Detection of Genuine and Posed Facial Expressions of Emotion: Databases and Methods

Can deepfakes be used to study emotion perception? A comparison of dynamic face stimuli

Video recordings accurately capture facial expression movements; however, they are difficult for face perception researchers to standardise and manipulate. For this reason, dynamic morphs of photographs are often used, despite their lack of naturalistic facial motion. This study aimed to investigate how humans perceive emotions from faces using real videos and two different approaches to artificially generating dynamic expressions - dynamic morphs, and AI-synthesised deepfakes. Our participants perceived dynamic morphed expressions as less intense when compared with videos (all emotions) and deepfakes (fearful, happy, sad). Videos and deepfakes were perceived similarly. Additionally, they perceived morphed happiness and sadness, but not morphed anger or fear, as less genuine than other formats. Our findings support previous research indicating that social responses to morphed emotions are not representative of those to video recordings. The findings also suggest that deepfakes may offer a more suitable standardized stimulus type compared to morphs. Additionally, qualitative data were collected from participants and analysed using ChatGPT, a large language model. ChatGPT successfully identified themes in the data consistent with those identified by an independent human researcher. According to this analysis, our participants perceived dynamic morphs as less natural compared with videos and deepfakes. That participants perceived deepfakes and videos similarly suggests that deepfakes effectively replicate natural facial movements, making them a promising alternative for face perception research. The study contributes to the growing body of research exploring the usefulness of generative artificial intelligence for advancing the study of human perception.

The Emotion Authenticity Recognition (EAR) test: normative data of an innovative test using dynamic emotional stimuli to evaluate the ability to recognize the authenticity of emotions expressed by faces

Despite research has massively focused on how emotions conveyed by faces are perceived, the perception of emotions' authenticity is a topic that has been surprisingly overlooked. Here, we present the Emotion Authenticity Recognition (EAR) test, a test specifically developed using dynamic stimuli depicting authentic and posed emotions to evaluate the ability of individuals to correctly identify an emotion (emotion recognition index, ER Index) and classify its authenticity (authenticity recognition index (EA Index). The EAR test has been validated on 522 healthy participants and normative values are provided. Correlations with demographic characteristics, empathy and general cognitive status have been obtained revealing that both indices are negatively correlated with age, and positively with education, cognitive status and different facets of empathy. The EAR test offers a new ecological test to assess the ability to detect emotion authenticity that allow to explore the eventual social cognitive deficit even in patients otherwise cognitively intact.

Automated facial recognition system using deep learning for pain assessment in adults with cerebral palsy

Objective

Assessing pain in individuals with neurological conditions like cerebral palsy is challenging due to limited self-reporting and expression abilities. Current methods lack sensitivity and specificity, underlining the need for a reliable evaluation protocol. An automated facial recognition system could revolutionize pain assessment for such patients.The research focuses on two primary goals: developing a dataset of facial pain expressions for individuals with cerebral palsy and creating a deep learning-based automated system for pain assessment tailored to this group.

Methods

The study trained ten neural networks using three pain image databases and a newly curated CP-PAIN Dataset of 109 images from cerebral palsy patients, classified by experts using the Facial Action Coding System.

Results

The InceptionV3 model demonstrated promising results, achieving 62.67% accuracy and a 61.12% F1 score on the CP-PAIN dataset. Explainable AI techniques confirmed the consistency of crucial features for pain identification across models.

Conclusion

The study underscores the potential of deep learning in developing reliable pain detection systems using facial recognition for individuals with communication impairments due to neurological conditions. A more extensive and diverse dataset could further enhance the models' sensitivity to subtle pain expressions in cerebral palsy patients and possibly extend to other complex neurological disorders. This research marks a significant step toward more empathetic and accurate pain management for vulnerable populations.

Should I trust you? Investigating trustworthiness judgements of painful facial expressions

Past research indicates that patients' reports of pain are often met with skepticism and that observers tend to underestimate patients' pain. The mechanisms behind these biases are not yet fully understood. One relevant domain of inquiry is the interaction between the emotional valence of a stranger's expression and the onlooker's trustworthiness judgment. The emotion overgeneralization hypothesis posits that when facial cues of valence are clear, individuals displaying negative expressions (e.g., disgust) are perceived as less trustworthy than those showing positive facial expressions (e.g., happiness). Accordingly, we hypothesized that facial expressions of pain (like disgust) would be judged more untrustworthy than facial expressions of happiness. In two separate studies, we measured trustworthiness judgments of four different facial expressions (i.e., neutral, happiness, pain, and disgust), displayed by both computer-generated and real faces, via both explicit self-reported ratings (Study 1) and implicit motor trajectories in a trustworthiness categorization task (Study 2). Ratings and categorization findings partly support our hypotheses. Our results reveal for the first time that when judging strangers' facial expressions, both negative expressions were perceived as more untrustworthy than happy expressions. They also indicate that facial expressions of pain are perceived as untrustworthy as disgust expressions, at least for computer-generated faces. These findings are relevant to the clinical setting because they highlight how overgeneralization of emotional facial expressions may subtend an early perceptual bias exerted by the patient's emotional facial cues onto the clinician's cognitive appraisal process.

Holistic processing and visual characteristics of regulated and spontaneous expressions

The rapid and efficient recognition of facial expressions is crucial for adaptive behaviors, and holistic processing is one of the critical processing methods to achieve this adaptation. Therefore, this study integrated the effects and attentional characteristics of the authenticity of facial expressions on holistic processing. The results show that both regulated and spontaneous expressions were processed holistically. However, the spontaneous expression details did not indicate typical holistic processing, with the congruency effect observed equally for aligned and misaligned conditions. No significant difference between the two expressions was observed in terms of reaction times and eye movement characteristics (i.e., total fixation duration, fixation counts, and first fixation duration). These findings suggest that holistic processing strategies differ between the two expressions. Nevertheless, the difference was not reflected in attentional engagement.

What is missing in the study of emotion expression?

While approaching celebrations for the 150 years of "The Expression of the Emotions in Man and Animals", scientists' conclusions on emotion expression are still debated. Emotion expression has been traditionally anchored to prototypical and mutually exclusive facial expressions (e.g., anger, disgust, fear, happiness, sadness, and surprise). However, people express emotions in nuanced patterns and - crucially - not everything is in the face. In recent decades considerable work has critiqued this classical view, calling for a more fluid and flexible approach that considers how humans dynamically perform genuine expressions with their bodies in context. A growing body of evidence suggests that each emotional display is a complex, multi-component, motoric event. The human face is never static, but continuously acts and reacts to internal and environmental stimuli, with the coordinated action of muscles throughout the body. Moreover, two anatomically and functionally different neural pathways sub-serve voluntary and involuntary expressions. An interesting implication is that we have distinct and independent pathways for genuine and posed facial expressions, and different combinations may occur across the vertical facial axis. Investigating the time course of these facial blends, which can be controlled consciously only in part, is recently providing a useful operational test for comparing the different predictions of various models on the lateralization of emotions. This concise review will identify shortcomings and new challenges regarding the study of emotion expressions at face, body, and contextual levels, eventually resulting in a theoretical and methodological shift in the study of emotions. We contend that the most feasible solution to address the complex world of emotion expression is defining a completely new and more complete approach to emotional investigation. This approach can potentially lead us to the roots of emotional display, and to the individual mechanisms underlying their expression (i.e., individual emotional signatures).

Genetic algorithms reveal profound individual differences in emotion recognition

Emotional communication relies on a mutual understanding, between expresser and viewer, of facial configurations that broadcast specific emotions. However, we do not know whether people share a common understanding of how emotional states map onto facial expressions. This is because expressions exist in a high-dimensional space too large to explore in conventional experimental paradigms. Here, we address this by adapting genetic algorithms and combining them with photorealistic three-dimensional avatars to efficiently explore the high-dimensional expression space. A total of 336 people used these tools to generate facial expressions that represent happiness, fear, sadness, and anger. We found substantial variability in the expressions generated via our procedure, suggesting that different people associate different facial expressions to the same emotional state. We then examined whether variability in the facial expressions created could account for differences in performance on standard emotion recognition tasks by asking people to categorize different test expressions. We found that emotion categorization performance was explained by the extent to which test expressions matched the expressions generated by each individual. Our findings reveal the breadth of variability in people's representations of facial emotions, even among typical adult populations. This has profound implications for the interpretation of responses to emotional stimuli, which may reflect individual differences in the emotional category people attribute to a particular facial expression, rather than differences in the brain mechanisms that produce emotional responses.

Facial expression recognition based on active region of interest using deep learning and parallelism

The automatic facial expression tracking method has become an emergent topic during the last few decades. It is a challenging problem that impacts many fields such as virtual reality, security surveillance, driver safety, homeland security, human-computer interaction, medical applications. A remarkable cost-efficiency can be achieved by considering some areas of a face. These areas are termed Active Regions of Interest (AROIs). This work proposes a facial expression recognition framework that investigates five types of facial expressions, namely neutral, happiness, fear, surprise, and disgust. Firstly, a pose estimation method is incorporated and to go along with an approach to rotate the face to achieve a normalized pose. Secondly, the whole face-image is segmented into four classes and eight regions. Thirdly, only four AROIs are identified from the segmented regions. The four AROIs are the nose-tip, right eye, left eye, and lips respectively. Fourthly, an info-image-data-mask database is maintained for classification and it is used to store records of images. This database is the mixture of all the images that are gained after introducing a ten-fold cross-validation technique using the Convolutional Neural Network. Correlations of variances and standard deviations are computed based on identified images. To minimize the required processing time in both training and testing the data set, a parallelism technique is introduced, in which each region of the AROIs is classified individually and all of them run in parallel. Fifthly, a decision-tree-level synthesis-based framework is proposed to coordinate the results of parallel classification, which helps to improve the recognition accuracy. Finally, experimentation on both independent and synthesis databases is voted for calculating the performance of the proposed technique. By incorporating the proposed synthesis method, we gain 94.499%, 95.439%, and 98.26% accuracy with the CK+ image sets and 92.463%, 93.318%, and 94.423% with the JAFFE image sets. The overall accuracy is 95.27% in recognition. We gain 2.8% higher accuracy by introducing a decision-level synthesis method. Moreover, with the incorporation of parallelism, processing time speeds up three times faster. This accuracy proves the robustness of the proposed scheme.

Review: Posed vs. Genuine Facial Emotion Recognition and Expression in Autism and Implications for Intervention

Different styles of social interaction are one of the core characteristics of autism spectrum disorder (ASD). Social differences among individuals with ASD often include difficulty in discerning the emotions of neurotypical people based on their facial expressions. This review first covers the rich body of literature studying differences in facial emotion recognition (FER) in those with ASD, including behavioral studies and neurological findings. In particular, we highlight subtle emotion recognition and various factors related to inconsistent findings in behavioral studies of FER in ASD. Then, we discuss the dual problem of FER – namely facial emotion expression (FEE) or the production of facial expressions of emotion. Despite being less studied, social interaction involves both the ability to recognize emotions and to produce appropriate facial expressions. How others perceive facial expressions of emotion in those with ASD has remained an under-researched area. Finally, we propose a method for teaching FER [FER teaching hierarchy (FERTH)] based on recent research investigating FER in ASD, considering the use of posed vs. genuine emotions and static vs. dynamic stimuli. We also propose two possible teaching approaches: (1) a standard method of teaching progressively from simple drawings and cartoon characters to more complex audio-visual video clips of genuine human expressions of emotion with context clues or (2) teaching in a field of images that includes posed and genuine emotions to improve generalizability before progressing to more complex audio-visual stimuli. Lastly, we advocate for autism interventionists to use FER stimuli developed primarily for research purposes to facilitate the incorporation of well-controlled stimuli to teach FER and bridge the gap between intervention and research in this area.