real-world constraints by augmenting these interactions. One example of such augmentation is Artificial Facial Mimicry (AFM), which holds significant potential to enhance social experiences. Mimicry, the unconscious imitation of verbal and non-verbal behaviors, has been shown to positively affect human-agent interactions, yet its role in avatar-mediated human-to-human communication remains under-explored. AFM presents various possibilities, such as amplifying emotional expressions, or substituting one emotion for another to better align with the context. Furthermore, AFM can address the limitations of current facial tracking technologies in fully capturing users' emotions. To investigate the potential benefits of AFM in SVR, an automated AM system was developed. This system provides AFM, along with other kinds of head mimicry (nodding and eye contact), and it is compatible with consumer VR devices equipped with facial tracking. This system was deployed within a test-bench immersive SVR application. A between-dyads user study was conducted to assess the potential benefits of AFM for interpersonal communication while maintaining avatar behavioral naturalness, comparing the experiences of pairs of participants communicating with AFM enabled against a baseline condition. Subjective measures revealed that AFM improved interpersonal closeness, aspects of social attraction, interpersonal trust, social presence, and naturalness compared to the baseline condition. These findings demonstrate AFM's positive impact on key aspects of social interaction and highlight its potential applications across various SVR domains.

Play Face (PF) and Full Play Face (FPF) in the great apes-homologous to human smile and laugh-face-have been considered a single phenomenon. However, if natural selection has preserved two expressions, probably their adaptive value differs. We collected video data on play interactions in two lowland gorilla groups (N = 21; Gorilla gorilla gorilla) housed at La Vallée des Singes and the ZooParc de Beauval (France). Lacking a tool tailored for gorillas during this study, we analyzed facial action-unit activation via chimpFACS and OpenFace. We found that PF and FPF activated partly different action units as it occurs for chimpanzees and humans' PF/FPF. We detected the rapid replication (Rapid Facial Mimicry [RFM]) of either PF or FPF that was associated with longer play sessions. Not-mimicked PF was linked to increased play session variability (different types of play patterns) measured via the Shannon Index, whereas not-mimicked FPF was associated with increased play asymmetry (imbalance between offensive/defensive patterns) measured via the Play Asymmetry Index. Lowland gorillas may use PF to manage sessions that are more complex in terms of pattern types and FPF-a more salient signal-to prevent misunderstandings when the session is imbalanced. RFM of both expressions may favor the prolongation of play sessions by increasing player synchronization and possibly emotional sharing. Our study opens the door to further comparative studies on playful expressions in humans and other primates as a way to fine-tune possible emotional communication and delineate potential evolutionary roots of Hominidae facial communication.

Social play requires communicative abilities that start developing since the first phases of life. Due to their paedomorphic nature, bonobos (Pan paniscus) tend to maintain a playful attitude also in adulthood. Here, we present the case of an adult bonobo (Congo) that has never had any contact with conspecifics in the first 40years of life. In 2023, Congo underwent an introduction protocol at the Wilhelma Zoo. Prior to the data collection period, he had spent 8months within the group. The aim was evaluating Congo's playful social competence and his ability to react in a resilient way to others' stimuli. The play sessions involving Congo and two juveniles (CPSs) were compared with play sessions occurring in other adult-juvenile pairs (OPSs). Compared to OPSs, CPSs were shorter and more punctuated by grooming pauses that were totally absent during OPSs. Despite increased grooming, CPSs had a higher risk of escalation into overt aggression from Congo frequently screaming during the interaction. Unlike other adults, Congo never rapidly replicated the partners' playful facial displays, a phenomenon known as rapid facial mimicry. Despite Congo appearing to struggle to manage social play, he responded to other's invitations, thus indicating a presence of play motivation. Albeit being just a case report, our findings emphasize the importance of adult play as a mechanism for social integration, even in individuals entirely unfamiliar with its nuances. Although animals engage in play early in life, it is never too late to play the game and strengthen social bonds.