Body, Space, and Emotion: A Perceptual Study

ISSN: 1795-6889 www.humantechnology.jyu.fi Volume 13(1), May 2017, 32–57 32 BODY, SPACE, AND EMOTION: A PERCEPTUAL STUDY Abstract: The present study aims at providing a systematic account of emotion perception applied to expressive full-body movement. Within the framework of the lens model, we identified the decoding process underlying one’s capacity to categorize emotions while watching others’ behaviors. We considered the application of Laban movement analysis, a method focusing on qualitative aspects of movement. An original © 2017 Donald Glowinski, Sélim Yahia Coll, Naëm Baron, Maëva Sanchez, Simon Schaerlaeken, & Didier Grandjean, and the Open Science Centre, University of Jyväskylä DOI: http://dx.doi.org/10.17011/ht/urn.201705272517 This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. Donald Glowinski Neuroscience of Emotion and Affective Dynamics (NEAD Lab), Faculty of Psychology and Educational Sciences, and the Swiss Center for Affective Sciences (SCAS), University of Geneva and Swiss Integrative Center for Human Health (SICHH), Fribourg Switzerland Didier Grandjean Neuroscience of Emotion and Affective Dynamics (NEAD Lab), Faculty of Psychology and Educational Sciences, and the Swiss Center for Affective Sciences (SCAS), University of Geneva Switzerland Simon Schaerlaeken Neuroscience of Emotion and Affective Dynamics (NEAD Lab), Faculty of Psychology and Educational Sciences, and the Swiss Center for Affective Sciences (SCAS), University of Geneva Switzerland Maëva Sanchez Neuroscience of Emotion and Affective Dynamics (NEAD Lab), Faculty of Psychology and Educational Sciences, University of Geneva Switzerland Naëm Baron Neuroscience of Emotion and Affective Dynamics (NEAD Lab), Faculty of Psychology and Educational Sciences, University of Geneva Switzerland Sélim Yahia Coll Neuroscience of Emotion and Affective Dynamics (NEAD Lab), Faculty of Psychology and Educational Sciences, and the Swiss Center for Affective Sciences (SCAS), University of Geneva Switzerland Body, Space, and Emotion: A Perceptual Study 33 experimental setup used a contemporary choreography interpreted by four expert dancers in an environment that restricted their movement to their peripersonal space. Each performance consisted of a subtle or intense emotional interpretation of the choreography (e.g., happiness, anger, surprise, fear, and sadness). Results showed that emotions being expressed in this confined environment could still be identified, categorized, and associated with a profile of movement qualities and specific body parts. Keywords: nonverbal expressive movement, emotion, peripersonal space, lens model, Laban movement analysis, dance, choreography. INTRODUCTION: BODY, SPACE AND EMOTIONS The body is a tool of unparalleled power for expressing emotion beyond the use of verbal utterances, and this issue has been the focus of numerous research studies (e.g., de Gelder, 2013; Glowinski et al., 2011; Wallbott, 1998). However, the relationship between space and the body’s emotion expression powers has been far less studied. This is due in part to prior difficulties in exploring and recording human interaction or body expression with a high degree of precision and flexibility. Most of the performances recorded for scientific experiments take place within a laboratory setting to allow absolute control over lighting conditions that otherwise might affect the robustness of the tracking. In fact, far fewer studies engage the natural setting of the performance (e.g., open-air performance and theatre). The role of space, and specifically the impact of a small space, on the progression of movement, particularly on the full-body expression of an individual’s emotion, remains to be clarified. A better insight into such relationships is valuable for a wide range of domains, including psychological research on emotional and aesthetical expression or the field of human–computer interaction (Pantic, Pentland, Nijholt, & Huang, 2007). In a context that combines psychological, aesthetic, and human–computer interaction research, studies focusing on emotion and space often refer to boundary conditions, extreme situations (e.g., flight space), or phobias (e.g., claustrophobia; see Palinkas, 2001). To address the space–emotion relationship, we present a case study of a dance choreography that was performed by professional dancers conveying various expressive emotional intents. This experimental method drew upon previous work by Camurri, Lagerlöf, and Volpe (2003). What characterizes these performances, and distinguishes the novelty of our contribution, is that the space made available to each dancer corresponded to their peripersonal space. This peripersonal space has been approximated by the kinesphere in dance theory as the environmental sphere surrounding the body whose periphery can be easily reached by extending a limb (Laban & Lawrence, 1947; Sutil, 2013). We are interested in understanding how external observers can discriminate the expressed emotions based on the expressive behavior of the dancer within the boundaries of his/her peripersonal space. Our contribution integrated the Laban movement analysis (LMA) categories, used to describe qualitative movement, to fit the decoding process underlying such observers’ emotional categorizations. We also were interested in evaluating other factors’ impact on the emotion recognition process: (a) the observers’ expertise (e.g., does being a trained dancer augment sensitivity to emotion expression?), (b) the performance expressive intensity (e.g., are emotions better recognized when expressed in an emphatic manner?), and (c) the potential role of body parts Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 34 in emotion recognition (e.g., are there specific relationships between expressed emotions and body parts?). This paper is organized in the following manner. We first review the existing approaches in the literature that attempt to recognize emotions through body postures and movements with a specific focus on dance as a test case. Next, we present the developed experimental framework, which includes the steps dedicated to the recording and presentation of the stimuli and details related to the statistical analysis methods applied to the participants’ answers. This is followed by the presentation of our results: how emotions are recognized and how they can be described in terms of the LMA categories. In the Discussion section, we examine the impact of the participants’ expertise, the intensity of the stimuli, and the relationship between specific emotions and identified body parts. Finally, we conclude this paper by addressing future research directions. Body as a Source of Emotional Expressivity The body is a key source of information for emotion recognition. An increasing trend in research relates to facial and vocal expression, gesture, and dynamic body motion recognition (e.g., Glowinski et al., 2011). The recent development of low-cost digital image recording equipment, together with the advent of professional motion-capture technologies, has enabled a close analysis of nonverbal modalities in human communication of emotions and, in particular, bodily behavior (Wallbott, 1998). Recently, affective computing, along with the wide range of related application areas, has led the way to meet an increasing demand for the creation of natural, intelligent, adaptive, and personalized multimodal environments (Vinciarelli et al., 2012). Until the turn of the 21st century, various coding systems were proposed by psychologists. The main focus has been on emotional facial expression due in part to the pioneering work of Ekman, who offered a systematic account for facilitating explicit coding and categorization (FACS; Ekman & Friesen, 1978). A realm of new standards is emerging in this domain, opening opportunities for commercial applications of automatic emotion recognition. Furthermore, alternative approaches have been developed more in recent decades. Research results in psychology suggest, in particular, that body movements do constitute a significant source of affective information (Wallbott, 1998). For example, body gesture, as a complement to facial expressions, can help disambiguate emotional information (de Gelder, 2006; Todorov, Baron, & Oosterhof, 2008). Yet, the vast number and combination of body postures and gestures offers a higher degree of freedom for expressions that can be difficult to easily manage during analysis. No standard body-movement coding scheme equivalent to the FACS for facial expressions exists to facilitate decomposing bodily expression into elementary components. Various systems have been suggested by psychologists (e.g., Bobick, 1997; Dael, Goudbeek, & Scherer, 2013) but none has reached the consensus achieved by the Ekman’s system (Ekman & Friesen, 1978) for facial expression analysis. Alternative approaches have been developed to fill this need. For instance, research on upper body expressivity took advantage of the clear conceptualization of sign language (e.g., Gunes & Piccardi, 2009). The world’s many sign languages, now being extensively documented, have become a resource for emotion recognition. In sign language, signs made with the hands work in complex coordination with signs made with the face, head movements, torso shifts, gaze, gestures, and mimetic moves. As upper-body movements also correspond to what can be Body, Space, and Emotion: A Perceptual Study 35 captured easily through a Webcam in the typical working environment (i.e., where people sit in front of their desktop computer), the recognition of such movements has stimulated a realm of applications in this domain. Yet, with the advent of wearable computing, that is, devices worn on the body giving the potential for digital interaction, particular attention is now being devoted to specific limb expressivity (e.g., arms, fingers; see Velloso, Bulling, & Gellersen, 2013). In this context, the lack of systematic coding has been successfully compensated for through a promising alternative approach developed by Caramiaux (2014). Caramiaux, Montecchio, Tanaka and Bevilacqua (2014) investigated and demonstrated how the variability of the body behavior itself can stand as a central cue for capturing expressivity. However, other than the seminal work by Pollick, Paterson, Bruderlin, and Sanford (2001) on motion of knocking, very few experiments have investigated emotional communication through specific limb variations. The key issue in recognition of affective bodily expression is to consider the level of information that determines the quality of movement. The view the body as a whole represents a complete source of affective information is now receiving increased attention in the scientific literature. The recent interest into full-body emotion expressivity can be related to the diffusion of advanced motion capture systems (e.g., Vicon) and especially to the larger dissemination of RGB-depth cameras (e.g., Kinect) that allow for an affordable and relatively fine-tuned tracking of an individual’s body movement. Existing studies on full-body movement have used coarse-grained posture features (e.g., leaning forward or slumping back) or low-level physical features of movements (i.e., kinematics, see, for example, Bianchi-Berthouze, Cairns, Cox, Jennett, & Kim, 2006). Other approaches have exploited the dynamics of gestures, referring to psychological studies reporting that temporal dynamics play an important role in interpreting emotional displays (e.g., Kapur, Kapur, Virji-Babul, Tzanetakis, & Driessen, 2005). Towards a Unified Approach One may note the disparity between the different approaches and, as pointed out earlier, the lack of broader and systematic view to address emotion recognition based on full-body expressivity. A few attempts have drawn inspiration from dance notation and theories (Camurri et al., 2003; Laban & Ullmann, 1971). The key issue is to consider the level of information that determines the quality of movement, that is to say the general characteristics about the way a movement is performed (e.g., the effort dimensions of LMA described below). This level of information may lie between the low-level features (e.g., position of and derivatives in a limb’s trajectory that report a mere displacement and inform about a specific gesture or motion activity) and a high level of information related to the emotional categories that people may use to infer emotional attitudes through observation (Glowinski et al., 2011). In this context, Laban’s (1947) conceptualization has proved useful in modeling what could be an intermediate level of information representing qualitative properties of movement where expressivity is embedded and conveyed. Emotion recognition may ultimately rely on the specific combination of these qualitative properties of movement (Glowinski et al., 2011). Since the pioneering work led by Zhang et al. (2006), an increasing number of computational implementations have been suggested but much more needs to be done on the perceptual side. Thus, our study aims to tackle this aspect by focusing on the perception of emotions Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 36 expressed through full-body movement in a restricted environment (i.e., the performer’s peripersonal space). Our specific goal is to provide a systematic account of the decoding process underlying such bodily emotion recognition and observe the respective impact of the observer’s expertise, the intensity of the performance expressivity, and role of body parts in this process. In particular, we are interested in understanding whether an intermediate level of emotional information discernment related to the perception of movement qualities based on the LMA conceptual framework (Laban & Ullmann, 1971) could help in tying the low and high levels of emotional information. We expect this study will shed light on to which level of information may be decisive in researchers’ understanding of the perceptual processes underlying emotion classification. The research approach to the expression of emotion employed in this study relies on the lens model initially developed by Brunswik (1956), adapted to the performing arts by Juslin and Laukka (2003), and recently reviewed by Glowinski et al. (2014; see Figure 1). According to this model, the analysis of emotion expression must take into account both the sender’s (e.g., the performer) and the receiver’s (e.g., the observer) perspectives. Two types of processes are thus considered: emotion expression/communication through body behavior (e.g., the proximal cues exhibited by a performer) and emotion recognition (e.g., the observer inferences based on the behavioral cues of that performer). Figure 1. An illustration of the revisited lens model (Brunswik, 1956) integrating the effort dimensions (i.e., factors and elements) of Laban movement analysis (Laban & Lawrence, 1947). This conceptual framework includes both the point of view of the sender (e.g., the dancer) and of the receiver (e.g., spectator). The respective encoding and decoding processes of emotion expression through body components can be analyzed in a systematic way. In particular, the attribution of a specific emotion category during the recognition phase of the movement can be related to the way a spectator combines bodily, effort-based, features of the dancer’s performance. Body, Space, and Emotion: A Perceptual Study 37 Perceptual Evaluation in Terms of Movement Qualities An analysis in terms of movement qualities helps identify the differences between the perspective of the person expressing the movement (i.e., the sender) and that of the person observing the same movement (i.e., the receiver). According to Laban (as cited by Sutil, 2013), movement can be experienced intuitively as a continuum, an indivisible flux of changes. Alternatively, movement can be rationalized as a series of snapshots that can be ordered, structured, and formalized as the building blocks of a representation (Glowinski, Camurri, Chiorri, Mazzarino, & Volpe, 2007). This conceptual framework facilitates investigating and understanding the implicit or explicit strategies an observer applies in manipulating the collected snapshots of movement in building up a complete, sensitive, and yet subjective representation of the (performer’s) sequence. Supported by a profound dualistic approach, Laban’s key movement concepts come in pairs of opposites (Laban & Lawrence, 1947). The theory of efforts developed by Laban aims at characterizing such dynamism in relation to four basic properties (effort factors): weight, space, time, and flow (Laban & Lawrence, 1947). Each of these factors is in turn divided into opposed subcategories known as effort elements (heavy–light, direct–indirect, quick–slow, and free– bound). These effort elements allow researchers to understand the fundamental qualitative differences in human movement. As stated by Sutil (2013, p. 5), “The difference between punching someone in anger and reaching for a glass is slight in terms of body organization—both rely on extension of the arm and the same spatial direction of the movement. The weight of the movement and the intensity of the movement are very different, though.” In this study, the Laban elements (Laban & Lawrence, 1947) are operationalized via measurable descriptions, described as follows: 1. The weight element considers the individual's movements in relationship to gravity and may describe its vigorousness. The two subcategories associated are heavy (i.e., powerful) and light (i.e., delicate). 2. The space element here considers the individual’s movements related to his/her peripersonal space. The two subcategories associated are direct and indirect. Indirect motion is interrupted and roundabout, and direct motion proceeds along a mostly straight line without deviation. 3. The time element is a measure of movement activity speed. The two associated subcategories are quick (i.e., sudden and urgent) and slow (i.e., sustained, continuous, and time stretching). 4. The flow element characterizes the continuity of the movement. The two associated subcategories with this element are free (i.e., a fluid and released movement) and bound (i.e., a controlled or contained movement). Emotion Categories, Peripersonal Space, and Laban’s Effort Dimensions Based on the qualitative approach of movement expressivity, recent computational studies have investigated how emotion could be rendered by integrating the effort dimensions (i.e., factors and elements) of Laban movement analysis (Laban & Lawrence, 1947). Initially, the objective in this domain was not to access perceptual processes only but also to create complementary strategies to organize and classify large databases on motion that included more subtle aspects Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 38 of expressions relating to emotion. A first attempt by Wakayama, Okajima, Takano, and Okada (2010), then Okajima, Wakayama, and Okada (2012), showed that motion retrieval can benefit from the use of a subset of LMA dimensions, especially when searching for data (indications, instances) on body movement in large research databases (Kapadia, Chiang, Thomas, Badler, & Kider, 2013). Recently, Aristidou, Charalambous, and Chrysanthou (2015) inspected the similarities among various emotional states classified according to the arousal and valence of Russell’s (1980) circumplex model and a subset of features that encode stylistic characteristics of motion based on the LMA. Overall, previous experimental results, based on video processing or body limbs’ trajectory dynamics, show that these features can be extracted using the LMA dimensions and thus allow researchers to encode body posture differences depicting emotion expression. The pertinence of Laban’s (Laban & Ullman, 1971) dimensions as descriptors for motion expressivity can be attested further by their use in avatar animation. Since the seminal work of Chi, Costa, Zhao, and Badler (2000), various studies have demonstrated that LMA-derived dimensions can be exploited efficiently in motion parameterization and expression (Zhao & Badler, 2005). Unfortunately, few studies have considered the application of Laban-based analysis in understanding perceptual processes during emotion recognition. Levy and Duke (2003) used LMA dimensions to score the capacity of nonprofessional dancers to improvise sequences of expressive movement. Correlation analyses in emotion recognition revealed relationships among the emotional states of depression and anxiety and certain movement qualities. Focusing on the specific case of walking, Crane and Gross (2013) explicitly instructed participants to use LMA-based analysis to classify the observed nuances of emotions. Sadness, anger, contempt, and joy were decoded with accuracy that ranged from 74% to 32 %, respectively; for most of the targeted emotions, decoding accuracy was related to the four effort factors (Serino, Annella, & Avenanti, 2009). Peripersonal Space, Kinesphere, and Emotions Peripersonal space has been defined in contrast to general space in a way similar to how space is defined in geometry or topology (Serino et al., 2009). In dance theory, this peripersonal space has been approximated further by a kinesphere, which refers to the space occupied by an outreaching body without it moving from one spatial location (Laban, 1966). Specifically, it can be represented as an area within reach of the body’s extended limbs which, when projected in all directions from the body’s center, can be conceived as a totality of movement or a sphere of movement (Sutil, 2013). Therefore, the kinesphere gives an operationalized way to analyze peripersonal space (see Figure 2) and how each individual might explore the surrounding environment to express emotion and, in turn, how such peripersonal space may reciprocally impact on emotional expression. Dance as a Test Case Dance appears as an ideal test case to study emotion in relation to movement expressivity. It condenses within a minimal amount of space and time the human capacity to express emotional information. Aesthetically, it is possible to create a minimally expressive dance or a dance that Body, Space, and Emotion: A Perceptual Study 39 Figure 2. Illustration of the peripersonal space as operationalized by the concept of kinesphere (Laban, 1966). The figure on the right represents the motion capture of the dancer’s movement (on the left) that was used in the perceptual experiment presented in this paper. unfolds slowly. Typically, dance provides no functional or utilitarian use of the space (i.e., related to a practical task achievement, such as opening a door or typing on a keyboard), but rather to express oneself. In contemporary Western dance art, however, the aesthetics have turned into quite everyday activities; therefore objects and space may be interrelated in very functional ways. Dance as a test case allows the experimenter to observe movement of an undetermined wide range of complexity that can vary according to the level of the dancer’s expertise (e.g., from beginner to accomplished dancer) as well as according to the choreography. Specific to our research interest, dance also can reveal exceptional and evolved ways of exploiting the surrounding space. By focusing on the expression of emotion, through movement in a limited space or constrained environment, we sought to observe how a dancer faces this challenge: Specifically, we were interested in how the dancer could transform a source of what could appear to most people as a stress or difficulty into a positive experience. Therefore, we experimentally defined a limiting condition where the performer was bounded by a choreography that limited the variety of possible movements within a further, restricted environment that constrained the dancer’s displacement within his/her personal space. As a consequence, we expected that this challenging situation could impact the performer’s creative capacity in expressing emotion through the only dance component that remained freely available: body expressivity. We explored this issue by assessing how external observers perceived such emotional body expression. Laban Movement Analysis and Dance The LMA is widely used in dance, either to annotate and generate choreography or to train dancers. With the advent of new systems for motion capture, dancers have shown an increased interest in using these new forms of interaction to map in real-time their expressive body movements to audio or visual feedback. The LMA dimensions have resurfaced as a source of inspiration in capturing key expressive variations in dance performance and for improving the “naturalness” during interaction with automatic systems (Mancas, Glowinski, Volpe, Coletta, & Camurri, 2010). Drawing upon Laban’s approach, Camurri et al. (2003), Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 40 and Van Dyck, Vansteenkiste, Lenoir, Lesaffre, and Leman (2014) developed a qualitative approach to human full-body movement for emotion recognition. In this study, LMA dimensions were approximated through a combination of low-level physical features to allow for a coarse description of an encoding process of emotion through body behaviors (e.g., emotion arousal revealed by acceleration peaks). Based on a more explicit computational modeling of Effort-Shape features, Alaoui, Caramiaux, Serrano, and Bevilacqua (2012) developed an interactive augmented dance performance that extracts movement qualities (energy, kick, jump/drop, verticality/height and stillness) to generate a visual simulation. However, these previous works overlooked the experiences that can be gained through perceptual studies based on the LMA. For dance examples, it is useful to consider the key conceptual distinctiveness embodied in this art form. Dance can be considered a specific case of stylized body movements in which the entirety of movement encodes a particular emotion. Stylized motions usually originate in laboratory settings, where subjects are asked to freely act on an emotion without any constraints. Another key distinction may refer to the propositional and nonpropositional aspects of movement (Boone & Cunningham, 1998). Raising one’s hand to indicate stop, for example, may be considered a propositional movement that constitutes established signs to transmit shared meaning. As stated in Camurri et al. (2003), emotions can be expressed through propositional movement (e.g. a clenched fist to show anger or raised arms to demonstrate joy), whereas nonpropositional movements are embodied not in discrete, easily segmented motions but rather through a subtle combination of movement qualities (e.g., lightness or heaviness). From the point of view of a perceiver, this distinction could be interpreted as a shift in attentional focus, whether on the configurational aspects (e.g., gesture as an explicit, well-delimited code) or the dynamic itself (i.e., how one expresses emotional intent). In this study, we focused on the nonpropositional style of movements as represented by dance sequences. In our research, we considered how external observers combine body movement based on the LMA to recognize the emotional intent of contemporary dancers during their performance. Specifically, we focused on assessing the significance of the time, weight, space, and flow factors featured in emotion recognition by external observers. METHODS Participants Dancers Two female (D1 and D2) and two male (D3 and D4) professional contemporary dancers were recruited for this study. Their average age was 28 years, and they were remunerated 100 CHF for their participation. All participants provided signed informed consent prior to study participation. The protocol for this study was approved by the University of Geneva, Switzerland, at the Faculty of Psychology and Educational Sciences. Dancers were asked to wear tight clothes, their hair was tied up, and all jewelry and accessories were removed for the recordings. Body, Space, and Emotion: A Perceptual Study 47 Figure 8. Plot of the interaction between the emotion expressed by the dancers and the observers’ perceived intensity of the dances in the fear scale. Vertical bars denote a 95% confidence interval. Figure 9. Histogram representing differences in Laban’s criteria evaluation (speed, weight, extension and flow) between each expressed emotion (anger, happiness, neutral, fear, surprise and sadness; N = 48). Vertical bars denote a 95% confidence interval. Time Factor As shown by the comparisons, anger was rated as significantly quicker, representing sudden and urgent movement, χ2(1, N = 48) = 99.65, p < .001, than the other emotions. Moreover, fear and surprise were rated as significantly quicker, χ2(1, N = 48) = 127.29, p < .001, than the happy, neutral, and sad expressions. Neutral and sad expressions were rated significantly Laban’s factors flowspaceweight time Evaluation of Laban’s factors Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 48 slower than happiness, χ2(1, N = 48) = 24.92, p < .001. However, no significant difference was observed between neutral and sad expressions, χ2(1, N = 48) = 1.03, p = .31, or between fearful and surprise expressions, χ2(1, N = 48) = 0.86, p = .36. Weight Factor Happiness was rated as significantly lighter than the other emotions, χ2(1, N = 48) = 4.56, p < .05. Conversely, sadness was rated as significantly more anchored and heavier than the other emotions, χ2(1, N = 48) = 19.50, p < .001. No significant difference was observed between anger and neutral, χ2(1, N = 48) = 0.32, p = .57, or anger and surprise, χ2(1, N = 48) = 0.59, p = .44) expressions. However, anger was recognized as significantly more anchored than fear, χ2(1, N = 48) = 5.66, p < .05. No significant difference was obtained between the expressions of neutral and fear, χ2(1, N = 48) = 1.99, p = .16, neutral and surprise, χ2(1, N = 48) = 0.002, p = .96, or fear and surprise, χ2(1, N = 48) = 2.71, p = .10. Space Factor Sadness was rated as significantly more indirect than the other emotions, χ2(1, N = 48) = 70.25, p < .001. Anger and happiness were rated as significantly more direct than neutral, fear, or surprise expressions, χ2(1, N = 48) = 19.99, p < .001. Between these two groups, happiness was rated as significantly more direct than anger, χ2(1, N = 48) = 8.58, p < .01, and neutral was considered significantly more indirect than the fear or surprise expressions, χ2(1, N = 48) = 7.45, p < .01. However, no significant difference was observed between fear and surprise, χ2(1, N = 48) = 0.99, p = .32. Flow Factor Happiness was rated as significantly more fluid, that is, referring to a free, relaxed, and uncontrolled movement, than the other emotions, χ2(1, N = 48) = 36, p < .001. By contrast, surprise and sadness were the two emotions recognized as the most bounded, χ2(1, N = 48) = 26.22, p < .001. No significant difference was observed between them, χ2(1, N = 48) = 0.74, p = .39. Moreover, anger was rated no differently than the neutral or fear expressions, χ2(1, N = 48) = 0.34, p = .56 and χ2(1, N = 48) = 0.78, p = .38, respectively, which were also not significantly different, χ2(1, N = 48) = 0.03, p = .86. Relationships Among Expressed Emotions and Body Parts Emotion showed a significant main effect on the dependent variables: the head, χ2(5, N = 48) = 71.23, p < .001; arms, χ2(5, N = 48) = 12.23, p < .05; shoulders, χ2(5, N = 48) = 11.23, p < .05; and pelvis, χ2(5, N = 48) = 12.01, p < .05. However, no significant difference was observed in the legs variable, χ2(5, N = 48) = 6.42, p = .27. For sake of clarity, we developed a visualization procedure to illustrate how each perceived emotion could be systematically related to a specific body area. Based on the analysis of the contrasts, we individuated three levels of gray-scale intensities (light grey, dark grey, and black) to designate the level of explicit attention the observers noted regarding the head, Body, Space, and Emotion: A Perceptual Study 49 shoulders, arms, pelvis, and legs with respect to the emotion expressed (see Figure 10). Black indicates the highest significant correlation between an emotion and the observed body part; a light gray code indicates the lowest one. The other body parts, for which correlation was neither the highest nor the lowest, were colored in dark gray. We report in the following the results of the contrast analyses. The observers noted the head as the focus during dances involving fear in comparison to neutral condition: χ2(1, N = 48) = 20.21, p < .001); shoulders (sadness in comparison to surprise: χ2(1, N = 48) = 7.37, p < .01); arms (anger in comparison to fear: χ2(1, N = 48) = 9.47, p < .01); pelvis (happiness in comparison to fear: χ2(1, N = 48) = 8.49, p < .01); legs (anger in comparison to sadness: χ2(1, N = 48) = 3.61, p = .06). Figure 10. Degree of reported observers’ attention to the different body areas depending on emotions expressed by the dancers. Neutral and Surprise did not appear to have a clear consensus regarding which body part most observers observed. Rather, the observers had responses across multiple body areas. Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 50 DISCUSSION In agreement with the literature (Camurri et al., 2003; Crane & Gross, 2013; Shikanai, Sawada, & Ishii, 2013), our results show that the emotions of happiness, anger, and sadness are typically identified correctly by the observers. Of the three, sadness is recognized most often, followed by happiness. This partially reproduces the results of the studies by Camurri et al. (2003) and Dittrich, Troscianko, Lea, and Morgan (1996), in which the highest recognition rate was obtained for sadness, followed by anger and joy. For the neutral performance, in which no emotion was expressed, observers also predominantly perceived it correctly. However, when the dancers expressed fear or surprise, these two emotions were not readily distinguished. This perceptual confusion finds support from two prior studies. In the seminal study by Ekman and Friesen (1974) on emotional facial expressions and in the study of a full-body behavior in a daily life scenario by Meijer (1989), expressions of surprise tended to be recognized as fear. In our study, the perceived expressions of fear and surprise shared most of the Laban factors: time, space, and weight. For the Laban elements associated with time, for example, speed changes were seen as particularly fast for these two emotions and for the weight factor, the performance was perceived as moderately anchored to the ground and light and delicate. Contrary to our expectations based on earlier studies (e.g., Bläsing et al., 2012; Broughton & Davidson, 2014; Cross, Kirsch, Ticini, & Schütz-Bosbach, 2011), the expertise of the observers had little influence on the emotion recognition in this study. Overall, our results suggest that the affective components of body expression are less driven by expertise in dance; we saw broadly consistent responses across observers, in line with van Paasschen, Bacci, & Melcher (2015). However, the use of a point-light display based on MoCap recordings could have diminished the impact of expertise. This type of minimalistic display transmits a sufficient amount of gestural information for emotion recognition when deployed in a restricted environment, and perhaps even those without dance experience were able to recognize the emotional expressions from these images. It is possible that additional physical evidence (e.g., facial expression and skin texture) may have allowed for one’s dance expertise to become more apparent. These findings also could be explained by the fact that the criterion chosen to classify our group (i.e., novice, expert, and nonexpert) was not sufficient enough to differentiate the levels of expertise. Our criterion for expertise was based on the standard levels of training period (8 years of regular practice) that Swiss dance institutions suggest to their most motivated students who are preparing for a professional career. Among the 22 experts who participated in our study, only four were dancing professionally. In future studies, we plan to include a larger number of experts, specifically professional dancers, to test whether sensorimotor expertise or aesthetic familiarity may impact emotion recognition in the specific context of this study (see also Bläsing et al., 2012). Results confirmed that emotions expressed in the high intensity condition (i.e., with a higher emphasis and demonstrative manner) during performances are better recognized than in the low emotional intensity, where emotion would be more subtlety expressed (Burger, Saarikallio, Luck, Thompson, & Toiviainen, 2013). Our results show a statistically significant and a marginal difference, respectively, for fear and surprise in that the emotion recognition was higher in the low intensity condition. However, anger, joy, and sadness were perceived in a similar way for Body, Space, and Emotion: A Perceptual Study 51 both intensities. This result goes against our expectations based on the literature and the results obtained by Hill and Pollick (2000). These authors showed that a particular increase in the expressive intensity of movements produced a better recognition of joy and anger. In their study, however, the differences between the expressive conditions were achieved through offline manipulation of the recorded point-light display used to render the dancers’ performances. More noteworthy is that fear was recognized by the observers as a more intense conveyed movement in our context of investigating the body gestures and movements. One can argue that, in real-life conditions, the experience of fear might be related to reduced motion in body parts in order to be less detectable by dangerous animals or conspecifics in a threatening situation. Such fear-related body movement might have been integrated as an internal representation of fear by dancers and spectators, explaining our results. Results showed that Laban’s effort dimensions can inform on an intermediate level of perceptual processing underlying the emotion recognition. They revealed that the movement qualities can be discriminated in a statistically significant way and be significantly associated with emotion portrayals, thus highlighting their expressive pertinence. Considering the time factor, the anger, fear, and surprise emotions were positively related to quicker element ratings in comparison to the sad and neutral conditions. This is in line with the results in the scientific literature (Camurri et al., 2003; Crane & Gross, 2013; Meijer, 1989). As previous research has shown, fear and surprise cannot be distinguished based on the time factor alone. Concerning the weight factor, performances expressing joy turned out to be lighter and delicate whereas those expressing sadness were seen as firmer and anchored to the ground. However, this single criterion may not be sufficient to discriminate between surprise, anger, and fear. For the space factor, the movement underlying the joy and anger emotions was considered as more direct than those of sadness, confirming recent studies on this specific issue (Crane & Gross, 2013; Shikanai et al., 2013). This factor alone, however, did not allow for any distinction between fear and sadness. Finally, the flow factor was positively rated in the happiness emotion, displaying the LMA elements of relaxed, free, and uncontrolled movement, whereas surprise and sadness seemed to include bound, tense, and controlled types of movement. These results replicate the findings in Camurri et al. (2003). In addition, our results show that the surprise and fear emotions, which prior search has found are typically confused by observers, could be distinguished on the flow factor: The movement related to fear was rated as more uncontrolled than that of surprise (see also Wallbott, 1998). These findings, in totality, are consistent with Crane and Gross’s observations that emotions displayed by dancers affect movement style in distinctive ways that could be described consistently with a specific combination of the effort dimensions (Crane & Gross, 2013). An original finding of this study concerns the how various body parts attract the attention of observers according to which emotions they perceive expressed by the dancers. It extends and confirms an original approach developed by Nummenmaa, Glerean, Hari, and Hietanen (2014) to observe how emotions may be preferentially related to specific body parts. Our results concerning the upper body parts (i.e., head, shoulders, arms) and the pelvis—and the absence of any significant effect concerning the legs—are confirmed partially by Sawada, Suda, & Ishii (2003) and by empirical observation from early contemporary dance. In this period, expressions and movements were confined essentially to the upper body, particularly the chest. Legs were less used. In the mid-20th century, choreographer Ted Shawn stated, Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 52 “The torso must become the most sensitive and expressive part of the body” (Shawn, 1963, p. 63). Our results, however, show that the head attracted more observers’ attention when watching dance performances expressing happiness, anger, or no emotion (the neutral condition) than fear, surprise, or sadness. Arms were primarily observed during performances expressing anger; dances expressing sadness attracted less attention to the arms. Concerning the shoulders, the happy, sad, and neutral conditions attracted the participants’ attention to this area of the body, whereas performances expressing surprise did significantly less. Finally, the pelvis was considered the most salient when watching happy and neutral performances and the least when sad dances were presented. Our results provide a further detailed insight on how differentiated emotion recognition is associated with specific body parts. The use of peripersonal space (Cléry, Guipponi, Wardak, & Ben Hamed, 2015; Serino et al., 2009) as an environmental restriction to constrain dancers’ expressive movement also is a unique component of this study. It allowed us to evaluate whether expressed motion within a limited environment could still be recognized by observers. Our findings showed that angry, fearful, happy, surprised, and sad emotions can still be well discriminated in constrained movement. In addition, our findings highlight that not only space-related features but also complementary movement qualities (e.g., flow, weight, and time) characterize the decoding process of emotion expression in a systematic manner within this specific context (see also Taffou & Viaud-Delmon, 2014). To better understand how emotion recognition may vary in relation to a space representation, a future experimental protocol could include systematic manipulation of the space rendering during the stimuli presentation. The same point-light displays could be presented in a virtual environment that differs in terms of space extension (e.g., either strictly matching an individual’s peripersonal space or placing a dancer’s rendered movement within a larger neutral space). On the other hand, these results reveal the dancer’s potential for exploiting his/her peripersonal space for emotion expression. According to Cléry et al. (2015), not only is the type of action performed in the representation of peripersonal space key, but also the emotional consequences of the actions can dynamically modify it. A computational analysis of the Laban effort dimensions may help clarify the encoding processes of emotion through expressive body behavior. Of specific interest will be the relationship among the expressed emotions, the exploration of the confined space, and the qualitative features of movement. An experimental manipulation also could consider the dance performance within a wider variety of environments (e.g., restricted to peripersonal space or to a larger space, such as a theater stage). Finally, this study reveals the potential of using dance as test case for investigating emotion expression and peripersonal space representation. CONCLUSIONS As novel digital environments increase the degree of freedom in movement expression, ongoing research would benefit from a conceptual framework and set of methodological procedures to consider the implicit effects of the space factor on emotion expressivity and perception. Using dance as a test case and a choreographed performance with a variety of emotional variations, we systematically considered the decoding processes underlying a spectator’s identification of the emotion expressed within the performer’s peripersonal space. Body, Space, and Emotion: A Perceptual Study 53 Within the integrated conceptual apparatus of the lens model and the LMA, we contribute by offering a better insight into the qualitative features of body expression on which external observers may rely. We further reveal commonalities in perceptual strategies developed by dance experts and nonexperts. This study sets the agenda for future developments in the thriving and quickly evolving field of motion analysis and on nonverbal communication in technologically mediated environments. Future work includes (a) extending and replicating with a greater number of participants the correlations observed so far, and (b) establishing relationships within the encoding process (i.e., those implemented by the dancers through their body behavior). The computational modeling of LMA features may be exploited in this perspective. IMPLICATIONS FOR THEORY OR APPLICATION Our study is advancing the technological breakthrough in the analysis of full-body expressivity but also acknowledges the embodied turn toward cognition research, confirming the prominent role of emotion and embodiment as essential components of cognitive processes. We used dance as a test case and bound the dancer’s movements to his/her peripersonal space in order to better study the crucial impact of space constraints on emotional body expressivity and observers’ perception of emotions expressed within those constraints. Drawing upon the Laban movement analysis framework seems an efficient approach for recognizing emotions and revealed underlying perceptual process. Used in a Brunswik’s (1956) lens model, we believe those qualities categorized along the Laban’s effort dimensions are intermediate-level key components in emotion recognition. This study could help advance a new generation of digital environments, allowing for natural and emotionally vivid interactions. Equally, this line of research could facilitate better automated recognition of emotion from bodily movement. In addition, it suggests how dance research can influence a wide variety of disciplines also interested in exploring the perception and interpretation of emotional conveyance. ENDNOTES 1. More information on the Nexus software is available at http://www.vicon.com/products/software/nexus 2. More information on the video capture software used to display the motion capture video https://www.techsmith.com/camtasia.html 3. More information on the software used to develop the online questionnaire at http://www.qualtrics.com REFERENCES Alaoui, S. F., Caramiaux, B., Serrano, M., & Bevilacqua, F. (2012). Movement qualities as interaction modality. In Proceedings of the Designing Interactive Systems Conference (pp. 761–769). Newcastle, UK: ACM Press. doi: 10.1145/2317956.2318071 Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 54 Aristidou, A., Charalambous, P., & Chrysanthou, Y. (2015). Emotion analysis and classification: Understanding the performers’ emotions using the LMA entities. Computer Graphics Forum, 34(6), 262–276. doi: 10.1111/cgf.12598 Banziger, T., & Scherer, K. R. (2007). Using actor portrayals to systematically study multimodal emotion expression: The GEMEP corpus. In A.C.R Paiva., R. Prada, & R. W. Picard (Eds.), Affective computing and intelligent interaction (pp. 476–487). Berlin, Germany: Springer. doi: 10.1007/978-3-540-74889-2_42 Bianchi-Berthouze, N., Cairns, P., Cox, A., Jennett, C., & Kim, W. (2006). On posture as a modality for expressing and recognizing emotions. In Proceedings of the Emotion in HCI Workshop (pp. 74–80). London, UK: University College of London Press. Bläsing, B., Calvo-Merino, B., Cross, E. S., Jola, C., Honisch, J., & Stevens, C. J. (2012). Neurocognitive control in dance perception and performance. Acta Psychologica, 139(2), 300–308. doi: 10.1016/j.actpsy.2011.12.005 Bobick, A. F. (1997). Movement, activity and action: The role of knowledge in the perception of motion. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 352, 1257–1265. doi: 10.1098/rstb.1997.0108 Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., & White, J. S. S. (2009). Generalized linear mixed models: A practical guide for ecology and evolution. Trends in Ecology and Evolution, 24(3), 127–135. doi: 10.1016/j.tree.2008.10.008 Boone, R. T., & Cunningham, J. G. (1998). Children’s decoding of emotion in expressive body movement: The development of cue attunement. Developmental Psychology, 34(7), 1007–1016. doi: 10.1037//00121649.34.5.1007 Broughton, M., & Davidson, J. W. (2014). Action and familiarity effects on self and other expert musicians’ Laban effort-shape analyses of expressive bodily behaviors in instrumental music performance: A case study approach. Frontiers in Psychology, 5(10), 1201–1206. doi: 10.3389/fpsyg.2014.01201 Brunswik, E. (1956). Perception and the representative design of psychological experiments. Berkeley, CA, USA: University of California Press. Burger, B., Saarikallio, S., Luck, G., Thompson, M. R., & Toiviainen, P. (2013). Relationships between perceived emotions in music and music-induced movement. Music Perception: An Interdisciplinary Journal, 30(5), 517–533. doi: 10.1525/mp.2013.30.5.517 Camurri, A., Lagerlöf, I., & Volpe, G. (2003). Recognizing emotion from dance movement: Comparison of spectator recognition and automated techniques. International Journal of Human Computer Studies, 59(1), 213–225. doi: 10.1016/S1071-5819(03)00050-8 Caramiaux, B. (2014). Motion modeling for expressive interaction: A design proposal using Bayesian adaptive systems. In Proceedings of the International Workshop on Movement and Computing (pp. 76–81). Paris, France: ACM Press. doi: 10.1145/2617995.2618009 Caramiaux, B., Montecchio, N., Tanaka, A., & Bevilacqua, F. (2014). Adaptive gesture recognition with variation estimation for interactive systems. ACM Transactions on Interactive Intelligent Systems, 4(4), 1– 34. doi:10.1145/2643204 Chi, D., Costa, M., Zhao, L., & Badler, N. (2000). The EMOTE model for effort and shape. In Proceedings of the 27th Annual Conference on Computer Graphics and Interactive Techniques (pp. 173–182). New York, NY, USA: ACM Press. doi: 10.1145/344779.352172 Cléry, J., Guipponi, O., Wardak, C., & Ben Hamed, S. (2015). Neuronal bases of peripersonal and extrapersonal spaces, their plasticity and their dynamics: Knowns and unknowns. Neuropsychologia, 70, 313–326. doi: 10.1016/j.neuropsychologia.2014.10.022 Crane, E. A., & Gross, M. M. (2013). Effort-shape characteristics of emotion-related body movement. Journal of Nonverbal Behavior, 37(2), 91–105. doi: 10.1007/s10919-013-0144-2 Cross, E. S., Kirsch, L., Ticini, L. F., & Schütz-Bosbach, S. (2011). The impact of aesthetic evaluation and physical ability on dance perception. Frontiers in Human Neuroscience, 5(9), 1–10. doi: 10.3389/fnhum.2011.00102 Body, Space, and Emotion: A Perceptual Study 55 Dael, N., Goudbeek, M., & Scherer, K. R. (2013). Perceived gesture dynamics in nonverbal expression of emotion. Perception, 42(6), 642–657. doi: 10.1068/p7364 Davidson, J. W. (1993). Visual perception of performance manner in the movements of solo musicians. Psychology of Music, 21, 103–113. doi: 10.1177/030573569302100201 de Gelder, B. (2006). Towards the neurobiology of emotional body language. Nature Reviews Neuroscience, 7(3), 242–249. doi: 10.1038/nrn1872 de Gelder, B. (2013). From body perception to action preparation: A distributed neural system for viewing bodily expressions of emotion. In K. Johnson & M. Shiffrar (Eds.), People watching: Social, perceptual, and neurophysiological studies of body perception (pp. 350–368). New York, NY, USA: Oxford University Press. Dittrich, W. H., Troscianko, T., Lea, S. E. G., & Morgan, D. (1996). Perception of emotion from dynamic pointlight displays represented in dance. Perception, 25(6), 727–738. doi: 10.1068/p250727 Ekman, P., & Friesen, W. V. (1974). Nonverbal behavior and psychopathology. In R. J. Friedman & H. M. Katz (Eds.), The psychology of depression: Contemporary theory and research (pp. 203–224). New York, NY, USA: Wiley. Ekman, P., & Friesen, W. V. (1978). Facial action coding system: A technique for the measurement of facial movement. San Francisco, CA, USA: Consulting Psychologists Press. Glowinski, D., Camurri, A., Chiorri, C., Mazzarino, B., & Volpe, G. (2007). Validation of an algorithm for segmentation of full-body movement sequences by perception: A pilot experiment. In Proceedings of the International Gesture Workshop (pp. 239–244). Berlin, Germany: Springer. Glowinski, D., Dael, N., Camurri, A., Volpe, G., Mortillaro, M., & Scherer, K. (2011). Toward a minimal representation of affective gestures. IEEE Transactions on Affective Computing, 2(2), 106–118. doi: 10.1109/T-AFFC.2011.7 Glowinski, D., Shirole, K., Torres-Eliard, K., Grandjean, D., Riolfo, A., & Chiorri, C. (2014). Is he playing solo or within an ensemble? How the context, visual information, and expertise may impact upon the perception of musical expressivity. Perception, 43(8), 825–828. doi: 10.1068/p7787 Gunes, H., & Piccardi, M. (2009). Automatic temporal segment detection and affect recognition from face and body display. IEEE Transactions on Systems, Man, and Cybernetics Part B, 39(1), 64–84. doi: 10.1109/TSMCB.2008.927269 Hill, H., & Pollick, F. E. (2000). Exaggerating temporal differences enhances recognition of individuals from point light displays. Psychological Science, 11(3), 223–228. doi: 10.1111/1467-9280.00245 Juslin, P. N., & Laukka, P. (2003). Communication of emotions in vocal expression and music performance: Different channels, same code? Psychological Bulletin, 129(5), 770–814. doi:10.1037/0033-2909.129.5.770 Kapadia, M., Chiang, I., Thomas, T., Badler, N. I., & Kider, J. T. (2013). Efficient motion retrieval in large motion databases. In Proceedings of the ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games (pp. 19–28). Orlando, FL, USA: ACM Press. Kapur, A., Kapur, A., Virji-Babul, N., Tzanetakis, G., & Driessen, P. F. (2005). Gesture-based affective computing on motion capture data. In J. Tao, T. Tan, & R.W. Picard (Eds.), Affective computing and intelligent interaction (pp. 1–7). Berlin, Germany: Springer. Laban, R. (1966). The language of movement: A guidebook to choreutics. Boston, MA, USA: Plays, Inc. Laban, R., & Lawrence, F. (1947). Effort. London, UK: Macdonald and Evans. Laban, R., & Ullmann, L. (1971). The mastery of movement. Boston, MA, USA: Plays, Inc. Levy, J. A., & Duke, M. P. (2003). The use of Laban movement analysis in the study of personality, emotional state and movement style: An exploratory investigation of the veridicality of “body language.” Individual Differences Research, 1(1), 39–63. Mancas, M., Glowinski, D., Volpe, G., Coletta, P., & Camurri, A. (2010). Gesture saliency: A context-aware analysis. Gesture in Embodied Communication and Human-Computer Interaction, 5934, 146–157. doi: 10.1007/978-3-642-12553-9_13 Glowinski, Coll, Baron, Sanchez, Schaerlaeken, & Grandjean 56 Meijer, M. (1989). The contribution of general features of body movement to the attribution of emotions. Journal of Nonverbal Behavior, 13(4), 247–268. doi: 10.1007/BF00990296 Nummenmaa, L., Glerean, E., Hari, R., & Hietanen, J. K. (2014). Bodily maps of emotions. Proceedings of the National Academy of Sciences, 111(2), 646–651. doi: 10.1073/pnas.1321664111 Okajima, S., Wakayama, Y., & Okada, Y. (2012). Human motion retrieval system based on LMA features using interactive evolutionary computation method. In T. Watanabe & L. C. Jain (Eds.), Innovations in intelligent machines 2: Intelligent paradigms and applications (pp. 117–130). Berlin, Germany: Springer. Palinkas, L. A. (2001). Psychosocial issues in long-term space flight: Overview. Gravitational and Space Biology Bulletin, 14(2), 25–33. Pantic, M., Pentland, A., Nijholt, A., & Huang, T. S. (2007). Human computing and machine understanding of human behavior: A survey. In T. Huang, A. Nijholt, M. Pantic, & A. Pentland (Eds.), Artificial intelligence for human computing (pp. 47–71). Berlin, Germany: Springer. Pollick, F. E., Paterson, H. M., Bruderlin, A., & Sanford, A. J. (2001). Perceiving affect from arm movement. Cognition, 82(2), 51–61. doi: 10.1016/S0010-0277(01)00147-0 Russell, J. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39, 1161–1178. doi: 10.1037/h0077714 Serino, A., Annella, L., & Avenanti, A. (2009). Motor properties of peripersonal space in humans. PLoS ONE, 4(8), e6582. doi: 10.1371/journal.pone.0006582 Shawn, T. (1963). Every little movement: A book about François Delsarte, the man and his philosophy, his science and applied aesthetics, the application of this science to the art of the dance, the influence of Delsarte on American dance. New York, NY, USA: Dance Horizons. Shikanai, N., Sawada, M., & Ishii, M. (2013). Development of the movements impressions emotions model: Evaluation of movements and impressions related to the perception of emotions in dance. Journal of Nonverbal Behavior, 37(2), 107–121. doi: 10.1007/s10919-013-0148-y Sutil, N. (2013). Rudolf Laban and topological movement: A videographic analysis. Space and Culture, 16(2), 173–193. doi: 10.1177/1206331213475776 Taffou, M., & Viaud-Delmon, I. (2014). Cynophobic fear adaptively extends peri-personal space. Frontiers in Psychiatry, 5, 122–128. doi:10.3389/fpsyt.2014.00122 Todorov, A., Baron, S. G., & Oosterhof, N. N. (2008). Evaluating face trustworthiness: A model based approach. Social Cognitive and Affective Neuroscience, 3(2), 119–127. doi: 10.1093/scan/nsn009 van Dyck, E., Vansteenkiste, P., Lenoir, M., Lesaffre, M., & Leman, M. (2014). Recognizing induced emotions of happiness and sadness from dance movement. PLoS ONE, 9(2), e89773. doi: 10.1371/journal.pone.0089773 van Paasschen, J., Bacci, F., & Melcher, D. P. (2015). The influence of art expertise and training on emotion and preference ratings for representational and abstract artworks. PLoS ONE, 10(8), e0134241. doi: 10.1371/journal.pone.0134241 Velloso, E., Bulling, A., & Gellersen, H. (2013). AutoBAP: Automatic coding of body action and posture units from wearable sensors. In Proceedings of the Humaine Association Conference on Affective Computing and Intelligent Interaction (pp. 135–140). Los Alamitos, CA, USA: IEEE. doi: 10.1109/ACII.2013.29 Vinciarelli, A., Pantic, M., Heylen, D., Pelachaud, C., Poggi, I., D’Errico, F., & Schröder, M. (2012). Bridging the gap between social animal and unsocial machine: A survey of social signal processing. IEEE Transactions on Affective Computing, 3(1), 69–87. doi: 10.1109/T-AFFC.2011.27 Wakayama, Y., Okajima, S., Takano, S., & Okada, Y. (2010). IEC-based motion retrieval system using Laban movement analysis. In R. Setchi, I. Jordanov, R. J. Howlett, & L. C. Jain (Eds.), Knowledge-based and intelligent information and engineering systems (pp. 251–260). Berlin, Germany: Springer. Wallbott, H. G. (1998). Bodily expression of emotion. European Journal of Social Psychology, 28(11), 879– 896. doi: 10.1002/(SICI)1099-0992(1998110)28:6<879::AID-EJSP901>3.0.CO;2-W