香京julia种子在线播放

We tend to feel greater empathy toward familiar individuals or individuals whom we perceive to be similar to us. This response is part of a general favorability toward similar others (Decety, 2010). This effect causes what can be called enhanced empathy toward ingroup members, due to positive biases toward them (Mathur et al., 2010). Conversely, it can lead individuals to behave unfairly toward outgroup members by comparison (Decety and Cowell, 2015).

The separation between “us” and “them” is not static and can be very subtle. On the one hand, group affiliation and ingroup empathy can be predicted by race (Chiao and Mathur, 2010). Phenotypes—gender, age, skin color—are clearly known to create intergroup barriers. On the other hand, environmental or social factors such as a mixed race group may predict social biases (Van Bavel and Cunningham, 2009) and empathic responses (Weisbuch and Ambady, 2008). For example, individuals from different races may present ingroup responses for others from different races but from the same basketball team (Weisbuch and Ambady, 2008). A classic behavioral experiment showed that football fans primed with the idea of interteam competition were less likely to help an individual in need if he wore a t-shirt of a rival team. Conversely, when primed by the idea of a uniting passion for football, they were more likely to engage in altruistic behavior, even for rival supporters (Levine et al., 2005). Another bonding factor (also used as a strategy to overcome biases) is to consider the observed individual separated from the group, humanizing them. In that way, the observer may present an empathic response to one observed individual, but not to his/her group (Kubota et al., 2012).

Explicit or implicit negative evaluation of an emoter considered an outgroup member can decrease or completely block any empathic response in the observer what can be caused by different constructs, such as negative stereotypes, negative biases and anxiety. These mechanisms can have a very high-speed automatic process that demands low cognitive resources whether the individual is aware or not (Amodio, 2009). Subjects with implicit, but not explicit, bias against outgroup members present reduced sensorimotor resonance to vicarious pain targeted to outgroup members (Avenanti et al., 2010). Stereotypes, biases and anxiety may block one’s empathic responses even among individuals with egalitarian values (Correll et al., 2002; Decety, 2010). Categorization of social groups, activation of group stereotypes and use of those stereotypes to form impressions of others is a common practice adopted by social perceivers that can vary in terms of chronic and situational, as well as cognitive and motivational factors, augmenting or reducing stereotypes to form a judgment (Quinn et al., 2003). This categorical thinking based on coded predictions saves cognitive resources on real life interactions, that are usually complex (Macrae and Bodenhausen, 2000). That means that our predictions are shaped by simplified associations present in our long-term memory based on stereotypes, fears, and personal experience (Macrae and Bodenhausen, 2000). Regarding empathy, coded predictions are deeply related to stereotypes (semantic associations) and to biases (visceral categorizations), that can be either negative (when related to outgroups) or positive (when related to ingroup members), respectively, blocking or enhancing one’s empathic response toward another (Amodio, 2009). Intergroup interactions are also modulated by the relation of power (resources) and prestige between groups (Fiske et al., 2016) and the perception of competition between groups may also intensify intergroup biases (Esses et al., 2001).

We refer to self-awareness as a combination of different phenomena related to the perception of one’s own emotions, body, and semantic constructs of themselves. First, the awareness of the external origin of one’s own vicarious emotions is what differentiates empathy from emotion contagion (de Vignemont and Singer, 2006). Furthermore, individuals vary in their ability to perceive and identify their own emotions and sensory states. For example, alexithymia, has been found to reduce empathic responses (Bird et al., 2010; Bernhardt and Singer, 2012). Second, the plasticity of the perception of the bodily and conceptual self may interact with other modulators such as affiliation, similarity and familiarity enhancing empathy by letting observers to feel more identified with emoters (for a complete review on plasticity of self-perception see Farmer and Maister, 2017).

Emotion regulation can act as a key factor to allow individuals to self-regulate their own stress, in order to direct affective empathy into empathic concern instead of personal distress (Decety, 2010; McCall and Singer, 2013). Emotion-regulation is a complex top-down process that allows individuals to initiate, inhibit or modulate their own emotional state or behavior in response to a given situation. Dispositional differences in the abilities of individuals to regulate their emotions have been shown to relate to differences in the tendency of experiencing empathic concern or distress (Rothbart et al., 1994). For example, children with a greater ability to focus or shift their attention have been found more able to present compassion responses; by modulating their negative vicarious emotions, children can keep their emotional arousal at a moderate level, thereby controlling their distress (Eisenberg and Eggum, 2009).

Implicit biases are likely to emerge even among individuals with egalitarian goals (Devine, 1989; Monteith, 1993; Greenwald et al., 1998; Devine et al., 2002; Amodio et al., 2003; Cunningham et al., 2004). Analysis of the self-regulation of stereotyping suggests that, in these cases, individuals experience guilt and redirect their behavior (Monteith, 1993; Czopp et al., 2006). Furthermore, the motivation to be consistent with egalitarian goals (Bargh et al., 2001; Moskowitz and Ignarri, 2009) can help individuals control non-empathic behaviors. Nevertheless, a great deal of research has shown that successfully suppressing unwanted thoughts or emotions is exceedingly difficult (Wegner and Erber, 1992; Gross and Levenson, 1993; Neil Macrae et al., 1994; Wegner, 1994; Monteith et al., 1998). The same likely holds for the self-regulation of behavior (e.g., Monteith, 1993; Monteith et al., 2002; Monteith and Mark, 2005). These self-regulatory mechanisms are based on two different components: monitoring and operating (e.g., Wegner, 1994). While monitoring operates in a relatively automatic manner and does not require deliberative thinking (Amodio et al., 2004), operating processes to control behavior require high motivation, attention and sufficient cognitive resources, and may not occur in complex situations, distraction, or cognitive load (Gilbert and Hixon, 1991; Spencer et al., 1998). Such situations may lead to a reduction in controlled empathic responses.

Empathic or non-empathic processes can also be defined by the quality of the interaction between emoter and observer (Main et al., 2017). On the one hand, to be properly recognized by the observer, the stimuli must be clear. That means that emoters with good communication skills and expressivity are more likely to trigger empathic responses (Greenson, 1960; Ickes et al., 1997; Halpern, 2001; Hollan, 2008, 2012). On the other hand, from the perspective of the observer, present attention and openness to understand the empathic stimuli are also needed (Main et al., 2017). Real time tuning to the mental states of others is needed to generate an accurate empathy response.

Even when feeling empathy, individuals may not express empathic behavior such as altruism. Altruistic motives (other-oriented rather than self-oriented motivation) are related to the amount of help one individual may offer to someone in need. Helping behavior is also related to the skills and abilities of individuals facing helping-tasks, as well as the power to offer effectiveness help (Clary and Orenstein, 1991). For a review on different theories of altruism see Feigin et al., 2014.

Table 1 summarizes highlighted empathic processes, abilities, and modulators discussed in section (A).

Research shows that many students lack social–emotional skills and become disengaged as they progress through school, which interferes with academic performance, behavior, and health (Klem and Connell, 2004). In one study from 2006 with 148,189 American students from sixth to twelfth grade, between only 29 and 45% reported having social competencies such as empathy, decision making, and conflict resolution skills; 71% indicated that their school did not provide an encouraging environment (Durlak et al., 2011). To address these topics, the process of SEL was created to enable learners to identify and manage their emotions, motivations, decisions, and social relations (Elias et al., 1997) through self-awareness, self-management, social awareness, relationship skills, and responsible decision making (Collaborative for Academic, Social and Emotional Learning, 2005—casel.org). These skills relate to some of the empathic phenomena discussed in section (A), specifically regarding perspective taking, empathic accuracy, and emotion regulation. A systematic and well implemented SEL program can allow students to learn, model, and practice these skills and apply them to diverse situations such that they become a part of their repertoire of behaviors (Ladd and Mize, 1983; Weissberg et al., 1989). A meta-analysis of SEL practices (Durlak et al., 2011; Taylor et al., 2017) showed that the conceptual model of targeting various social and emotional assets can be associated with significant improvement in students’ social and emotional skills, as well as academic performance and less risky behaviors. These empirical findings are in line with the educational literature on how intrapersonal and interpersonal competencies—such as self-regulation, problem solving, and relationship skills—may enhance academic performance and prosocial behavior of students. To have an optimal impact and to offer conditions for students to improve learning, behavior and wellbeing, SEL approaches use two great catalysts: long-term training, and a safe environment in a “whole school approach.” The latter engages all members of the school community (management, staff, students, parents, broader community) to work together and to create a safe environment through a sense of belonging and cohesion (Durlak et al., 2011).

Constructivism combines a series of approaches and methodologies developed to empower students by helping them to learn more than just content, but also to learn how to learn, and to develop cognitively, socially and emotionally (Karagiorgi and Symeou, 2005). It has been shown that constructivist learning environments enhance student’s emotional and social abilities, such as self-regulation and perspective taking (Karagiorgi and Symeou, 2005). Some of its strategies are specifically interesting for training perspective taking by offering realistic and plausible stimuli for understanding another’s point of view. These representations of reality avoid oversimplification by representing the natural complexity of the world. They present authentic tasks that contextualize rather than providing abstract instruction. They furthermore provide real world, case-based learning environments, rather than predetermined instructional sequences. Altogether, they enable context, and content, dependent knowledge construction (Jonassen, 1994). Constructivism approaches regularly involve cooperative tasks. Educators play the role of facilitators engaged in dialog with students, rather than monologs, which is especially interesting for instrumentalization of reflexivity, letting learners find their own answers, and leading with their misconceptions (Karagiorgi and Symeou, 2005).

To explore the interaction between observer and targeted outgroups can be an effective way to overcome fear and stereotypes. Intergroup contact (Allport, 1955) can improve positive intergroup attitudes when focused on equal status, cooperation, and common goals (Tropp and Page-Gould, 2015). Affective connection to outgroups (liking one outgroup individual) can decrease prejudice by stimulating perceptions of familiarity (Zajonc, 1968, 1980). Friendship with outgroup individuals may also reduce prejudice through prosocial contact (Allport, 1955; Collins and Ashmore, 1970). For example, one quasiexperimental study combined content and intergroup interaction (Rudman et al., 2001) in a 14-week conflict seminar held by an African American male professor. The intervention reduced prejudice and stereotyping among participants, a pattern that was mediated by cognitive factors (the content of the seminar) but also by affective experiences (liking the professor). Similarly, by combining intergroup interaction with perspective taking training (Malhotra and Liyanage, 2005), a 4-day intergroup workshop conducted between Sri Lankan Singhalese and Tamils resulted in enhanced empathy toward outgroup members 1 year after the intervention. Although intergroup interaction may be effective when it leads to appraisal, this approach presents some strong limitations. In real life, individuals tend to interact within their own social group, missing the opportunity to accumulate personal experiences via social interactions with outgroup members. Moreover, prejudiced people avoid intergroup contact (Pettigrew, 1998). In fact, obligatory courses in diversity have been reported to enhance racial bias in comparison to control students (Bigler, 1999). After enforced multicultural training, individuals with high external motivation but low internal motivation actually responded in anger to behavioral measures targeting outgroups (Plant and Devine, 1998). Furthermore, adult interventions based on the contact hypothesis (Allport, 1955) rarely diminish bias against outgroups in general, but only improve responses to outgroup members present in these interventions (Hewstone, 1996). Moreover, interventions based in “color-blind” strategies (encouraging individuals to suppress their category-based stereotypes in favor to more personalized judgments) have been shown ineffective (Wolsko et al., 2000), and actually enhance negative bias showing a backfire effect (Schofield, 1986; Wegner, 1994). Together these data suggest that although intergroup interactions may reduce bias, interventions along these lines must account for the motivations of the individual.

Table 2 summarizes the methodologies for educational contexts reviewed in this section, highlighting respective strategies on promoting catalysts factors and natural abilities.

Playing the role of a movie character, such as Superman, can be one of the most effective types of play for developing perspective taking (Whitebread et al., 2012). As previously discussed, perspective taking is fundamental for understanding accurately the point of view of another person. Accordingly, perspective taking of an outgroup individual decreases explicit and implicit stereotypes toward the individual and increases positive evaluations toward their group (Galinsky and Mussweiler, 2001). Kwon and Yawkey (2000) also show interesting relations between emotional development and role-playing. Using basic foundations of psychoanalytical and learning theories they discuss how emotional development can be enhanced through role-playing tasks that enhance skills such as interactive levels of expression, control and modeling of emotion, and emotional intelligence. Role playing has also been used in digital games focused on empathy. Real Lives allows players to inhabit the lives of individuals around the world. In a quasiexperimental study with high school students in three schools in the USA (Bachen et al., 2012), students who played Real Lives as part of their curriculum, expressed more global empathy (observed in their identification with the characters played) and greater interest in learning about other countries.

Besides education, role playing techniques have been used in therapeutic contexts, conflict mediation, restorative justice, and many other fields. In each of these fields, different practices are proposed to help participants visualize events and conflicts from the perspective of others. These practices use physical dynamics (e.g., changing seats with another participant) and narratives related to real life (e.g., reporting on conflict from the protagonist’s point of view) to provide multisensory experiences. Successful perspective taking tasks tend to involve more immersive techniques, such as writing an essay about the other’s perspective (Todd and Burgmer, 2013) or taking the role of an outgroup member in a computer game (Gutierrez et al., 2014).

In recent years, mindfulness practices have been getting more attention due to putative therapeutic benefits for depression, anxiety, and chronic pain (e.g., Baer, 2003; Grossman et al., 2004; Galante et al., 2014) and even burnout in the workplace (Krasner et al., 2009). The term mindfulness has been used to describe states, traits, psychological functions, cognitive processes, and different types of meditation practices or intervention programs (Vago and Silbersweig, 2012). In this article, we use mindfulness to describe secular methods of mental training such as Mindfulness-Based Stress Reduction (Kabat-Zinn, 1982), Mindfulness-Based Cognitive Therapy (Segal et al., 2002), and Compassion Cultivation Training (Jazaieri et al., 2013). Each of these methods are driven by different goals, and the article will address specific outcomes relating to empathy-related phenomena, in enhancing natural abilities such as perspective taking and compassion (Klimecki et al., 2013; Hildebrandt et al., 2017) and to address modulators of empathy such as anxiety control and non-judgmental thinking (Lueke and Gibson, 2014). Mindfulness practices often help the subject to focus their attention to their own breath (Bishop et al., 2004), suggesting that interoceptive awareness may be used as an effective strategy in developing empathy-related abilities. Another common approach in many mindfulness practices is to observe thoughts without suppressing them (Bishop et al., 2004). Mindfulness practices enable individuals to focus their attention on automatic cognitions, such as implicit race biases, and can therefore modulate explicit social judgments and behaviors (Payne, 2005). Many mindfulness approaches consist in daily practices that are intended to generate positive outcomes after months of practice (Hildebrandt et al., 2017), but some studies have shown significant benefits after one single intervention (Klimecki et al., 2013; Lueke and Gibson, 2014). Using implicit association tests (Greenwald et al., 1998), Lueke and Gibson (2014) showed that one single 10 minute intervention of mindfulness—aiming to let subjects simply observe thoughts and events in a nonjudgmental way—helped individuals reduce biases against outgroups. Mindfulness practices to develop kindness and compassion have also been associated with lower levels of implicit bias (Kang et al., 2014) and to enhancing altruistic motivations (Condon et al., 2013). Different mindfulness practices appear to have different effects in these domains. Hildebrandt and colleagues (Hildebrandt et al., 2017) analyzed self-reported effects of mindfulness training comparing different types of interventions, focusing, respectively, on present attention, perspective taking, and compassion motivated practices. Results showed that the present attention training was able to significantly increase ratings in self-reported mindfulness, but not improve reports of perspective taking and compassion. Conversely, interventions focusing on perspective taking and compassion enhanced subjects self-reported ratings in all three domains. Specifically, the compassion motivated practice revealed the broadest effects, leading to enhanced abilities of present attention, perspective taking, compassion, and self-compassion, showing the great potential of cultivating compassion mind states to enhance empathy-related abilities. Another study by Klimecki et al. (2013) tested the behavioral and neural effects of one single 6-h training session on compassion-based therapy. The results showed an increase in positive affect, even in response to others’ suffering. The phenomenon was observed in brain areas related to positive evaluation (Kringelbach and Berridge, 2009), love (Bartels and Zeki, 2000, 2004; Aron et al., 2005) and affiliation (Vrticka et al., 2008; Strathearn et al., 2009). Taken together, these studies reveal how mindfulness practices can have powerful effects in training empathy-related abilities.

Even when biases and stereotypes initially preempt one’s empathic responses, it is possible to upregulate empathy to adopt behaviors in line with internal and social values. At least three different strategies can be implemented in these cases (for review on the strategies listed see Kubota et al., 2012):

Mental-scanning of immediate responses: self-awareness of non-empathic responses (biases, stereotypes, anxiety) does not require high cognitive resources and can be learned and practiced. Recognizing these responses is the first step in controlling non-empathic expressions. When non-empathic responses are not identified, it is impossible to control them.

Different studies have explored priming methods to implement internal and social goals in order to overcome negative responses to outgroups, changing the attitudes and perceptions of outgroup members. Three interesting approaches are listed below:

Exposure to non-stereotypical associations (Kawakami et al., 2005).

Although these methods were found to be efficient in lab conditions, they may not be feasible or effect in everyday life. For example, they may not be effective when the observer is repeatedly exposed to stereotypical and biased information in their everyday environments.

Table 3 summarizes training methods discussed, highlighting the environmental context in which they can be applied, and the natural abilities involved.

While the concept of “immersion” refers to the physical nature of a system, presence is its subjective correlative (Slater and Sanchez-Vives, 2016). The term presence has been used to convey many alternative meanings (Slater, 2009). In real life, “presence” is the state of being present (Hildebrandt et al., 2017) or is the state of existing in the world and, fundamentally, to have a body. In VR, the term presence is not necessarily related to having a body, but as the feeling of “being there” (Held and Durlach, 1992; Sheridan, 1992). This phenomenon has been referred to by scientists such as Mel Slater as “PI” in order to distinguish it from different concepts and is defined by “the strong illusion of being in a place in spite of the sure knowledge that you are not there.” Slater defines Psi as a different concept generally associated with presence. Psi stands for the illusion that the environment exhibited in VR is actually taking place. While PI is constrained by sensorimotor contingencies of the VR system, Psi relates to the credibility of the scenario. In both cases, users know that they are not “there” and that the events are not happening, but they feel as if they are, leading them to adopt behaviors as if they were really inhabiting the virtual environment (Slater and Sanchez-Vives, 2016). The interrelation of presence, engagement, and empathy has been observed in immersive VR experiences that teletransport the user to the environment of one emoter (Schutte and Stilinović, 2017).

Immersive EVR or immersive VR with body ownership illusions (Maselli and Slater, 2013) refers to an adaptation of the technique of the Rubber Hand Illusion (Botvinick and Cohen, 1998) to create full body illusions in VR (Petkova and Ehrsson, 2008; Maselli and Slater, 2013). Using VR, researchers apply multisensory and motor stimuli in synchronicity with the first-person perspective of an avatar—using computer generated imaging (Maselli and Slater, 2013), or the image of real humans through stereoscopic video (Petkova and Ehrsson, 2008). In these studies, the evidence shows that subjects feel that they have swapped bodies with another person (Petkova and Ehrsson, 2008), a plastic mannequin (Petkova and Ehrsson, 2008), a Barbie doll (Van der Hoort et al., 2011), a digital avatar (Maselli and Slater, 2013), an invisible body (Guterstam et al., 2015), and even a body located in the front of them (Lenggenhager et al., 2007). These multisensory stimuli elicit a blurriness in the identity perception of self and other (Paladino et al., 2010) and may even drive participants to present a subjective anxiety to threats targeted at their virtual hand (Zhang and Hommel, 2016). Body Ownership, or the sense of embodiment, is comprised of the sense of self-location, the sense of agency and the sense of body ownership (Kilteni et al., 2012).

The most explored stimuli for inducing embodiment are visuomotor synchronicity, seeing oneself in the body of an avatar that mimics one’s movement in real time, and visuotactile synchronicity, seeing tactile stimuli applied to the avatar at the same time that it is applied to the hidden body part of the user (Maselli and Slater, 2013) with the avatar in a congruent posture with the subject (Tsakiris et al., 2007). Visuomotor synchronicity can be applied only to movements of the head, or also to movements of the whole body (Maselli and Slater, 2013), and visuotactile synchronicity can be passive (e.g., being touched) or active (e.g., touching an object) (Tajadura-Jiménez et al., 2013). Maselli and Slater (2013) have shown that a proper combination of stimuli to promote strong embodiment illusions that includes realistic images and wide field of view may not require visuomotor or visuotactile stimulation. In fact, incongruent perceptual cues may not break the embodiment during strong illusions. In one experiment, they showed that an avatar with realistic skin tone placed congruently to the user in immersive VR does not require visuomotor or visuotactile synchronicity to produce the embodiment. Moreover, full body visuomotor synchronicity using the image of an avatar with realistic skin tone can induce Body Ownership Illusion even under asynchronous visuotactile stimulation, which does not occur when the image of the avatar has a non-realistic skin tone (Maselli and Slater, 2013). Other combinations of stimuli, such as congruent full body first-person perspective and visuotactile synchronicity, can also be sufficient to create strong embodiment illusions, even with non-realistic human images (mannequins) and no head movements (Petkova and Ehrsson, 2008).

As well as visuomotor and visuotactile synchronicity, congruent first-person images, and realistic images, there are other variables that may induce or enhance manipulations in the perception of the body. Sound has been shown to alter the perception of the body, even without the use of VR. For example, altering the sound feedback when touching objects may alter the perception of the arm length (Tajadura-Jiménez et al., 2012, 2015a) and/or its strength (Tajadura-Jiménez et al., 2017). Similarly, manipulation of the sound feedback of a hammer hitting the user’s virtual hand can make the participants hand feel stiffer and heavier (Senna et al., 2014). Furthermore, manipulations of the sound of someone’s steps may also alter the perception of one’s own body weight and change the pattern of their gait (Tajadura-Jiménez et al., 2015b). Sound manipulation techniques have also been applied to EVR by changing the frequency of the user’s voice feedback to become more similar to the avatar (childlike) causing changes regarding the user’s voice recalibration toward the auditory feedback (Tajadura-Jiménez et al., 2017). These experiments show the potential of sound manipulation to enhance EVR experiences. Manipulations of interoceptive signals can also modulate embodied illusions. Evidence suggests that feedback of biosignals such as a heartbeat may enhance embodiment illusions (Suzuki et al., 2013). Synchronous cardiovisual signals increased self-identification and self-location in relation to the subject’s virtual body, shifting their perception of touch toward the virtual body (Aspell et al., 2013).

Recent theories of body cognition offer an interesting perspective on the potential nature of these processes. Tsakiris (2017) bases his model on extensive reviews and experiments, suggesting that one’s perception about one’s own body combines the interaction of bottom up brain phenomena (from the body to the brain) and top down processes (from the brain to the body). In this model, real time information of interoceptive states (such as proprioception, breathing, heart rate, and arousal) and real time information of exteroceptive sensations (such as vision, touch, and taste) inform the brain’s predictions of the perception of the body. The regulation between internal bodily states, external environment information, and mental concepts give us the sense of ourselves and the space that surrounds us. Neuroimaging research suggests that a significant prediction error is required to update the predictive internal models of the body matrix (O’Reilly et al., 2013; Riva et al., 2017).

Table 4 summarizes highlighted concepts related to Body Ownership Illusion.

Although correlated, evidence suggests that Agency and Body Ownership are two different phenomena (Sato and Yasuda, 2005) that can occur in different circumstances. Agency requires voluntary action, while body ownership may occur under both voluntary action and passive events (Tsakiris et al., 2007). The subjective perception of agency over a body part is different from the subjective perception of agency over a physical action, that by being voluntary, involves a combination of efferent (top down) and afferent (bottom up) information (Tsakiris et al., 2007). The rubber hand illusion with visuotactile stimulation is a classic example of Body Ownership Illusion without agency (Botvinick and Cohen, 1998), that actually involves a subjective perception of agency, with affirmative agreements with the sentences “it seemed like I could have moved the rubber hand if I had wanted” and “it seemed like I was in control of the rubber hand” in self-reported questionnaires used to measure the illusion. But this subjective perception of agency decreased, for example, with visuomotor delay, without changing the sense of ownership (Kalckert and Ehrsson, 2012). Even though it is not necessary for inducing Body Ownership Illusions, agency may contribute to the extent that this illusion is felt. Using proprioceptive drift, the perception that the real hand is closer to the displaced rubber hand, Tsakiris et al. (2006) suggest that voluntary movements of body parts induce a global change in proprioceptive awareness. While localized proprioceptive drifts were found in passive stimulation, during active movement of one digit the proprioceptive drift was observed in the whole hand.

More recently, different studies have used the Body Ownership Illusion with visuomotor synchronicity and voluntary movement to induce an illusion of agency over the user actions such as walking (Kokkinara et al., 2016) or speaking (Banakou and Slater, 2014). By embodying a digital avatar that could be controlled by the user’s movements, researchers observed self-attribution of agency to subjects over actions taken by the avatar, even without any prior intention, prediction, priming, and cause preceding effect (Banakou and Slater, 2014). In this experiment, the digital avatar would speak independently of the user’s action creating not only the perception that subjects were themselves talking, but also changing the fundamental frequency of the user’s voice after the experience. This illusion was found to be even stronger when a vibration stimulus was applied to the user’s throat in synchronicity with the avatar’s voice.

As with the model of body cognition discussed in Section “Embodied VR or Full Body Ownership Illusion: Feeling That You Have a Different Body, with Different Traits,” several theories define the sense of agency as a result of the comparison between prediction of efferent and afferent information. For a cognitive and neural perspective of the sense of agency see the review of David et al. (2008). These concepts were implemented in an experiment in which researchers were able to induce the illusion of walking in subjects who were actually seated (Kokkinara et al., 2016) through a combination of priming and body ownership illusion. In this experiment, subjects could see themselves walking while perceiving an optic flow in the environment and a sway movement of the head due to the walking motion. Subjects presented high levels of body ownership in self-report questionnaires and based on physiological data.

Taken together, these findings suggest that Agency Illusions could be combined with Body Ownership Illusions to create experiences in which the avatars perform prosocial actions that could be perceived as voluntary actions of the subjects themselves. Even so, this hypothesis is yet to be investigated before being implemented into empathy-related training.

Some experiments have used EVR to allow participants to step into the shoes of outgroup members. Peck et al. (2013) conducted a study in which subjects with light skin could see themselves in a dark-skinned avatar. The manipulation decreased negative implicit associations toward black individuals immediately after the experiment. A similar setting was conducted by Banakou et al. (2016) showing that a decrease of implicit bias was sustained even 1 week after the intervention. In another experiment (Hasler et al., 2017), levels of implicit biases did not change, but the intervention nevertheless increased mimicry between a subject with white skin embodying an avatar with dark skin and another digital character with dark skin.

In one non-experimental setting, de la Peña et al. (2010) used the concepts of multisensory perspective to allow participants to step into the shoes of a character confined in a Guantanamo Bay. Visuomotor synchronicity of the user’s head, binaural audio of the environment and haptic feedback of breathing were provided to create the illusion of being in the stress position of the prisoners (a position described in reports on prisoner treatment). Although no scientific data were collected, participants reported feeling anxiety and discomfort and expressed an emotional connection with the situation of the prisoners.

Using a different approach and without immersive interfaces, researchers used a different technique called “enfacement,” that stimulates a mirror touch synesthesia (Fini et al., 2013). Participants see the image of faces of different avatars on screen being stroked by a brush. Mimicking the tactile stimuli provided in the avatar, researchers stroked the user’s face with an identical brush, in synchrony with the image observed. After seeing the image of avatars with different phenotypes, subjects revealed a greater self-identification with more diverse phenotypic characteristics. Using enfacement illusions with EEG measurements, Serino et al. (2015) observed activation of face-specific regions correlating to the increased identification with the avatar’s face.

In a recent experiment without VR (Azevedo et al., 2017a,b), researchers showed that tactile feedback of a slow heartbeat-like rhythm can make subjects more relaxed before performing a stressful task such as speaking in public, showing its potential implications in emotion regulation (an important ability for converting empathic distress into empathic concern). As mentioned in Section “Embodied VR or Full Body Ownership Illusion: Feeling That You Have a Different Body, with Different Traits,” interoceptive signals can interfere in embodiment. It has also been shown that interoceptive signals such as heartbeat can correspond to bias behavior (Azevedo et al., 2017a,b) and that awareness of heartbeat signals relate in significant part to cognitive-affective processing (Dunn et al., 2010). These findings reveal a great potential for integrating biosignal manipulations to help participants to control their anxiety when faced with the stress of others, to interfere in anxiety triggered by an outgroup threat and to interfere in emotional processing such as affective empathy.

In virtual environments, digital self-representations (i.e., avatars) may influence users and lead their behaviors to be consistent with the avatar’s appearance. This behavioral modulation, known as the Proteus Effect (Yee and Bailenson, 2007; Yee et al., 2009), has been observed in several studies. In their seminal work, Yee and Bailenson (2007) have shown that attractive avatars lead to a more intimate behavior with a confederate in terms of self-disclosure and interpersonal distance. In a second study, they also observed that tall avatars lead to more confident behavior than short avatars in a negotiation task. More recent studies have also shown that the appearance of the embodied avatars could influence attitudes, beliefs (Fox et al., 2013) and actions of the users (Peña et al., 2009; Guegan et al., 2016). For instance, it has been shown in non-immersive VR environments that the use of an avatar resembling a member of the Ku Klux Klan activates more negative thoughts and leads users to participate more aggressively (e.g., murder, vengeance) in the stories created (Peña et al., 2009). Other studies have shown how the Proteus Effect may enhance stereotypes against outgroup members. It has also been shown that users that play a black avatar in a computer game present greater aggressive cognition and affect (Eastin et al., 2009; Ash, 2016).

From a theoretical point of view, the Proteus effect is based on self-perception principles (Bem, 1972), under which the individual explains his attitudes and internal states based on observation of external cues. In this way, the profile of the avatar could lead the user to make implicit inferences about his/her personal dispositions (e.g., I am an empathic person). The influence of avatars is also compatible with the priming process (Peña et al., 2009), which refers to “the incidental activation of knowledge structures, such as trait concepts and stereotypes, by the current situational context” (Bargh et al., 1996). For instance, perceiving the characteristics of an avatar nurse or a humanitarian worker could activate some related concepts (e.g., altruism, empathy) as well as inhibit more antithetical concepts such as aggression or violence. Moreover, these situational cues may lead to behavioral assimilation via an increase in the likelihood of behaviors congruent with the primed concept. Whatever the underlying mechanism, and to the best of our knowledge, no study has directly investigated the links between the Proteus Effect and the empathic processes. However, previous work shows that the digital self-representations can influence behavior in a pro or antisocial way. For example, the embodiment of an avatar resembling an inventor led engineering students to show higher creative fluency and originality of ideas during a face-to-face brainstorming session conducted after immersion in a virtual environment (Guegan et al., 2016). Another example shows that embodying a casually dressed black avatar enhances users’ performance in playing drums in comparison to embodying a formally dressed white avatar (Kilteni et al., 2013), probably due to the positive stereotypic association of black individuals and rhythm. In another experiment, it was demonstrated that embodying a Sigmund Freud-like avatar talking to a scanned version of their own body can help users improve their mood after self-counseling, in comparison to self-counseling in a self-representing avatar (Osimo et al., 2015). Another case of positive stereotyping showed that embodying a superhero who helps the population of a virtual city led to increased prosocial behavior in an offline interaction (Rosenberg et al., 2013). Conversely, using a villainous avatar (Voldemort) led both to an increase in antisocial behavior and to a significant decrease in prosocial behavior (Yoon and Vargas, 2014).

It has been shown that the Proteus effect is mediated by the level of embodiment felt by users in relation to their avatar (Ash, 2016), suggesting that EVR can enhance this effect. Given this set of findings, one might expect that training methods using avatars designed and pretested to improve empathy would induce beneficial behavioral changes, improve positive perceptions among users, and so on. Another possible implication of this concept is to embody a digital avatar of an outgroup member that presents traits that contradict stereotypes. These hypotheses have yet to be tested.

Table 5 summarizes highlighted strategies that can be applied using VR and Body Ownership Illusion for empathy learning and their potential effects.