Facial Action Research Articles

Higher social tolerance is associated with more complex facial behavior in macaques.

The social complexity hypothesis for communicative complexity posits that animal societies with more complex social systems require more complex communication systems. We tested the social complexity hypothesis on three macaque species that vary in their degree of social tolerance and complexity. We coded facial behavior in >3000 social interactions across three social contexts (aggressive, submissive, affiliative) in 389 animals, using the Facial Action Coding System for macaques (MaqFACS). We quantified communicative complexity using three measures of uncertainty: entropy, specificity, and prediction error. We found that the relative entropy of facial behavior was higher for the more tolerant crested macaques as compared to the less tolerant Barbary and rhesus macaques across all social contexts, indicating that crested macaques more frequently use a higher diversity of facial behavior. The context specificity of facial behavior was higher in rhesus as compared to Barbary and crested macaques, demonstrating that Barbary and crested macaques used facial behavior more flexibly across different social contexts. Finally, a random forest classifier predicted social context from facial behavior with highest accuracy for rhesus and lowest for crested, indicating there is higher uncertainty and complexity in the facial behavior of crested macaques. Overall, our results support the social complexity hypothesis.

Domain-Incremental Continual Learning for Mitigating Bias in Facial Expression and Action Unit Recognition

As Facial Expression Recognition (FER) systems become integrated into our daily lives, these systems need to prioritise making <i>fair</i> decisions instead of only aiming at higher individual accuracy scores. From surveillance systems, to monitoring the mental and emotional health of individuals, these systems need to balance the <i>accuracy versus fairness</i> trade-off to make decisions that do not unjustly discriminate against specific under-represented demographic groups. Identifying <i>bias</i> as a critical problem in facial analysis systems, different methods have been proposed that aim to mitigate bias both at data and algorithmic levels. In this work, we propose the novel use of Continual Learning (CL), in particular, using Domain-Incremental Learning (Domain-IL) settings, as a potent bias mitigation method to enhance the <i>fairness</i> of Facial Expression Recognition (FER) systems. We compare different non-Continual Learning (CL)-based and CL-based methods for their <i>performance</i> and <i>fairness scores</i> on expression recognition and Action Unit (AU) detection tasks using two popular benchmarks, the RAF-DB and BP4D datasets, respectively. Our experimental results show that CL-based methods, on average, outperform other popular bias mitigation techniques on both <i>accuracy</i> and <i>fairness</i> metrics.

Toward Robust Facial Action Units’ Detection

Facial action unit (AU) detection plays an important role in performing facial behavioral analysis of raw video inputs. Overall, there are three key factors that contribute toward the optimal performance of AU detectors: 1) being able to capture local AU-centered features; 2) exploiting the fact that some AUs co-occur with others; and 3) utilizing appearance changes across frames. We briefly review current techniques addressing each factor and discuss the challenges they meet. Given that very few works consider how to effectively and efficiently merge them all into a single framework that can be trained in an end-to-end manner, we propose (), a simple yet strong baseline for landmark-based AU detection. implements the abovementioned key factors by: 1) using the intermediate layers of a pretrained face alignment model to act as our AU features’ space; 2) optimized to satisfy a correlation constraint, derived from the AU labels, and 3) temporal constraint, derived from variations in the contents of consecutive frames in the input videos. The proposed model, with its three key components, remains simple in nature and aligns with the primary AU detection task. Experiments on several benchmarks show that it substantially improves the AU detector’s accuracy and achieves new state-of-the-art AU detection results on popular benchmarks: BP4D and DISFA. Code is available at https://github.com/jingyang2017/AU-Net.

Caregiver Ratings of Toddler Pain: The Role of Caregiver Psychological Predictors.

Young children's limited ability to self-report pain necessitates an understanding of the factors that influence pain ratings. The current paper examines the relative prediction of caregiver psychological factors and toddler pain behaviors on caregiver pain ratings post-vaccination. One hundred fifty-six parent-toddler dyads were video recorded during pediatric vaccinations. Child pain behaviors were coded before, during, and after the needle using the Face, Legs, Activity, Cry, Consolability Scale and the Neonatal Facial Coding System). Caregivers rated their child's pain after the needle, reported pre- and post-needle worry during the visit, and completed rating scales assessing other areas of psychological functioning within 2 weeks after the appointment. Regression models were estimated to examine the relative contribution of child and caregiver factors to the prediction of caregiver pain ratings. The regression model predicting caregiver pain ratings from the toddlers' pain-related distress (facial activity immediately after the needle, overall pain-related behavior immediately after, 1-min and 2-min post-needle) and caregiver worry were significant (adjusted R-square = 0.21), with caregiver pre- and post-needle worry being the only significant predictors of caregiver pain ratings. This study outlines that although child distress behavior remains a significant influence on pain ratings during toddlerhood, when caregiver worry (pre- and post-needle) was entered into the model, they were the only significant predictors of caregiver pain ratings.

Distinct physiological responses to social-evaluative stress in patients with major depressive disorder reporting a history of peer victimization

Peer victimization is a risk factor for the development of major depressive disorders, but little is known about the mechanisms. This study examined whether peer victimization alters physiological and affective responses to potentially threatening social stimuli. For this purpose, reactions to socially evaluative stimuli of depressive patients and healthy controls with varying histories of peer victimization were compared. In a social conditioning task, we studied heart rate responses to unconditioned socially negative and neutral evaluative video statements, followed by the heart rate reactions to conditioned stimuli, i.e. still images of the faces of the same actors. Diagnosis of depression and peer victimization were both associated with a more pronounced heart rate deceleration in response to unconditioned stimuli, irrespective of valence. The effect of peer victimization was stronger in depressive patients than in healthy controls. However, heart rate responses to the CSs were not related to depression or peer victimization. The results indicate a hypervigilant processing of social stimuli in depressive patients reporting histories of peer victimization. This distinct processing may be associated with inappropriate behavioral and emotional responses to social challenges, putting individuals at risk for depressive symptoms.

Analyzing facial action units in children to differentiate genuine and fake pain during inferior alveolar nerve block: a cross-sectional study

This study aimed to investigate the association between facial action units and pain levels in Syrian children, focusing on both genuine and fake pain expressions. A total of 300 Syrian children aged 6–9 years participated in the study. Pain levels were assessed using the validated Face, Legs, Activity, Cry, Consolability scale, and facial expressions were analyzed using the Facial Action Coding System. The children were asked to mimic their feelings after receiving a dental injection to elicit fake pain expressions. Statistical analysis, including multinomial logistic regression and chi-square tests, was conducted to determine the Action Units (AUs) associated with each pain level and to compare the differences between real and fake pain expressions. The results revealed significant associations between specific AUs and pain levels. For real pain expressions, the most activated AUs across different pain levels with positive coefficient values of correlation (P-value < 0.01) were analyzed. In contrast, for fake pain expressions, AU12 and AU38 were consistently observed to be the most activated. These findings suggest that certain AUs are uniquely associated with fake pain expressions, distinct from those observed in real pain expressions. Furthermore, there were no significant differences between boys and girls in terms of their genuine and fake pain expressions, indicating a similar pattern of AU activation (P-value > 0.05). It was concluded that AUs 4, 6, 41, and 46 were associated with mild pain, and AUs 4, 6, 41, 46, and 11 were associated with moderate pain cases. In severe pain, AUs 4, 6, 7, 9, 11, and 43 were associated. In fake pain feelings, AU43, AU38, and AU12 were the most activated with no difference between boys and girls.

AI-ASSISTED emotion analysis during complementary feeding in infants aged 6–11 months

This study aims to explore AI-assisted emotion assessment in infants aged 6–11 months during complementary feeding using OpenFace to analyze the Actions Units (AUs) within the Facial Action Coding system. When infants (n = 98) were exposed to a diverse range of food groups; meat, cow-milk, vegetable, grain, and dessert products, favorite, and disliked food, then video recordings were analyzed for emotional responses to these food groups, including surprise, sadness, happiness, fear, anger, and disgust. Time-averaged filtering was performed for the intensity of AUs. Facial expression to different food groups were compared with neutral states by Wilcoxon Singed test. The majority of the food groups did not significantly differ from the neutral emotional state. Infants exhibited high disgust responses to meat and anger reactions to yogurt compared to neutral. Emotional responses also varied between breastfed and non-breastfed infants. Breastfed infants showed heightened negative emotions, including fear, anger, and disgust, when exposed to certain food groups while non-breastfed infants displayed lower surprise and sadness reactions to their favorite foods and desserts. Further longitudinal research is needed to gain a comprehensive understanding of infants' emotional experiences and their associations with feeding behaviors and food acceptance.

Modulation of facial muscle responses by another person’s presence and affiliative touch during affective image viewing

ABSTRACT Stimulating CT-afferents by forearm caresses produces the subjective experience of pleasantness in the receiver and modulates subjective evaluations of viewed affective images. Receiving touch from another person includes the social element of another person’s presence, which has been found to influence affective image evaluations without involving touch. The current study investigated whether these modulations translate to facial muscle responses associated with positive and negative affect across touch-involving and mere presence conditions. Female participants (N = 40, M(age) = 22.4, SD = 5.3) watched affective images (neutral, positive, negative) while facial electromyography was recorded (sites: zygomaticus, corrugator). Results from ANOVAs showed that providing touch to another person or oneself modulated zygomaticus site responses when viewing positive images. Providing CT-afferent stimulating touch (i.e., forearm caresses) to another person or oneself dampened the positive affective facial muscle response to positive affective images. Providing touch to another person generally increased corrugator facial muscle activity related to negative affect. Receiving touch did not modulate affective facial muscle responses during the viewing of affective images but may have effects on later cognitive processes. Together, previously reported social and touch modulations of subjective evaluations of affective images do not translate to facial muscle responses during affective image viewing, which were differentially modulated.

A facial expression and expert recommendation fuzzy decision-making approach for sustainable business investments within the metaverse world

Commercial transactions in the metaverse have become more popular in recent years. Sustainable investments are a prerequisite to metaverse survival and growth. There are various factors that have an impact on the performance of sustainable business investments, such as financial analysis and organizational effectiveness. However, each of these improvements leads to increased investment and costs. These new investments may create a heavy financial burden on businesses. In this context, there is a great need to conduct a priority analysis of the criteria affecting these investments in the metaverse. This study uses a balanced scorecard to analyze the alternatives of sustainable business investments in the metaverse. For this purpose, a hybrid quantum spherical fuzzy decision-making approach is proposed by using the facial action coding system and collaborative filtering to measure the weights of the balanced scorecard perspectives with the extended decision-making trial and evaluation laboratory (DEMATEL) and to rank the alternatives of sustainable business investments within the metaverse with the extended technique for order preference by similarity to ideal solution (TOPSIS). The analysis perspectives are finance, customers, internal process, learning, and growth. The alternatives are determined as real estate, commerce, tourism, education, and community development for sustainable virtual investment decisions in the metaverse. The novelty of this study is to construct a new decision-making approach based on facial expressions, expert recommendation systems, and the quantum spherical fuzzy sets, as well as a new concept of sustainable business investments within the metaverse. It has been determined that the organizational effectiveness of the enterprises is very important in realizing sustainability goals and increasing sustainability performance. Similarly, tourism is the most suitable sustainable virtual investment area in the Metaverse universe. Using resources to ensure organizational effectiveness helps businesses reduce energy, water, and raw material consumption. This situation contributes to achieving sustainability targets as it allows the consumption of fewer natural resources.

Entrepreneurs beyond neoliberalism: Municipally owned corporations and climate change mitigation in German cities

Urban studies have shown that the affordances, constraints and forms of urban climate change mitigation often emerge from ‘entrepreneurial’ modes of governance that have developed in post-Fordist cities. However, comparative research stresses that concepts focused on the growing power of private capital under neoliberalisation are inadequate to comprehend developments in German cities. In this article, we argue that municipally owned corporations occupy critical positions in climate change mitigation governance. While municipal ownership of utilities in principle increases local governance capacities, municipally owned corporations’ roles are shaped by fiscal relations and asymmetric organisational capacities between ‘agents’ and ‘principals’. In the case of Cologne, we show that the city has failed to leverage ownership over its public energy utility to decarbonise energy provision. Managers were able to assert corporate interests in the face of fragmented political actors and entrenched fiscal crisis. In this context of political incapacity and fragmentation, environmental social movements become important actors for translating widely shared decarbonisation objectives into concrete political demands towards municipally owned corporations.

Artificial intelligence-generated virtual influencer: Examining the effects of emotional display on user engagement

Focusing on the application of artificial intelligence, this study investigates the impact of emotional display on user engagement with computer-generated imagery influencers through the lens of the computers are social actors (CASA) framework. It breaks down emotions into individual muscle movements (i.e., facial action units). By using facial recognition based on 1,028 pictures shared by Lil Miquela, the findings disclose the significance of happiness, sadness, disgust, and surprise in triggering user engagement when promoting diverse products with visually captivating content. The findings highlight the importance of balancing the intensity of muscle movement to streamline the interplay between technology, human behaviour, and digital communication.

Detecting depression based on facial cues elicited by emotional stimuli in video

Recently, depression research has received considerable attention and there is an urgent need for objective and validated methods to detect depression. Depression detection based on facial expressions may be a promising adjunct to depression detection due to its non-contact nature. Stimulated facial expressions may contain more information that is useful in detecting depression than natural facial expressions. To explore facial cues in healthy controls and depressed patients in response to different emotional stimuli, facial expressions of 62 subjects were collected while watching video stimuli, and a local face reorganization method for depression detection is proposed. The method extracts the local phase pattern features, facial action unit (AU) features and head motion features of a local face reconstructed according to facial proportions, and then fed into the classifier for classification. The classification accuracy was 76.25%, with a recall of 80.44% and a specificity of 83.21%. The results demonstrated that the negative video stimuli in the single-attribute stimulus analysis were more effective in eliciting changes in facial expressions in both healthy controls and depressed patients. Fusion of facial features under both neutral and negative stimuli was found to be useful in discriminating between healthy controls and depressed individuals. The Pearson correlation coefficient (PCC) showed that changes in the emotional stimulus paradigm were more strongly correlated with changes in subjects’ facial AU when exposed to negative stimuli compared to stimuli of other attributes. These results demonstrate the feasibility of our proposed method and provide a framework for future work in assisting diagnosis.

Facing emotional politicians: Do emotional displays of politicians evoke mimicry and emotional contagion?

Emotional displays of politicians can be persuasive. According to prominent psychological theories, we can easily "catch" the emotional displays of others through mimicry and emotional contagion. Do these processes work for politicians too, or is it conditional on what voters think of the politician making the display? In a preregistered within-subjects laboratory experiment, participants observed images of neutral and manipulated emotional displays of politicians. We measured emotional mimicry (facial electromyography) and emotional contagion (self-reports). We do not find evidence for the matched motor hypothesis. Our findings are in line with the emotional mimicry in social context model. Namely, we find that the happy displays of in-party politicians elicit congruent facial activity (a positive facial index). Furthermore, the displays of the out-party politicians do not elicit mimicry, but instead our findings suggest a reactive response: Participants smiled in response to angry out-party politicians. The self-reported emotions indicated a small effect of emotional contagion. Taken together, our study provides insights in how voters are emotionally affected by politicians' emotional displays and highlights that our polarized prior beliefs color our emotional responses to politics. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

How Brow Rotation Affects Emotional Expression Utilizing Artificial Intelligence.

It is well known that brow position affects emotional expression. However, there is little literature on how and to what degree this change in emotional expression happens. Previous studies on this topic have utilized manual rating; this method of study remains small and labor intensive. Our objective is to correlate manual brow rotations with emotional outcomes using artificial intelligence to objectively determine how specific brow manipulations affected human expression. We included 53 brow-lift patients in this study. Pre-operative patients' brows were rotated to - 20, - 10, +10, and +20 degrees in respect to the central axis of their existing brow using PIXLR, a cloud-based set of image editing tools and utilities. These images were analyzed using FaceReader, a validated software package that uses computer vision technology for facial expression recognition. The primary facial emotion and intensity of facial action units (0 = no action unit detected to 4 = most intense action unit detected) generated by the software were recorded. 265 total images [5 images (pre-operative, - 20 degree brow rotation, - 10, +10, and +20) per patient] were analyzed using FaceReader. The primary emotion detected in the majority of images was neutral. The percentage of disgust in patients' expressions, as detected by FaceReader, increased with increased positive brow rotation (1.76% disgust detected at - 20 degrees, 2.09% at - 10 degrees, 2.65% at neutral, 2.61% at +10 degrees, and 2.95% at +20 degrees). In contrast, the percentage of sadness in patients' expressions decreased with increased positive brow rotation (29.92% sadness detected at - 20 degrees, 21.5% at - 10 degrees, 11.42% at neutral, 15.75% at +10 degrees, and 12.86% at +20 degrees). Our facial action unit analysis corresponded with primary emotion analysis. The intensity of the inner brow raiser decreased with increased positive brow rotation 8.54% at - 20 degrees, 4.21% at - 10 degrees, 1.48% at neutral, 0.84% at +10 degrees, and 0.76% at +20 degrees). The intensity of the outer brow raiser increased with increased positive brow rotation (0.97% at - 20 degrees, 0.45% at - 10 degrees, 1.12% at neutral, 5.45% at +10 degrees, and 11.19% at +20 degrees). We demonstrated that increasing the degree of brow rotation correlated positively with the percentage of disgust and inversely with the percentage of sadness detected by FaceReader. This study demonstrated how different manipulated brow positions affected emotional outcomes using artificial intelligence. Physicians can use these findings to better understand how brow-lifts can affect the perceived emotion of their patients. This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .

BCI-Walls: A robust methodology to predict if conscious EEG changes can be detected in the presence of artefacts.

Brain computer interfaces (BCI) depend on reliable realtime detection of conscious EEG changes for example to control a video game. However, scalp recordings are contaminated with non-stationary noise, such as facial muscle activity and eye movements. This interferes with the detection process making it potentially unreliable or even impossible. We have developed a new methodology which provides a hard and measurable criterion if conscious EEG changes can be detected in the presence of non-stationary noise by requiring the signal-to-noise ratio of a scalp recording to be greater than the SNR-wall which in turn is based on the highest and lowest noise variances of the recording. As an instructional example, we have recorded signals from the central electrode Cz during eight different activities causing non-stationary noise such as playing a video game or reading out loud. The results show that facial muscle activity and eye-movements have a strong impact on the detectability of EEG and that minimising both eye-movement artefacts and muscle noise is essential to be able to detect conscious EEG changes.

Facial Regulation During Dyadic Interaction: Interpersonal Effects on Cooperation

This study investigated interpersonal effects of regulating naturalistic facial signals on cooperation during an iterative Prisoner’s Dilemma (IPD) game. Fifty pairs of participants played ten IPD rounds across a video link then reported on their own and their partner’s expressed emotion and facial regulation in a video-cued recall (VCR) procedure. iMotions software allowed us to auto-code actors’ and partners’ facial activity following the outcome of each round. We used two-level mixed effects logistic regression to assess over-time actor and partner effects of auto-coded facial activity, self-reported facial regulation, and perceptions of the partner’s facial regulation on the actor’s subsequent cooperation. Actors were significantly less likely to cooperate when their partners had defected on the previous round. None of the lagged scores based on auto-coded facial activity were significant predictors of cooperation. However, VCR variables representing partner’s positive regulation of expressions and actor’s perception of partner’s positive regulation both significantly increased the probability of subsequent actor cooperation after controlling for prior defection. These results offer preliminary evidence about interpersonal effects of facial regulation in interactive contexts and illustrate how dynamic dyadic emotional processes can be systematically investigated in controlled settings.

Vicarious facilitation of facial responses to pain.

Observing facial expressions of pain has been shown to lead to increased subjective, neural and autonomic pain responses. Surprisingly, these vicarious facilitation effects on its corresponding response channel, namely facial responses to pain have mostly been neglected. We aim to examine whether the prior exposure to facial expressions of pain leads to a facilitation of facial responses to experimental pain; and whether this facilitation is linked to the valence (pain vs. neutral expression) or also linked to specific motor-features of the facial pain expressions (different facial muscle movements). Subjective (intensity and unpleasantness ratings) and facial responses (Facial Action Coding System) of 64 participants (34 female) to painful and non-painful heat stimuli were assessed. Before each heat stimulus, video clips of computer-generated facial expressions (three different pain expressions and a neutral expression) were presented. The prior exposure to facial expressions of pain led to increased subjective and facial responses to pain. Further, vicarious pain facilitation of facial responses was significantly correlated with facilitation of unpleasantness ratings. We also found evidence that this vicarious facilitation of facial responses was not only linked to the presentation of pain versus neutral expressions but also to specific motor-features of the pain cue (increase in congruent facial muscle movements). Vicarious pain facilitation was found for subjective and facial responses to pain. The results are discussed with reference to the motivational priming hypothesis as well as with reference to motor priming. Our study uncovers evidence that facial pain responses are not only influenced by motivational priming (similar to other types of pain responses), but also by motor-priming. These findings shed light on the complexity - ranging from social, affective and motor mechanisms - underling vicarious facilitation of pain.

Facial Expressions Track Depressive Symptoms in Old Age.

Facial expressions play a crucial role in the diagnosis of mental illnesses characterized by mood changes. The Facial Action Coding System (FACS) is a comprehensive framework that systematically categorizes and captures even subtle changes in facial appearance, enabling the examination of emotional expressions. In this study, we investigated the association between facial expressions and depressive symptoms in a sample of 59 older adults without cognitive impairment. Utilizing the FACS and the Korean version of the Beck Depression Inventory-II, we analyzed both "posed" and "spontaneous" facial expressions across six basic emotions: happiness, sadness, fear, anger, surprise, and disgust. Through principal component analysis, we summarized 17 action units across these emotion conditions. Subsequently, multiple regression analyses were performed to identify specific facial expression features that explain depressive symptoms. Our findings revealed several distinct features of posed and spontaneous facial expressions. Specifically, among older adults with higher depressive symptoms, a posed face exhibited a downward and inward pull at the corner of the mouth, indicative of sadness. In contrast, a spontaneous face displayed raised and narrowed inner brows, which was associated with more severe depressive symptoms in older adults. These findings suggest that facial expressions can provide valuable insights into assessing depressive symptoms in older adults.

Toddlers’ affective responses to sociomoral scenes: Insights from physiological measures

A growing literature suggests that preverbal infants are sensitive to sociomoral scenes and prefer prosocial agents over antisocial agents. It remains unclear, however, whether and how emotional processes are implicated in infants’ responses to prosocial/antisocial actions. Although a recent study found that infants and toddlers showed more positive facial expressions after viewing helping (vs. hindering) events, these findings were based on naïve coder ratings of facial activity; furthermore, effect sizes were small. The current studies examined 18- and 24-month-old toddlers’ real-time reactivity to helping and hindering interactions using three physiological measures of emotion-related processes. At 18 months, activity in facial musculature involved in smiling/frowning was explored via facial electromyography (EMG). At 24 months, stress (sweat) was explored via electrodermal activity (EDA). At both ages, arousal was explored via pupillometry. Behaviorally, infants showed no preferences for the helper over the hinderer across age groups. EMG analyses revealed that 18-month-olds showed higher corrugator activity (more frowning) during hindering (vs. helping) actions, followed by lower corrugator activity (less frowning) after hindering (vs. helping) actions finished. These findings suggest that antisocial actions elicited negativity, perhaps followed by brief disengagement. EDA analyses revealed no significant event-related differences. Pupillometry analyses revealed that both 18- and 24-month-olds’ pupils were smaller after viewing hindering (vs. helping), replicating recent evidence with 5-month-olds and suggesting that toddlers also show less arousal following hindering than following helping. Together, these results provide new evidence with respect to whether and how arousal/affective processes are involved when infants process sociomoral scenarios.

Emotion-specific AUs for micro-expression recognition

The Facial Action Coding System (FACS) comprehensively describes facial expressions with facial action units (AUs). It is a well-used technique by researchers in emotions research to understand human emotions better. Most micro-expression datasets provide FACS-coded AU ground truths corresponding to micro-expressions classes. It is commonly accepted in computer vision-based emotions research that certain emotions are reliably revealed when specific combinations of AUs occur. However, the reliability of the ground truth AUs in the micro-expression datasets is lower than that of normal expressions, as they have lower AU intensities. Moreover, these micro-expression datasets only report the overall reliability of all AUs. It could not be identified which AUs had been accurately coded. This work aims to revisit the ground truth AUs of popular micro-expression datasets, namely CASME II, SAMM and CAS(ME)2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$^2$$\\end{document}, and inspect whether any AUs crucial for micro-expression recognition may need to be reconsidered. This paper also provides a detailed AU analysis which yields new AU-based RoIs for each dataset. These new RoIs improve the micro-expression recognition performances compared to the baselines considered in this work. The proposed RoIs for CASME II, SAMM and CAS(ME)2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$^2$$\\end{document} improve the recognition rates by 2%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$2\\%$$\\end{document}, 1%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$1\\%$$\\end{document} and 4%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$4\\%$$\\end{document}, respectively, when compared with the existing RoIs.