Full length article
Social fidelity in virtual agents: Impacts on presence and learning

https://doi.org/10.1016/j.chb.2020.106562Get rights and content

Highlights

  • We suggest that Social Fidelity has two divisions: Functional and Physical.

  • Social Fidelity may impact teaching effectiveness of pedagogical agents in non-social domains.

  • Tradeoffs exist when implementing some forms of Social Fidelity in pedagogical agents.

  • Personalized language, politeness, personality, and gestures in pedagogical agents have had positive learning impacts.

Abstract

Teaching and training are increasingly moving from real world venues to computerized environments, with human instructors often being replaced or joined by virtual pedagogical agents. While system fidelity and immersive properties of virtual learning environments are frequently discussed in the literature, less often addressed is the fidelity of the social components of teaching and their inclusion in pedagogical agent design. Teaching is inherently a social process, making the social fidelity of virtual agents a potential factor affecting learning outcomes. In this paper, we explore the concept of social fidelity as it pertains to the teaching effectiveness of pedagogical agents. We define the term, distinguish two subcategories, and discuss representative examples of research in these domains. Promising avenues for improving learning outcomes with social fidelity include personalized language, politeness, personality, attention, feedback, social memory, and gestures. Key conclusions are that: Social fidelity is important to learning in non-social domains, tradeoffs exist when implementing certain forms of social fidelity, individual user differences need to be more widely considered, and more focused studies are needed to compare different levels of social fidelity to uncover how they impact learning outcomes.

Introduction

Computerized learning environments, including simulations and virtual reality environments, can be found in many diverse and applied domains (e.g., medical training, military training, second language learning). Benefits of virtual environment or computer-aided training are many (Lisetti, 2012), and include the potential for greater flexibility for students and teachers, potential cost savings, more options for personalization, and increased safety and accessibility for certain training scenarios, as compared to real-world training. Reaping these benefits requires careful elucidation of factors that influence learning outcomes in computerized learning environments.

A variety of factors have been shown to affect learning in computerized and virtual environments. These factors range from physical attributes such as the quality of display systems and the fidelity of visual and other sensory items in the environment (Alexander, Brunyé, Sidman, & Weil, 2005; Cummings & Bailenson, 2016; Parong et al., 2020) to functional attributes, such as teaching strategies (e.g., Mayer et al., 2003, Mayer et al., 2003; Moreno & Mayer, 2004). For example, a given virtual learning environment will have a series of attributes, including its exact graphical resolution, frame rate, audio quality, and user interface (e.g., a way to provide feedback to the user), and may be played on different hardware (such as an immersive virtual reality (VR) headset). Physical and functional hardware, software, and pedagogical attributes can each contribute to the level of presence (the feeling of being there) experienced by the user (Witmer & Singer, 1998) in the virtual learning environment, which may increase engagement, time on task, attention, or other states relevant to learning (Alexander et al., 2005) and may affect learning outcomes (e.g., Parong et al., 2020).

An integral component of many computer-aided learning environments is a virtual pedagogical agent, a simulated character that socially interacts with the user to facilitate learning. The pedagogical agent may play the role of a teacher, coach, student, or peer, and/or the agent may provide emotional support during the learning process. The physical and functional attributes of these characters can influence learning in profound ways. Simply adding a social entity to a computerized environment impacts affective and learning outcomes, a phenomenon labeled the persona effect by Lester et al. (1997).

Simulated social characters imbued with social behaviors are particularly important in the learning context, when learning is the goal of a computerized environment. Human-human teaching is an inherently social act. For thousands of years, humans have been learning from one another in a richly social context. It is reasonable to suppose, then, that the details of the simulated social interaction could play a significant role in affecting learning outcomes and learning-related behaviors.

Here, we focus on socially-relevant attributes of virtual characters and the realism of these social attributes, which we denote as their social fidelity. Much as fidelity is generally defined as the extent to which a virtual environment emulates the real world (Alexander et al., 2005), we define social fidelity as the extent to which the social aspects of a virtual character (e.g., pedagogical agent) emulate social aspects of the real world.

This is somewhat related to the term behavioral realism, however, behavioral realism deals with a wide range of non-social behaviors and characteristics. The term has not been consistently defined, with definitions ranging from realistic behavioral models of targets within a virtual environment (Evertsz et al., 2014), to naturalistic behaviors being displayed by avatars (Herrera, Oh, & Bailenson, 2018), and even a combination of beliefs that the behaviors are consistent with real people and beliefs that an avatar is controlled by a real person (Guadagno, Blascovich, Bailenson, & McCall, 2007). We suggest that social fidelity is distinct as it is focused on how close the social interaction with the virtual character is to a real social interaction. Furthermore, we propose that social fidelity can be divided into two main categories that cover functional attributes of social fidelity (e.g., the realism of the functional social interactions) and physical attributes of social fidelity (e.g., the quality of perceptual elements of the character's social behavior). This parallels the distinction drawn in previous work on general fidelity of virtual environments, including functional (if the simulation acts like the real world) and physical (if the simulation looks like the real world) attributes (Alexander et al., 2005).

In this paper, we focus on the effects of social fidelity on learning outcomes. We consider both learning performance and learning-related behaviors. Learning performance can be measured in a variety of ways, including pre-test/post-test designs, direct tests of skills or knowledge, and the use of transfer tasks which require users to apply their learning in the performance of novel tasks. Learning-related behaviors can also be measured, many of which have been shown to correlate with learning performance. Examples include state measures of motivation and engagement, measures of presence, objective measures of time-on-task, and subjective reports of preferences and affect. Learning-related behaviors are not 100% consistent in their correlation with objective performance measures, so they must be interpreted with caution. For example, a learning intervention that increases engagement is suggestive but not definitive in regards to its effects on performance.

Interest is only growing as computerized technologies, including artificial intelligence algorithms and cutting-edge virtual reality equipment, become ever more powerful and accessible (Hobert & von Wolff, 2019). Computer-aided learning and virtual training environments are becoming an increasingly popular topic of study, leading to several reviews published in the last few years.

Virtual environment presentation properties and virtual reality equipment have been reviewed by Cummings and Bailenson (2016) and Jensen and Konradsen (2018), respectively. Cummings and Bailenson (2016) reviewed studies that related fidelity, and other virtual environment parameters to learning outcomes across a number of different learning tasks-- including navigation, decision making and game-based tasks (Cummings & Bailenson, 2016). In their meta-analysis, Cummings and Bailenson (2016) identified immersive properties of displays and virtual environments, including image quality, field of view, update rate, and sound quality as features contributing to immersion levels of the systems. Their meta-analysis did not focus on pedagogical agents and did not examine social fidelity.

Jensen and Konradsen (2018), reviewed studies of virtual reality hardware for educational virtual environments, and called out the dearth of studies, especially high quality studies, addressing this question in the literature. Using a metric of quality, Jensen and Konradsen (2018) report that studies in this topic area scored relatively low, with many not using randomized samples or not using validated survey instruments. Nonetheless, virtual reality was found useful for spatial, visual, psychomotor, and affective skills training. Their review did not focus on pedagogical agents or social fidelity.

Other reviews have focused on pedagogical agents. Martha and Santoso (2019) review pedagogical agent research published between 2001 and 2017, focusing on the modality, appearance, and roles of agents. Their review does not look at social fidelity or call out specific aspects of successful social fidelity design elements. Importantly, they point out a dearth of significant results in the literature broadly.

A review by Hobert and von Wolff (2019) looked at trends in use domains and target platforms for conversational pedagogical agents (one type of pedagogical agent). In agreement with many authors of pedagogical agent reviews, Hobert and von Wolff (2019) mention that there are few comprehensive evaluations and a lack of uniform processes for guiding research studies in this area.

Jamet, Masson, Jacquet, Stilgenbauer, and Baratgin (2018) focus their review on only one pedagogical agent role – student/protégé – and only on physical robot agents. In their review on learning by teaching, the authors note promising results but do not break down aspects of social fidelity. The review describes a few studies that vary aspects of what we would refer to as social fidelity but the focus is more on learning domains. Importantly, it is noted that experimental paradigms vary widely. This echoes the lack of consistency mentioned by Davis (2018).

Richards and Dignum (2019) performed an important review focused on ethical issues in pedagogical agents, including discussions of privacy, accessibility, and cultural appropriateness. This paper does not review learning effectiveness, but nonetheless it calls out a need for greater focus on individual differences and also learning transfer, a suggestion with which we strongly agree. We echo many authors’ concerns about lack of consistency and rigor in the pedagogical agent literature broadly.

Kim and Baylor (2016) discussed their work examining perceived social and affective support of pedagogical agents for learners. While Kim and Baylor (2016) provided distinctions in types of agent personas (e.g. expert, motivator, mentor), they did not discuss the implications of each of the individual characteristics that was included in each type of persona (e.g., politeness, gestures, etc.). As noted above, a systematic literature review about the design of pedagogical agents (Martha & Santoso, 2019) revealed overall patterns and approaches that have been used to study the impact of pedagogical and virtual agents. However, these works have not specifically focused on the implications of social fidelity, or if there are distinctions between the characteristics of the social interaction, or the representativeness of the social interaction to a real-world interaction.

We are aware of one review that examines an aspect of social fidelity in pedagogical agents and looks at its effects on learning outcomes. A meta-analysis of gesture use in pedagogical agents (Davis, 2018) found that including gestures generally improves transfer and retention, but was based on only 20 experiments. The review makes clear that there is a relative lack of studies that vary gesture parameters, particularly on a fine scale, and quantitatively measure learning outcomes. The lack of consistency in outcome tests is called out as a problem for drawing broad conclusions about how gestures should best be used in agents and points out that gestures need to become the direct focus of future pedagogical agent research.

While many aspects or parameters of computerized learning environments have been recently reviewed, social aspects have received comparatively less attention. Teaching can be seen as an inherently social process that relies on social connection between individuals (Frymier & Houser, 2000; Heyes, 2016). The teacher-student, trainer-trainee, or learner-peer relationship can impact the transmission and retention of knowledge, both in the real world and when virtual or computer-based teachers, trainers, or simulated peers are used (Saerbeck, Schut, Bartnek, & Janse, 2010; Kasap; Magnenat-Thalmann, 2012). However, there has not yet been a concerted effort to call out elements of social fidelity and discuss how they impact learning, nor have types of social fidelity (physical vs. functional) been explicitly categorized. The goal of the current paper is to define types of social fidelity, provide examples of elements that can contribute to social fidelity, and discuss the impacts of these elements on learning outcomes.

In a computer-based context, fidelity can be succinctly described as “the extent to which the virtual environment emulates the real world” (p.4, Alexander et al., 2005). As noted by Alexander et al. (2005), two types of fidelity that are relevant in a virtual environment are: functional (how closely the items match the real functions, and react as they would in the real world) and physical (how close the environment looks to the actual environment). Building on Alexander et al. (2005)'s definition of computer-based fidelity, we define social fidelity in a computer-based context as the extent to which the social aspects of the virtual character emulate social aspects of the real world. We furthermore suggest that there are two subtypes of social fidelity: functional social fidelity, which includes how close the interaction is to a real-world interaction in terms of the content of that interaction, and physical social fidelity, which includes how close the auditory, and communication methods are to a real social interaction. We propose that both of these categories are relevant when considering the development of an educational social interaction with a virtual agent, and that both types of social fidelity may influence learning outcomes.

The term presence generally refers to a user's feeling of actually “being there” in a virtual environment (Witmer & Singer, 1998). The term social presence has been discussed in the literature but generally concerns the level of connectedness a user experiences in online learning environments through the use of email and online message boards for the course (Garrison, 2007; Kehrwald, 2008; Richardson & Swan, 2003); i.e., the term is most often used to denote a feeling derived from mediated social interactions with other humans. However, Heeter (1992) initially described social presence in virtual worlds more broadly as the feeling of existing in the world by interacting with others in that virtual world. When other personas (whether they are other human players or virtual agents) acknowledge the presence of the user and interact appropriately, this can lead to a feeling that the environment itself is more real. Here we build on Heeter's (1992) definition of social presence and on Witmer and Singer's (1998) definition of presence, by describing social presence as the feeling of “being there” socially in the virtual environment or other computerized learning environment, i.e. the feeling that the user is engaging in real social interactions. We suggest that both the functional social fidelity and physical social fidelity of virtual agents can lead to feelings of increased social presence.

Aside from social presence and specific questionnaires to measure it (De Kort; IJsselsteijn, W. A., & Poels, 2007; Lin, 2004), is the general concept of social agency theory, the idea that seeing a pedagogical agent as a social partner should lead a learner to put more effort into understanding the lesson and should ultimately lead to deeper encoding and better learning outcomes (Mayer & DaPra, 2012; Mayer, Sobko, & Mautone, 2003; Moreno, Mayer, Spires, & Lester, 2001). A recent illustration of this phenomenon can be found in the work of Ruan et al. (2019), who used a within-subjects design to compare the use of a dialogue-based pedagogical agent vs. the standard flash card method for learning factual knowledge in several topical domains. When using the pedagogical agent, students invested more time on task, had more correct answers, and had better learning gains in their recall of the learning material.

Having a social agent involved, feeling that the virtual agent is a social entity, and experiencing feelings of social presence, are all factors expected to increase engagement and buy-in in computerized training environments (Moreno et al., 2001; Alexander et al., 2005), which in turn may increase factors like time-on-task (e.g., Ruan et al., 2019), which are strong predictors of knowledge acquisition (Alexander et al., 2005). Social presence would be expected to improve learning outcomes and transfer, potentially via increased engagement and motivation. This was observed by Zhan and Mei (2013), who found a positive relationship between undergraduates’ social presence in an online digital design course and their scores on final exam questions. Similarly, in a study of graduate students taking an online course in education administration, students who experienced higher levels of social presence performed better on the course writing assignment (Picciano, 2002). However, they did not score better on the course exam (Picciano, 2002). The course writing assignment required the use of social skills, such as perspective-taking, while the exam (rote knowledge) did not (Picciano, 2002). In this case, learning in a social task domain (e.g., perspective taking) was enhanced by social presence in the computerized learning environment, but learning in a non-social task domain (rote knowledge) was not. This leads to a key question: are the learning benefits of social presence and social fidelity confined to the learning of social tasks?

It has generally been argued that presence or fidelity in virtual or computerized learning environments may not be helpful if it is outside the domain of, or not directly relevant to, the learning task (Whitelock, Romano, Jelfs, & Brna, 2000; Moreno & Mayer, 2004; Alexander et al., 2005; Krämer & Bente, 2010; Lane, Hays, Core, & Auerbach, 2013; Liew, Tan, & Jayothisa, 2013). However, if teaching and learning are inherently social acts (e.g., Saerbeck, Schut, Bartneck, & Janse, 2010; Kasap & Magnenat-Thalmann, 2012; Richards & Dignum, 2019), it follows that social fidelity is always “in domain,” particularly when pedagogical agents are employed (Lester et al., 1997). Indeed, social fidelity has been found to enhance non-social forms of presence (Moreno & Mayer, 2004), which may improve learning outcomes in non-social domains. Furthermore, if social fidelity increases the feeling of being “with” a virtual teacher, tutor, or co-learning peer, this may increase attention (Moreno & Mayer, 2004) or motivation (Saerbeck et al., 2010) and thus learning outcomes, regardless of the learning domain.

An ever-increasing number of studies have varied social fidelity levels and reported the resulting effects on learning and transfer in non-social domains (e.g., Yadollahi, Johal, Paiva, & Dillenbourg, 2018; Wang, Li, Mayer, & Liu, 2018; de Wit et al., 2018; Müller-Wuttke & Müller, 2019; Ruan et al., 2019; Moon & Ryu, 2020). For example, in a study of language learning in school children (Saerbeck et al., 2010), several aspects of social fidelity were implemented simultaneously by a physically embodied virtual agent (a teaching robot). The high social fidelity condition included non-verbal social behaviors such as head nods, gaze direction, simple emotional empathy (i.e., happy face when the student performed correctly, sad face when incorrect), and simulated extrovert personality. These behaviors were modeled after social behaviors used by human tutors. The low social fidelity implementation lacked these aspects and displayed a more “robotic” personality. The high social fidelity condition improved user motivation as well as scores on a post-test. The results lend support to the idea that the inherently social nature of teaching means that social fidelity can improve learning and transfer of skills that are not themselves strictly within the social domain (here, rote learning of vocabulary). It should be noted, however, that the social fidelity aspects in this study were not varied separately, so the independent contributions of each aspect are unclear—other studies have manipulated individual aspects of social fidelity. To provide more scaffolding for understanding, and to help guide future research, it would be helpful to categorize different aspects and types of social fidelity. Our paper proposes categories of social fidelity and discusses examples of studies and summarizes findings in these areas.

Many computerized learning environments include on-screen characters or virtual agents that are intended for users to interact with. Similarly, virtual characters are sometimes physically embodied (e.g., interactive robots) and placed in the real world to assist learning (e.g., Blanson Henkemans et al., 2013; Kasap; Magnenat-Thalmann, 2012; Saerbeck et al., 2010; Tapus; Tapus, & Mataric, 2008). We suggest that an embodied or unembodied virtual agent has high functional social fidelity if it appropriately reproduces cues, behaviors, and content of speech that are socially realistic in real world human-human interaction. For example, backchannel body movements such as head nods are important for natural turn-taking during human-human conversation and for giving the impression that a listener is paying attention and engaged in the interaction (e.g., von der Pütten; Krämer, & Gratch, 2009). A virtual agent that includes such backchannel behaviors could be considered to have higher functional social fidelity than a virtual agent that does not include these behaviors or that employs them inappropriately. Other relevant social cues and behaviors include appropriate use of facial expressions, emotions, body postures, proximity, gaze direction, word choice, empathy, politeness, personality, and many others. An agent that exhibits relevant social behaviors appropriately (i.e., high functional social fidelity) can increase the feeling of social presence experienced by users (e.g., Moreno & Mayer, 2004; von der Pütten; Krämer, & Gratch, 2009; Lee & Nass, 2005; Kasap; Magnenat-Thalmann, 2012). We suggest that the congruence and socially-appropriate use of such attributes constitutes functional social fidelity (e.g., displaying a smile and a happy vocal tone when encouraging the student), while the accuracy of the sensory implementation of the cues (e.g., the visual accuracy of the smile animation and the acoustic realism of the vocal tone) constitutes physical social fidelity. If a virtual agent has realistic physical characteristics such as sounding like a real person, and exhibiting natural sounding speech, then it has higher physical social fidelity than a virtual agent that does not.

Section snippets

Contributors to functional social fidelity

Research has suggested that there are a number of different potential contributors to functional social fidelity in virtual agents. Functional social fidelity includes the content of the interaction between the virtual agent and learner, and how close the social interaction is to a real instructor-learner (or learner-peer) interaction. Studied elements that contribute to this include personalized language, politeness, attention, feedback, social memory, personality, interactivity, and gestures.

Contributors to physical social fidelity

Physical social fidelity describes how representative an interaction between a virtual agent and a human learner is to a real world interaction in terms of the physical fidelity. Major aspects of physical social fidelity include the acoustic quality or type of speech generation that is used for the interaction (e.g., computer-generated vs. human), and how closely the visual cues of mouth movement match the speech. These can contribute to a learner's feelings of social presence and can influence

Conclusions and future directions

Some aspects of social fidelity seem particularly promising as means of improving learning transfer in both social and non-social task domains. For functional social fidelity, personalized language and politeness have yielded fairly consistent positive effects on learning outcomes and social presence (Moreno & Mayer, 2000, 2004; Ogan et al., 2012; Schneider et al., 2015A). These particular factors are used in many VR agent applications for education, therapy, and other domains that require

CRediT authorship contribution statement

Anne M. Sinatra: Writing - original draft, Conceptualization. Kimberly A. Pollard: Writing - original draft, Conceptualization. Benjamin T. Files: Writing - original draft. Ashley H. Oiknine: Writing - original draft. Mark Ericson: Writing - original draft. Peter Khooshabeh: Writing - original draft, Writing - review & editing.

Declaration of competing interest

None (Note, 5 of the authors are U.S. Government employees).

Acknowledgements

This work was funded by the US Army Research Laboratory's Human Research and Engineering Directorate. The authors would like to acknowledge the entire Immersion Training Effectiveness Team who contributed to research focused on social fidelity. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Army Research Laboratory, Soldier Center, or U.S. Government. The U.S.

References (100)

  • S. Schneider et al.

    Introducing the familiarity mechanism: A unified explanatory approach for the personalization effect and the examination of youth slang in multimedia learning

    Computers in Human Behavior

    (2015)
  • S. Schneider et al.

    Mind your Ps and Qs! How polite instructions affect learning with multimedia

    Computers in Human Behavior

    (2015)
  • C. Stevens et al.

    On-Line experimental methods to evaluate text-to-speech (TTS) synthesis: Effects of voice gender and signal quality on intelligibility, naturalness and preference

    Computer Speech & Language

    (2005)
  • A. Thangthai et al.

    Synthesising visual speech using dynamic visemes and deep learning architectures

    Computer Speech & Language

    (2019)
  • N. Wang et al.

    The politeness effect: Pedagogical agents and learning outcomes

    International Journal of Human-Computer Studies

    (2008)
  • Z. Zhan et al.

    Academic self-concept and social presence in face-to-face and online learning: Perceptions and effects on students' learning achievement and satisfaction across environments

    Computers & Education

    (2013)
  • A.L. Alexander et al.

    From gaming to training: A review of studies on fidelity, immersion, presence, and buy-in and their effects on transfer in pc-based simulations and games

    DARWARS Training Impact Group

    (2005)
  • P. Alves-Oliveira et al.

    Empathic robot for group learning: A field study

    ACM Transactions on Human-Robot Interaction

    (2019)
  • P. Arnold et al.

    Bisensory augmentation: A speechreading advantage when speech is clearly audible and intact

    British Journal of Psychology

    (2001)
  • P. Arnold et al.

    Bisensory augmentation of complex spoken passages

    British Journal of Audiology

    (2001)
  • E.T. Auer et al.

    Enhanced visual speech perception in individuals with early-onset hearing impairment

    Journal of Speech, Language, and Hearing Research

    (2007)
  • A. Baylor et al.

    The effects of pedagogical agent voice and animation on learning, motivation and perceived persona

  • P. Brown et al.

    Politeness: Some universals in language usage

    (1987)
  • Y.-Y. Chang

    Evaluation of TTS systems in intelligibility and comprehension tasks

  • S.D. Craig et al.

    Animated pedagogical agents in multimedia educational environments: Effects of agent properties, picture features, and redundancy

    Journal of Educational Psychology

    (2002)
  • S.D. Craig et al.

    Text-to-Speech software and learning: Investigating the relevancy of the voice effect

    Journal of Educational Computing Research

    (2019)
  • J.J. Cummings et al.

    How immersive is enough? A meta-analysis of the effect of immersive technology on user presence

    Media Psychology

    (2016)
  • R.O. Davis et al.

    Sometimes more is better: Agent gestures, procedural knowledge and the foreign language learner

    British Journal of Educational Technology

    (2019)
  • R.O. Davis et al.

    Reconsidering the voice principle with non-native language speakers

    Computers & Education

    (2019)
  • Y.A. De Kort et al.

    Digital games as social presence technology: Development of the social presence in gaming questionnaire (SPGQ)

    Proceedings of PRESENCE

    (2007)
  • S.A. Duffy et al.

    Comprehension of synthetic speech produced by rule: A review and theoretical interpretation

    Language and Speech

    (1992)
  • A. Eriksson et al.

    Perception of vocal effort and distance from the speaker on the basis of vowel utterances

    Perception & Psychophysics

    (2002)
  • R. Evertsz et al.

    January). Enhanced behavioral realism for live fire targets

  • B.T. Files et al.

    Visual speech discrimination and identification of natural and synthetic consonant stimuli

    Language Sciences

    (2015)
  • C. Frechette et al.

    The roles of animated pedagogical agents' presence and nonverbal communication in multimedia learning environments

    Journal of Media Psychology

    (2010)
  • A.B. Frymier et al.

    The teacher-student relationship as an interpersonal relationship

    Communication Education

    (2000)
  • D.R. Garrison

    Online community of inquiry review: Social, cognitive, and teaching presence issues

    Journal of Asynchronous Learning Networks

    (2007)
  • P. Gebhard et al.

    Exploring interaction strategies for virtual characters to induce stress in simulated job interviews

  • R.E. Guadagno et al.

    Virtual humans and persuasion: The effects of agency and behavioral realism

    Media Psychology

    (2007)
  • M. de Haas et al.

    The effects of feedback on children's engagement and learning outcomes in robot-assisted second language learning

    Frontiers in Robotics and AI

    (2020)
  • C. Heeter

    Being there: The subjective experience of presence

    Presence: Teleoperators and Virtual Environments

    (1992)
  • F. Herrera et al.

    Effect of behavioral realism on social interactions inside collaborative virtual environments

    Presence: Virtual and Augmented Reality

    (2018)
  • C. Heyes

    Born pupils? Natural pedagogy and cultural pedagogy

    Perspectives on Psychological Science

    (2016)
  • M. Hildebrand et al.
    (1971)
  • S. Hobert et al.

    Say hello to your new automated tutor – a structured literature review on pedagogical conversational agents

    Proceedings of the 14th International Conference on Wirtschaftsinfomatik

    (2019)
  • F. Jamet et al.

    Learning by teaching with humanoid robot: A new powerful experimental tool to improve children's learning ability

    Journal of Robotics

    (2018)
  • L. Jensen et al.

    A review of the use of virtual reality head-mounted displays in education and training

    Education and Information Technologies

    (2018)
  • K. Johnsen et al.

    Experiences in using immersive virtual characters to educate medical communication skills

  • Z. Kasap et al.

    Building long-term relationships with virtual and robotic characters: The role of remembering

    The Visual Computer

    (2012)
  • H. Kato et al.

    On the human ability to auditorily perceive human speaker's facing angle

  • Cited by (25)

    • How pedagogical agents communicate with students: A two-phase systematic review

      2022, Computers and Education
      Citation Excerpt :

      Learning is a socially engaging activity, and PAs can bring relational closeness to SAC, thus, PAs also should avoid intimidating language and, instead, be polite and sensitive. Social fidelity through personalized language, politeness, personality, and social memory, are promising ways to enhance closeness in the relationship between student and agent (Sinatra et al., 2021). Overall, the bulk of empirical studies show that PAs possess some interpersonal level capabilities that are relevant for pedagogical purposes, stressing the importance of instructional communication research in PA development.

    • The evaluation of problem-solving oriented e-module in learning computer-based subject

      2024, International Journal of Evaluation and Research in Education
    View all citing articles on Scopus
    View full text