Search

Research on the sense of agency has shown that being the author of an action changes the way we estimate the timing and the intensity of the action-effect. Yet, there is a dearth of attempts to assess the influence of agency on perception per se. The present study used the Representational Momentum paradigm to measure participants' visual anticipation of movement while manipulating their agency. In line with previous literature emphasizing the impact of social factors on visual anticipation and on the sense of agency, we additionally investigated the modulating power of social affordances on the relationship between agency and visual anticipation. We conducted two experiments where participants viewed a virtual agent directing a handshake gesture toward a second virtual agent. In a first experiment, we addressed the role of agency on visual anticipation by comparing a condition in which participants triggered the virtual agent's gesture with a condition where the computer triggered the gesture. Results showed greater forward movement anticipation when participants triggered the gesture. The second experiment investigated how altering social interaction parameters (interindividual distance and body posture) modulated the relationship between agency and visual anticipation. The outcome contrasted with the first experiment, with participants anticipating a backward movement of the hand when the computer triggered the gesture and displaying a null anticipation when participants triggered the gesture. Those two experiments highlighted how active involvement and environmental affordance interact to shape perception and allowed us to propose an updated model of agency processing.

International audience; Sign language (SL) motion contains information about the identity of a signer, as does voice for a speaker or gait for a walker. However, how such information is encoded in the movements of a person remains unclear. In the present study, a machine learning model was trained to extract the motion features allowing for the automatic identification of signers. A motion capture (mocap) system recorded six signers during the spontaneous production of French Sign Language (LSF) discourses. A principal component analysis (PCA) was applied to time-averaged statistics of the mocap data. A linear classifier then managed to identify the signers from a reduced set of principal components (PCs). The performance of the model was not affected when information about the size and shape of the signers were normalized. Posture normalization decreased the performance of the model, which nevertheless remained over five times superior to chance level. These findings demonstrate that the identity of a signer can be characterized by specific statistics of kinematic features, beyond information related to size, shape, and posture. This is a first step toward determining the motion descriptors necessary to account for the human ability to identify signers.

New technologies for autism focus on the training of either social skills or motor skills, but not both. Such a dichotomy omits a wide range of joint action tasks that require the coordination of two persons (e.g. moving a heavy furniture). The training of these physical tasks performed in dyad has great potential to foster inclusiveness while having an impact on both social and motor skills. In this paper, we present the design of a tangible and virtual interactive system for the training of children with Autism Spectrum Disorder (ASD) in performing joint actions. The proposed system is composed of a virtual character projected onto a surface on which a tangible object is magnetized: both the user and the virtual character hold the object, thus simulating a joint action. We report and discuss preliminary results of a field training study, which shows the potential of the interactive system.

International audience; Previous research has shown that human perceivers can identify individuals from biological movements, such as walking or dancing. It remains to be investigated whether sign language motion, which obeys to other constraints than pure biomechanical ones, also allows for person identification. The present study is the first to investigate whether deaf perceivers recognize signers based on motion capture (mocap) data only. Point-light displays of 4 signers producing French Sign Language utterances were presented to a group of deaf participants. Results revealed that participants managed to identify familiar signers above chance level. Computational analysis of the mocap data provided further evidence that morphological cues were unlikely to be sufficient for signer identification. A machine learning approach aiming to evaluate the motion features that can account for human performance is currently being developed. First results of the model reveal high accuracy for signer identification based on the same stimulus material, even after having normalized for size and shape. The present behavioral and computational findings suggest that mocap data contain sufficient information to identify signers, and this beyond simple cues related to morphology.

To estimate the time-to-contact (TTC) of a moving object, numerous studies have focused on the type of information or gaze strategy used by the observer. However, it remains to be determined whether and how attention could affect TTC estimation. In particular, how does TTC estimation operate when less attention is available? To answer this question, we conducted two experiments in which the participants had to perform an absolute (Experiment 1) or relative (Experiment 2) prediction-motion task, either alone (i.e., in single-task condition) or along with a secondary, visual working-memory task (i.e., in dual-task condition). In both experiments, we found that TTC estimation was superior in dual-task condition relative to single-task condition. This finding suggests that the reduction of available attention actually improves TTC estimation. We discuss possible explanations as well as theoretical implications for this seemingly counter-intuitive finding. Further research is needed to investigate if (in)attention facilitates or only shifts TTC estimation.

International audience; Virtual signers (or signing avatars) play an important role in the accesibility of information in sign languages. They have been developed notably for their capability to anonymize the signer's appearance and to enable dynamic or interactive scenarios. Recording real movements thanks to motion capture provides human-like, realistic and comprehensible signing animations. However, such accurate systems may also convey extralinguistic information such as identity, gender or emotional state. In the present work, we want to address the problem of gestural identity in the context of animated agents in French Sign Language (LSF). On the one hand, person identification from signing motion is assessed through psychophysical experiments, using point-light displays. On the other hand, a computational framework is developed for the analysis of LSF motion in order to investigate which features are critical for identification. For some applications, determining these movement parameters will enable controlling the gestural human traits of virtual signers.

Background.Through real-time behavioral observation systems, pain behaviors are commonly used by clinicians to estimate pain intensity in patients with low back pain. However, little is known about how clinicians rely on pain-related behaviors to make their judgment. According to the Information Integration Theory (IIT) framework, this study aimed at investigating how clinicians value and integrate information from lumbopelvic kinematics (LK), a protective pain behavior, and facial expression intensity (FEI), a communicative pain behavior, to estimate pain in patients with chronic low back pain (cLBP).Methods.Twenty-one experienced clinicians and twenty-one novice clinicians were asked to estimate back pain intensity from a virtual character performing a trunk flexion-extension task.Results.Results revealed that both populations relied on facial expression and that only half of the participants in each group integrated FEI and LK to estimate cLBP intensity. Among participants who integrated the two pain behaviors, averaging rule predominated among others. Results showed that experienced clinicians relied equally on FEI and LK to estimate pain, whereas novice clinicians mostly relied on FEI.Discussion.The use of additive rule of integration does not appear to be systematic when assessing others’ pain. When assessing pain intensity, communicative and protective pain behaviors may have different relevance.

International audience; Humans have developed a specific capacity to rapidly perceive and anticipate other people’s facial expressions so as to get an immediate impression of their emotional state of mind. We carried out two experiments to examine the perceptual and memory dynamics of facial expressions of pain. In the first experiment, we investigated how people estimate other people’s levels of pain based on the perception of various dynamic facial expressions; these differ both in terms of the amount and intensity of activated action units. A second experiment used a representational momentum (RM)paradigm to study the emotional anticipation (memory bias) elicited by the same facial expressions of pain studied in Experiment 1. Our results highlighted the relationship between the level of perceived pain (in Experiment 1) and the direction and magnitude of memory bias (in Experiment 2): When perceived pain increases, the memory bias tends to be reduced (if positive) and ultimately becomes negative. Dynamic facial expressions of pain may reenact an "immediate perceptual history" in the perceiver before leading to an emotional anticipation of the agent’s upcoming state. Thus, a subtle facial expression of pain (i.e., a low contraction around the eyes) that leads to a significant positive anticipation can be considered an adaptive process—one through which we can swiftly and involuntarily detect other people’s pain.

Objectives Two studies examined whether observers’ personality traits contribute to prosocial responses to others’ facial expression of pain. Experiment 1 examined the personality traits that could account for observers’ variability in estimating others’ pain intensity. Experiment 2 questioned to what extent the contribution of personality traits on inclination to help people in pain depend on observers’ beliefs about pain’ characteristics. Method 59 (experiment 1) and 76 (experiment 2) participants observed to 3-D realistic synthetic face movements mobilizing action units of pain, in order to estimate others’ pain. In experiment 2, painful localizations (e.g., chest, hand) were also manipulated. In each experiment, Big Five personality traits were assessed. Results Experiment 1 revealed that agreeableness and conscientiousness contributed to observers’ pain estimates across the increase of facial expression intensity. Experiment 2 showed that conscientiousness contributed to observers’ judgments whatever pain’ characteristics. Neuroticism was only salient for pain referring to life-threatening pain. Conclusion Prosocial response to others’ pain depends on agreeableness, conscientiousness and neuroticism. However, these links are modulated by the pain behavior elicited and observers’ belief about the characteristic of pain.

International audience; Advances in the use of virtual affective agents for therapeutic purposes in mental health opened a research avenue to improve the way patients interpret other's behavior as helpful instead of menacing. Here, the authors propose an original paradigm based on affective computing and virtual reality technologies requiring the assessment of helping intentions as well as self-monitoring metacognition. Sixteen healthy subjects played a 38-turns card games with a virtual affective agent (MARC) during which they had to guess between two cards the one that would be color-matched with another card. Their guesses could be oriented by the agent's emotional displays. Three subjective ratings on a percentage analog scale were recorded after each trial: helpfulness, self- monitoring, and sympathy. Help recognition and self-monitoring metacognitive ratings raise the question of the importance to enhance both components in therapeutic situations within psychiatric populations. Overall, this study exemplifies the promising use of virtual reality settings for future studies in the medical psychology field.

To predict where a thrown ball will land, an observer may use visual information about its trajectory. However, in addition, the thrower's body language (i.e., body movement and facial expression) may contain useful information that could be used by the observer to understand intention and emotional state. Here, we investigated how observers estimated a ball's landing point thrown by a virtual agent with different amounts of information from body language. In addition, occlusion time was varied to examine how it potentiates the use of body-language information. Results showed that body movement and facial expression carry information about thrower's effort. However, once the ball has left the thrower's hand, advance information on facial expression does contribute to judgments only if consistent with the amplitude of the throw. Moreover, as the occlusion time increases, a stronger influence of the body movement is observed for estimating the landing point. The overriding effect of ball's trajectory availability over body language is discussed.

Alzheimer’s patients display a wide variety of emotional and behavioral disorders (apathy, aggressiveness, depression, etc.) that evolve with the disease. When dealing with this type of patient, medical and nursing staff must be trained to manage their own non-verbal behavior. They must also be trained to understand the patient’s behavior in order to improve the quality of interactions and avoid difficult situations. Standardized patients (human actors playing the role of the patient) are among the most commonly used traditional training tools. However, they are expensive and not widely available. As for digital training materials (online courses, videos...), they lack interactivity. Virtual patients, which are interactive animated characters that simulate patient behaviors, are increasingly used to train caregivers to interact with patients. They enable a precise and dynamic control of realistic behaviors on the virtual patient, while ensuring consistency with the educational objectives of the training. In this thesis, we introduce the method used to design such an interactive and expressive virtual patient to simulate a virtual patient with Alzheimer’s disease. Our work is part of the field of affective computing. This multidisciplinary field is at the intersection of computer science and cognitive psychology. We conducted a Wizard of Oz experiment and collected a corpus of interaction data between 31 medical and nursing staff members and a controlled version of our virtual patient. These data were analyzed and integrated into the design of an automatic model aiming at controlling automatically the reactions of the virtual patient. We thus introduce the COPALZ model (COmputational model of Pathological appraisal biases for a virtual Alzheimer’s patient), which is based on the Appraisal Bias Model from cognitive psychology. Our COPALZ model generates pathological appraisals of our virtual patient in order to reproduce the emotional and behavioral disorders associated with Alzheimer’s disease. The evaluation results suggest that our model can generate pedagogically relevant situations by adjusting the severity of the disease and the mood instability of the virtual patient.; Les patients atteints d’Alzheimer présentent une grande variété de troubles des émotions et des comportements (apathie, agressivité, dépression . . .) qui évoluent avec la maladie. Face à ce type de patients, les membres du personnel soignant et médical doivent être formés à gérer leur propre comportement non-verbal et interpréter celui du patient afin d’améliorer la qualité des interactions et éviter les situations difficiles. Parmi les outils de formation traditionnels les plus utilisés, les patients standardisés (acteurs humains jouant le rôle du patient) sont chers et peu disponibles, et les nouveaux supports de formation numérique (cours en ligne, vidéos . . .) manquent d’interactivité. Les patients virtuels, qui sont des personnages animés interactifs simulant les comportements d’un patient, sont de plus en plus utilisés pour former les soignants à interagir avec des patients. Ils permettent un contrôle fin et dynamique de comportements réalistes à afficher sur ces personnages virtuels, tout en assurant la cohérence avec les objectifs pédagogiques de la formation. Nous présentons dans ces travaux de thèses la méthode employée afin de concevoir un tel patient virtuel interactif et expressif pour simuler une patiente virtuelle atteinte d’Alzheimer. Ces travaux sont inscrits dans le domaine de l’informatique affective, qui représente un contexte multidisciplinaire à la frontière entre informatique et psychologie cognitive. Une expérimentation de Magicien d’Oz nous a permis de récolter un corpus de données d’interaction entre 31 membres du personnel médical et soignant, et une version contrôlée de notre patiente virtuelle. Ces données ont été analysées et intégrées à la conception d’un modèle automatique visant à contrôler automatiquement les réactions de la patiente virtuelle. Nous introduisons ainsi notre modèle COPALZ (COmputational model of Pathological appraisal biases for a virtual Alzheimer’s patient), qui se base sur le modèle Appraisal Bias Model issu de la psychologie cognitive. Notre modèle COPALZ permet de générer des évaluations pathologiques de notre patiente virtuelle afin de reproduire les troubles des émotions et des comportements associés à la maladie d’Alzheimer. Les résultats de l’évaluation suggèrent que notre modèle permet de générer des situations pédagogiquement intéressantes en adaptant le niveau d’avancement de la maladie et l’instabilité de l’humeur de la patiente virtuelle.

Many technological barriers must be tackled in order to provide tools in Sign Languages (SLs) in the same way as for spoken languages. For that aim, further insights must be gained into multiple disciplines, in particular motion science. More specifically, the present thesis aims to gain insights into the possibility of anonymizing the movements of a signer, in the same way as a speaker can remain anonymous by modifying specific aspects of the voice.First, this thesis sheds light on general kinematic properties of spontaneous SL in order to improve the models of natural SL. Using 3D motion recordings of multiple signers, we show that the kinematic bandwidth of spontaneous SL highly differs from that of signs made in isolation. Furthermore, a Principal Component Analysis reveals that the spontaneous SL discourses can be described by a reduced set of simple, one-directional, movements (i.e., synergies).Furthermore, combining human data and computational modelling, we demonstrate that signers can be identified from their movements, beyond morphology- and posture-related cues. Finally, we present machine learning models able to automatically extract identity information in SL movements and to manipulate it in generated motion. The models developed in this thesis could allow producing anonymized SL messages via virtual signers, which would open new horizons for deaf SL users.; De nombreux obstacles technologiques doivent être surmontés afin d'outiller les Langues des Signes (LS) de la même manière que les langues parlées. Pour ce faire, il est nécessaire d'approfondir les connaissances dans de multiples disciplines, en particulier les sciences du mouvement. Plus précisément, cette thèse vise à étudier la possibilité d'anonymiser les mouvements d'un signeur, de la même manière qu'un locuteur peut rester anonyme en modifiant des aspects spécifiques de sa voix.Premièrement, cette thèse met en lumière les propriétés cinématiques de la LS spontanée afin d'améliorer les modèles de LS naturelle. En utilisant des données de mouvements en 3D de plusieurs signeurs, nous montrons que la bande passante cinématique de la LS spontanée diffère fortement de celle des signes isolés. Ensuite, une analyse en composantes principales révèle que les discours spontanés peuvent être décrits par un ensemble réduit de mouvements simples (i.e., synergies).De plus, en combinant données humaines et modélisation informatique, cette thèse démontre que les signeurs peuvent être identifiés à partir de leurs mouvements, au-delà de la morphologie et de la posture. Enfin, nous présentons des modèles d'apprentissage automatique capables d'extraire automatiquement l'information d'identité dans les mouvements de la LS, puis de la manipuler lors de la génération. Les modèles développés dans cette thèse pourraient permettre de produire des messages de LS anonymes via des signeurs virtuels, ce qui ouvrirait de nouveaux horizons aux signeurs sourds.