skip to main content

Title: Micro-expressions in Animated Agents

The overall goal of our research is to develop a system of intelligent multimodal affective pedagogical agents that are effective for different types of learners (Adamo et al., 2021). While most of the research on pedagogical agents tends to focus on the cognitive aspects of online learning and instruction, this project explores the less-studied role of affective (or emotional) factors. We aim to design believable animated agents that can convey realistic, natural emotions through speech, facial expressions, and body gestures and that can react to the students’ detected emotional states with emotional intelligence. Within the context of this goal, the specific objective of the work reported in the paper was to examine the extent to which the agents’ facial micro-expressions affect students’ perception of the agents’ emotions and their naturalness. Micro-expressions are very brief facial expressions that occur when a person either deliberately or unconsciously conceals an emotion being felt (Ekman &Friesen, 1969). Our assumption is that if the animated agents display facial micro expressions in addition to macro expressions, they will convey higher expressive richness and naturalness to the viewer, as “the agents can possess two emotional streams, one based on interaction with the viewer and the other based more » on their own internal state, or situation” (Queiroz et al. 2014, p.2).The work reported in the paper involved two studies with human subjects. The objectives of the first study were to examine whether people can recognize micro-expressions (in isolation) in animated agents, and whether there are differences in recognition based on the agent’s visual style (e.g., stylized versus realistic). The objectives of the second study were to investigate whether people can recognize the animated agents’ micro-expressions when integrated with macro-expressions, the extent to which the presence of micro + macro-expressions affect the perceived expressivity and naturalness of the animated agents, the extent to which exaggerating the micro expressions, e.g. increasing the amplitude of the animated facial displacements affects emotion recognition and perceived agent naturalness and emotional expressivity, and whether there are differences based on the agent’s design characteristics. In the first study, 15 participants watched eight micro-expression animations representing four different emotions (happy, sad, fear, surprised). Four animations featured a stylized agent and four a realistic agent. For each animation, subjects were asked to identify the agent’s emotion conveyed by the micro-expression. In the second study, 234 participants watched three sets of eight animation clips (24 clips in total, 12 clips per agent). Four animations for each agent featured the character performing macro-expressions only, four animations for each agent featured the character performing macro- + micro-expressions without exaggeration, and four animations for each agent featured the agent performing macro + micro-expressions with exaggeration. Participants were asked to recognize the true emotion of the agent and rate the emotional expressivity ad naturalness of the agent in each clip using a 5-point Likert scale. We have collected all the data and completed the statistical analysis. Findings and discussion, implications for research and practice, and suggestions for future work will be reported in the full paper. ReferencesAdamo N., Benes, B., Mayer, R., Lei, X., Meyer, Z., &Lawson, A. (2021). Multimodal Affective Pedagogical Agents for Different Types of Learners. In: Russo D., Ahram T., Karwowski W., Di Bucchianico G., Taiar R. (eds) Intelligent Human Systems Integration 2021. IHSI 2021. Advances in Intelligent Systems and Computing, 1322. Springer, Cham. https://doi.org/10.1007/978-3-030-68017-6_33Ekman, P., &Friesen, W. V. (1969, February). Nonverbal leakage and clues to deception. Psychiatry, 32(1), 88–106. https://doi.org/10.1080/00332747.1969.11023575 Queiroz, R. B., Musse, S. R., &Badler, N. I. (2014). Investigating Macroexpressions and Microexpressions in Computer Graphics Animated Faces. Presence, 23(2), 191-208. http://dx.doi.org/10.1162/

« less
Authors:
; ;
Award ID(s):
1821894
Publication Date:
NSF-PAR ID:
10340819
Journal Name:
AHFE International
Volume:
22
ISSN:
2771-0718
Sponsoring Org:
National Science Foundation
More Like this
  1. The goal of this research is to develop Animated Pedagogical Agents (APA) that can convey clearly perceivable emotions through speech, facial expressions and body gestures. In particular, the two studies reported in the paper investigated the extent to which modifications to the range of movement of 3 beat gestures, e.g., both arms synchronous outward gesture, both arms synchronous forward gesture, and upper body lean, and the agent‘s gender have significant effects on viewer’s perception of the agent’s emotion in terms of valence and arousal. For each gesture the range of movement was varied at 2 discrete levels. The stimuli of the studies were two sets of 12-s animation clips generated using fractional factorial designs; in each clip an animated agent who speaks and gestures, gives a lecture segment on binomial probability. 50% of the clips featured a female agent and 50% of the clips featured a male agent. In the first study, which used a within-subject design and metric conjoint analysis, 120 subjects were asked to watch 8 stimuli clips and rank them according to perceived valence and arousal (from highest to lowest). In the second study, which used a between-subject design, 300 participants were assigned to two groups ofmore »150 subjects each. One group watched 8 clips featuring the male agent and one group watched 8 clips featuring the female agent. Each participant was asked to rate perceived valence and arousal for each clip using a 7-point Likert scale. Results from the two studies suggest that the more open and forward the gestures the agent makes, the higher the perceived valence and arousal. Surprisingly, agents who lean their body forward more are not perceived as having higher arousal and valence. Findings also show that female agents’ emotions are perceived as having higher arousal and more positive valence that male agents’ emotions.« less
  2. The paper reports ongoing research toward the design of multimodal affective pedagogical agents that are effective for different types of learners and applications. In particular, the work reported in the paper investigated the extent to which the type of character design (realistic versus stylized) affects students’ perception of an animated agent’s facial emotions, and whether the effects are moderated by learner characteristics (e.g. gender). Eighty-two participants viewed 10 animation clips featuring a stylized character exhibiting 5 different emotions, e.g. happiness, sadness, fear, surprise and anger (2 clips per emotion), and 10 clips featuring a realistic character portraying the same emotional states. The participants were asked to name the emotions and rate their sincerity, intensity, and typicality. The results indicated that for recognition, participants were slightly more likely to recognize the emotions displayed by the stylized agent, although the difference was not statistically significant. The stylized agent was on average rated significantly higher for facial emotion intensity, whereas the differences in ratings for typicality and sincerity across all emotions were not statistically significant. A significant difference in ratings was shown in regard to sadness (within typicality), happiness (within sincerity), fear, anger, sadness and happiness (within intensity) with the stylized agent ratedmore »higher. Gender was not a significant correlate across all emotions or for individual emotions.« less
  3. We introduce a novel framework for emotional state detection from facial expression targeted to learning environments. Our framework is based on a convolutional deep neural network that classifies people’s emotions that are captured through a web-cam. For our classification outcome we adopt Russel’s model of core affect in which any particular emotion can be placed in one of four quadrants: pleasant-active, pleasant-inactive, unpleasant-active, and unpleasant-inactive. We gathered data from various datasets that were normalized and used to train the deep learning model. We use the fully-connected layers of the VGG_S network which was trained on human facial expressions that were manually labeled. We have tested our application by splitting the data into 80:20 and re-training the model. The overall test accuracy of all detected emotions was 66%. We have a working application that is capable of reporting the user emotional state at about five frames per second on a standard laptop computer with a web-cam. The emotional state detector will be integrated into an affective pedagogical agent system where it will serve as a feedback to an intelligent animated educational tutor.
  4. Abstract The enhancement hypothesis suggests that deaf individuals are more vigilant to visual emotional cues than hearing individuals. The present eye-tracking study examined ambient–focal visual attention when encoding affect from dynamically changing emotional facial expressions. Deaf (n = 17) and hearing (n = 17) individuals watched emotional facial expressions that in 10-s animations morphed from a neutral expression to one of happiness, sadness, or anger. The task was to recognize emotion as quickly as possible. Deaf participants tended to be faster than hearing participants in affect recognition, but the groups did not differ in accuracy. In general, happy faces were more accurately and more quickly recognized than faces expressing anger or sadness. Both groups demonstrated longer average fixation duration when recognizing happiness in comparison to anger and sadness. Deaf individuals directed their first fixations less often to the mouth region than the hearing group. During the last stages of emotion recognition, deaf participants exhibited more focal viewing of happy faces than negative faces. This pattern was not observed among hearing individuals. The analysis of visual gaze dynamics, switching between ambient and focal attention, was useful in studying the depth of cognitive processing of emotional information among deaf and hearing individuals.
  5. The paper reports progress on an NSF-funded project whose goal is to research and develop multimodal affective animated pedagogical agents (APA) for different types of learners. Although the preponderance of research on APA tends to focus on the cognitive aspects of online learning, this project explores the less-studied role of affective features. More specifically, the objectives of the work are to: (1) research and develop novel algorithms for emotion recognition and for life-like emotion representation in embodied agents, which will be integrated in a new system for creating APA to be embedded in digital lessons; and (2) develop an empirically grounded research base that will guide the design of affective APA that are effective for different types of learners. This involves conducting a series of experiments to determine the effects of the agent’s emotional style and emotional intelligence on a diverse population of students. The paper outlines the work conducted so far, e.g., development of a new system (and underlying algorithms) for producing affective APA. It also reports the findings from two preliminary studies.