Navigating the design of simulated exercising peers: insights from a participatory design study

Background: To fight sedentary lifestyles, researchers have introduced various technological interventions aimed at promoting physical activity through social support. These interventions encourage people to exercise together, maintaining high levels of motivation. However, the unpredictable nature of human peers makes it challenging to control behavior and balance these interventions effectively. Artificial intelligence agents, on the other hand, can provide consistent social support and are more controllable. Hence, we propose Simulated Exercising Peers (SEPs) as a promising solution for providing agent-based social support for physical activity.

Method: Participatory design sessions were conducted, involving young adults in the creation of SEP-based interventions. Sixteen participants generated four prototypes that varied in aesthetics, behavior, and communication style, with outcomes analyzed through the lens of Self-Determination Theory to better understand the motivational implications of each design.

Results: Findings highlight key components crucial for designing SEPs that enhance acceptance and efficiently integrate into physical activity interventions. Additionally, the study revealed how the aesthetics and behavior of SEPs could potentially deceive users, which can lead to user disengagement from interventions involving SEPs. Participants also defined two distinct social roles for the SEPs, i.e., coach, and companion, each associated with unique communication styles.

Conclusion: This study offers five design guidelines for the development of SEPs, AI agents aimed at promoting physical activity through social support, and highlights opportunities for their integration into broader physical activity interventions.

1 Introduction

Engaging in regular exercise sessions of moderate-to-vigorous intensity can have positive effects on human health and overall well-being. However, according to a 2022 World Health Organization (WHO) report, more than 1.4 billion people (i.e., more than a quarter of Earth’s population) do not meet the recommended physical activity levels (1). WHO’s report also highlights that physical inactivity will represent a yearly cost of 27 billion US dollars in disease treatment between 2020 and 2030.

Physical activity does not exhibit a linear decline across all age groups. Rather, researchers have identified adolescence and young adulthood as crucial periods when individuals’ physical activity habits undergo significant changes (2–4) and bad habits might crystallize (5).¹ Hence, young adulthood becomes a pivotal period for promoting long-term physical activity behaviors within this demographic (6), as changes during this period are likely to persist into adulthood.

Studies have highlighted that male adolescents and university students mention the absence of social support as a barrier to engage in more physical activity (7–9). To address such barriers, Human-Computer Interaction (HCI) research leverages persuasive digital technologies to promote behavior change, highlighting the crucial role these technologies play in enabling interventions based on various psychological theories (10). In particular, several interventions have demonstrated the effectiveness of promoting peers social support in increasing engagement in physical activity (11–15). One of the most obvious advantage of these interventions was the ability to connect peers who might be interested in exercising together, but who could not be available at specific times or places.

However, these seminal studies have also revealed the limits of these approaches. Firstly, even if technology could bridge the gap of finding an exercising partner, several users might still be reluctant to engage in physical activity due to self-consciousness and fear of being judged (16). Also, it is hard to match people effectively. Exercising with someone who might be over- or under-performing might prove disengaging over the long run (17). This highlights how individuals’ idiosyncratic preferences make it complex to predict the effectiveness of these interventions as individuals might be more receptive to certain kinds of partners (e.g., family, friends, strangers) (18), or social strategies (e.g., competition, cooperation), as studied by Orji et al. (19).

Given the challenges in designing interventions that could scale to large populations of users, researchers in HCI have introduced the use of agents, i.e., computer-based entities that can act autonomously and interact with users. AI agents have the advantage that they are always available and that they can adapt their appearance [e.g., animals (20, 21), humans (22)], behavior, and communication abilities (23) according to the user preferences. The main objective of these agents is to understand and assist individuals in their physical activity, providing their human peers with tailored feedback, instructions, education, and a social presence (24). In this paper, we are going to refer to this type of agents as Simulated Exercising Peers (or SEPs, in short). Unfortunately, the design of SEPs remains largely undefined. Their effectiveness heavily depends on their ability to establish a relationship with individuals, provide relevant and context-sensitive feedback (25, 26), and create meaningful social connections (20, 27–29). Previous research on agents for physical activity focused on conversations, and has overlooked the potential implications of the agents’ visual characteristics in physical activity interventions (30). Unfortunately, to this day, there are no guidelines to be found in the scientific literature that can help designing SEPs. These studies failed to develop specific design recommendations for SEPs. Most importantly, prior research has designed interventions deductively, moving from psychological theories of human motivation into design embodiments.

This study aims to address current gaps in drafting design guidelines for SEPs by exploring aesthetics, behavior, and communication aspects of these agents. Contrarily to prior work, we aim at generating these recommendations inductively, using a bottom-up approach. This approach could greatly benefit the field and the specific design space of SEPs by revealing nuanced considerations in design (31). Therefore, we conducted a participatory design study: a collaborative approach to design that actively involves stakeholders—especially end-users—throughout the design process to ensure the resulting technologies or solutions align with their needs, values, and contexts (32). We involved 16 university students (19–28 y.o.) ensuring participants’ background heterogeneity (33). Participants were asked to co-design their ideal SEP specifically focusing on their visual aspects, their behavior, and communication style. To analyze the solutions produced by the participants, we decided to use the Self-Determination Theory (34). This theory is particularly useful in studying SEPs as one of its main constructs is relatedness, namely the need to feel connected to other people in a meaningful way (35), which is a fundamental aspect of the SEPs’ mission. Also this theory has already been scientifically validated through technology-based empirical studies (36–39).

As a first contribution of this study, we provide an in-depth analysis of the requirements and expectations of young adults regarding AI agents designed to support their daily physical activity. Our work emphasizes three core aspects crucial for creating believable and relatable agents: emotional intelligence, personalization, and contextual relevance. Specifically, our research extends the concept of agent believability from virtual worlds literature, demonstrating that an agent’s visual appearance significantly influences user expectations of its behavior and communication skills (40–42). Through participatory design, we discovered that integrating SEPs’ behavior with their aesthetics is crucial for creating engaging and motivating agents to support physical activity. Unlike previous studies on relational agents (43) and virtual coaching (44), our participants envisioned SEPs as entities sharing the physical activity journey with users as peers rather than acting as third-party advisors.

Secondly, this study reveals the critical impact of deception on SEPs, which can arise from misleading aesthetics, behavior, and communication skills of virtual agents. The young adults defined deception as believing that a user is a human being when it is actually an AI agent. To this end, the implications of this study suggest avoiding human-like appearances for SEPs, and preferring animal or cyborg aesthetics instead. Participants highlighted that deception not only undermines trust but also discourages engagement and induces an eerie feeling. Additionally, our study highlights the challenges of creating human-like agents, as users expect higher levels of believability, leading to frustration (45) and uncanny feelings when these expectations are not met (46).

Thirdly, our work provides actionable guidelines for designing SEPs that foster basic psychological needs, a key concept in SDT for promoting intrinsic motivation, which is the strongest type of motivation. Our findings emphasize the importance of relatedness, or social connectedness, identifying that SEPs can motivate participants by establishing mutual care relationships. This completes the existing literature as it only typically explores one-sided care relationships (22, 28, 47).

Finally, this paper provides specific and practical design implications for developing AI agents that effectively support physical activity and promote overall well-being, filling the gap in the fields of HCI and behavior change technologies.

2 Background

Behavior change theories provide a framework for understanding and influencing how individuals alter their behaviors, making them essential for developing effective interventions (48) in different fields. The use of these theories has been highly increasing in the context of physical activity—notably since the WHO expressed alarming concerns on the detrimental effects of physical inactivity and sedentary behavior (1). The landscape of behavior change theories in the context of physical activity is broad and continually expanding (49, 50). This growth makes it challenging to determine which of the theories is best suited for a given intervention. Relying on theoretical frameworks is essential for assessing and predicting the impact of technologies and features on behavior change (51). Additionally, employing behavior change theories can facilitate the development of interventions that have long-term effects on individuals’ behavior (52). Integrating these theories is even more important in the context of the third-wave of HCI (53), opening new horizons for technology supported behavior change research (51, 54).

2.1 Behavior change theories in the context of physical activity

Although a broad range of theories to explain human behavior exists, researchers often focus on a narrow selection when investigating physical activity promotion (55, 56). These predominant theories include the Social Cognitive Theory (SCT) (57), the Transtheoretical Model (TTM) (58), the Theory of Planned Behavior (TPB) (59), and the Self-Determination Theory (SDT) (34).

SCT, as introduced by Bandura (57), emphasizes the significant influence of social and environmental factors on individual behavior. This theory suggests that behavior is learned by observing others within a context of continuous interaction among environmental, behavioral, and personal cognitive factors. Central to SCT is the concept of self-efficacy, which is an individual’s belief in their ability to succeed in specific situations. Self-efficacy is shaped by four main sources: mastery experiences, where successes and failures respectively strengthen or undermine personal efficacy beliefs; vicarious experiences, where observing peers succeed can enhance one’s belief in their own abilities; social persuasion, where verbal encouragement from others persuades individuals of their capability to succeed; and somatic and emotional states, where individuals rely on their physical and emotional conditions to judge their capability in achieving a given activity. In SCT, self-efficacy, alongside personal goals, is a critical determinant of physical activity behaviors, as highlighted in the literature (60). However, the application of SCT in addressing physical inactivity has yielded inconsistent results, particularly concerning the impacts of outcome expectations and socio-structural factors (61, 62). Moreover, the effectiveness of SCT varies with the age of participants, with older individuals generally showing more positive behavioral changes. Due to these mixed findings, we decided against basing our study on SCT.

TPB rather posits that an individual’s intention to perform a behavior is the primary predictor of actual engagement in that behavior (59). This intention is influenced by three key factors: attitude towards the behavior, subjective norms, and perceived behavioral control. Attitude refers to one’s evaluation (either positive or negative) of performing the behavior, subjective norms is defined as the perceived peer pressure to perform or not the behavior, and perceived behavioral control represents an individual’s belief in their capabilities to execute the behavior. Despite the significance of these factors in shaping intentions, research indicates that even substantial changes in intentions often lead to only modest changes in actual behavior (63). This gap highlights a limitation of TPB: it does not incorporate volitional strategies such as planning and self-regulation, which are crucial for behavior maintenance (55, 64). Additionally, TPB primarily focuses on the initiation of behaviors rather than their sustained execution and lacks comprehensive longitudinal studies to effectively differentiate between individuals who maintain behaviors and those who do not (65). Given these limitations, particularly the theory’s inadequate focus on long-term behavior maintenance, we decided not to use TPB as a framework for designing SEPs, which aim to promote sustained physical activity.

TTM offers a structured, six-stage approach to understanding behavior change, which includes a focus on maintenance, unlike TPB (58). TTM outlines a progression through distinct stages that reflect an individual’s readiness to change. The initial stage is pre-contemplation, a stage where individuals’ are not willing to change their behaviors. Whereas, individuals in the next stage, contemplation, are considering making a behavioral change but have not taken any actions yet. Preparation is the first intentional stage, where individuals have started taking actions towards changing their behavior, for example by increasing their daily physical activity levels. This stage is reached after regular repetitions of their actions for at least six months. When the actions keep to be repeated regularly for a longer period (i.e., six months or more), individuals are considered to be reaching the maintenance stage. TTM acknowledges that progression through these stages is not necessarily linear but can be cyclical, with individuals potentially moving back and forth between stages. Despite its structured approach and inclusion of a maintenance stage, TTM faces criticism for several reasons such as the lack of evidences on the positive effects of this theory (66), and the lack of validated algorithms to assess the individuals’ current stage of change (67). More importantly, TTM does not account for key external factors that could influence individuals’ behavior (68) such as the social factors (69), a crucial element of SEPs’ design. Given these limitations, particularly the model’s insufficient consideration of external social factors fundamental for SEPs’ design, TTM, while insightful, may not fully address the needs of designing effective SEPs.

SDT is a broadly applied theory in technological behavior change interventions that do not necessarily target physical activity (70); however, research relying on it has been consistently growing (71). SDT is an organic theory composed of multiple sub-theories, enabling the application of its constructs at different stages of the interventions, demonstrating the high versatility of this theory. These observations have motivated us to use SDT as a foundational theory for designing SEPs, AI agents specifically designed to support individuals’ physical activity and foster relatedness.

2.2 Self-determination theory applied to physical activity support

SDT offers an insightful framework for understanding motivation through its organic and dynamic structure, which is articulated by multiple sub-theories (34). According to SDT, motivation moves along a continuum (35) going from amotivation to intrinsic motivation. The first state, amotivation, defines the reluctance or disinterest in the task or activity. Then, extrinsic motivation, is a state where the individual’s motivation depends on external stimuli and is further decomposed into multiple types, ranging from external: regulation to the task or activity driven by the external stimuli, to self-regulation: where individuals feel as the owners and the initiators of the task or activity. The continuum ends with intrinsic motivation, where individuals perform a task out of self-interest and their own volition. Interventions seek to support the transition of individuals from an extrinsically motivated behavior to an intrinsically motivated one.

Furthermore, Ryan and Deci (34) emphasize that reaching self-regulated motivation depends on satisfying Basic Psychological Needs (BPNs). Individuals are likely to take on activities that best promote the three BPNs: autonomy, competence, and relatedness. Autonomy is satisfied when an activity is performed out of self-interest and volition. Competence relates to believing in one’s ability to successfully accomplish a task or activity. Relatedness involves feeling cared for, caring for others, and being connected with others. Feeling that a peer is genuinely interested has a crucial effect on relatedness. Specifically, researchers have observed that the level of attachment to the care provider influences the promotion of relatedness (18). The feeling of relatedness is not limited to direct relationships and can appear online. For example, researchers have demonstrated the positive effect of social media health platforms on relatedness (72). Similarly, positive and encouraging comments have been observed to increase engagement and the promotion of relatedness in behavior change activities (15).

Autonomy is tied to the freedom of choice available to individuals, and relatedness hinges on the supportive presence of peers during tasks; however, fostering a sense of competence remains a challenging task. Therefore, SDT introduces the concept of optimal challenge as a construct to support the BPN of competence. Optimal challenge posits that a task should be neither too easy nor too difficult for individuals to perform; it should always be adapted to their capabilities (34, 73). By fostering competence, optimal challenge is likely to lead individuals to a feeling of mastery while performing a task and increase the chances of intrinsic motivation towards the task. Additionally, optimal challenge can facilitate experiencing flow—a state defining a complete immersion and focus into an activity—which leads to greater enjoyment and fulfillment (74).

The focus on competence within SDT exemplifies how its principles can be leveraged to enhance motivation and facilitate behavior change in technology-based interventions. SDT has been extensively applied and validated across various HCI domains related to behavior change (36). Specifically, in the realm of physical activity, SDT has deepened insights into motivational dynamics and behavior change processes in HCI interventions (71). However, a significant challenge identified in the literature is identifying optimal moments to provide support within fully integrated behaviors. This difficulty has prompted researchers to advocate for the integration of SDT principles early in the design phase of interventions, aiming to enhance the likelihood of creating impactful user experiences that effectively satisfy BPNs (75, 76). In response to this, taxonomies have been developed to assist in the design of HCI interventions, categorizing mobile app features based on the specific BPNs they target (77). Additionally, SDT principles have been incorporated into the design stages of interventions, proving beneficial in structuring persona designs (78) and in formulating design guidelines for conversational agents that support BPNs fulfillment (79).

Our decision to utilize SDT in designing SEPs is informed by these empirical findings and the theory’s applicability in exploratory contexts like ours. At this stage, we employ SDT as a framework to interpret key literature insights for designing SEPs that promote physical activity through social support. Further discussions will elaborate on how SDT’s constructs are integrated with SEPs, as they appear in our findings.

3 Related work

Designing virtual agents requires three things: a. defining the visual characteristics of the agent; b. establishing the behavioral guidelines, describing what it can or cannot do within the scope of the interaction with its human counterpart; and c. defining the rules that govern how it should interpret and respond to human language. In this section, we are going to review past research that covered these three areas.

3.1 Aesthetics of agents

The visual appearance of agents is a pivotal factor in user interaction, making it an essential consideration in the design of SEPs. Research distinguishes between two main types of supportive agents for physical activity: physical and virtual agents (80). Both types are capable of engaging in social interactions with humans through dialogues, natural cues like gestures, and emotional expressions (81). Studies indicate that both young and older adults tend to enjoy interactions with physical agents more, attributing a greater sense of presence to them compared to their virtual counterparts (82, 83).

Despite their advantages, physical agents come with higher costs, require specific materials, and need controlled environments for safe operation, which restricts their mobility and usability. These constraints make virtual agents, particularly those integrated into mobile interventions, a more viable option due to their pervasiveness and flexibility. Virtual agents allow for precise control over their characteristics, such as emotions, gestures, and visual appearance, enabling diverse aesthetic representations including dogs (20, 27), dragons (28), fish (84), and abstract creatures (29).

Moreover, the ability to personalize these virtual agents has been demonstrated to significantly boost user engagement (85, 86). This adaptability and personalization potential make virtual agents particularly suited for SEPs, offering a dynamic and user-centric approach to supporting physical activity.

The digital form of agents can also be used to provide feedback on users’ physical activity engagement by modifying their shape (e.g., making them wider or thinner) (20) or displaying specific emotional states (e.g., sick, unhappy) (28, 84). These emotional displays can directly impact users’ feelings of guilt (84), motivating them to engage in physical activity as they care for their agent.

Despite these findings, many studies offer limited justification for their visual appearance choices, making it difficult to assess their long-term impact on behavior change. This is concerning, as research in virtual worlds highlights that aesthetics play a crucial role in shaping human-agent relationships. Users form expectations about an agent’s behavior and capabilities based on its visual appearance (87, 88). In video game research, this is examined through the concept of agent believability—the perceived contrast between user expectations and the agent’s actual behavior (41, 42, 89). Significant discrepancies can undermine believability, negatively affecting user experience, immersion, and ultimately, motivation. However, current research on the believability of agents in physical activity contexts remains insufficient to provide definitive design guidelines for SEPs.

In addition to limited research on agents’ visual appearance, their design is further complicated by the need to accommodate idiosyncratic preferences, which can lead to unmet expectations. Video game researchers address this challenge by enabling players to personalize their avatars, allowing them to tailor avatars to their liking. Empirical studies show that personalization features strengthen users’ psychological connection with digital entities (90–92), increase attachment (93), and enhance engagement by supporting BPNs (85, 86). These findings reinforce the relevance of SDT as a foundational theory for SEP design.

Given the importance of aesthetics in shaping user interactions and motivation, it is crucial to investigate users’ expectations regarding the visual characteristics of SEPs. Despite a large interest for agents’ visual appearance in the video game literature, there are no existing studies that adequately inform the effects of visual appearance on promoting motivation in physical activity interventions. To address this gap, we aim to answer the following research question:

RQ1. What are users’ expectations in terms of visual appearance for a simulated exercising peer that can promote motivation in physical activity interventions?

3.2 Agent behavior

The use of agents as peers in group-based physical activity is a novel area with limited research on their specific behaviors. While extensive research has investigated the benefits of group-based physical activities involving human peers, the behaviors exhibited by agents in these contexts remain unclear. Grouping peers can foster the BPNs of autonomy, competence, and relatedness if implemented correctly (94–96). However, the success of group-based physical activity heavily relies on the behavior of each peer, as the idiosyncratic nature of individuals’ training habits and performance levels complicates the creation of interventions with appropriate social strategies. This can lead to unfair social comparisons. Consequently, agents have the potential to promote physical activity by adapting their behavior to individual preferences, but their behavior must align with users’ expectations based on their visual appearance.

To define agent behavior effectively, it is essential to consider the context of interaction, particularly the social strategies of competition and cooperation. These strategies can significantly influence the dynamics of promoting physical activity in group interventions (97). In competitive settings, individuals aim to outperform their peers, which has been observed to enhance individual performance (98). Conversely, in cooperative settings, individuals work together towards a common goal (99). Both strategies can lead to positive outcomes in physical activity interventions, depending on the social interdependence of the task (100).

Social interdependence exists when individuals share common goals, and their outcomes are affected by their own and others’ actions (101). For example, highly interdependent tasks (e.g., team sports such as football, volleyball, or basketball) require cooperation from the entire team, while tasks with low interdependency favor competition (e.g., fencing, tennis, or activities that can be done alone). Tauer and Harackiewicz (102) suggest using intergroup competition as a hybrid solution, benefiting from cooperation within a group and competition against another group, an approach known as “coopetition” in business and industrial strategy (103, 104).

However, group-based activities can present risks when rewards are contingent upon performance, leading to social comparisons where individuals compare their results with those of their teammates (105). Social comparison naturally occurs when tasks are tied to contingent goals and the intervention provides means for comparison (34, 106). Individuals may engage in upward comparison (comparing themselves with higher-performing peers) or downward comparison (comparing themselves with lower-performing peers) (107). The literature is unclear on the positive effects of social comparison, as it can undermine motivation if individuals consistently lose against their peers (34, 108). For example, research has shown that students frequently exposed to social comparison when comparing exam results may experience decreased motivation, particularly among underperforming students (109). To mitigate the negative effects of social comparison, designers should carefully consider the adopted social strategy and user behavior.

Determining the most suitable social strategy for SEPs remains challenging, as existing empirical results show both advantages and disadvantages for each strategy. If not properly addressed, SEPs’ behavior could impact social comparison and lead to disengagement from the intervention.

Another crucial factor in determining agents’ behavior is the concept of believability, which was also discussed in the previous subsection on aesthetics. Ensuring the believability of agents requires that their behavior is consistent with their visual appearance (41, 42, 89). Lankoski and Björk (87) categorize design patterns that define believable human agents in video games, noting that agents should have their own agendas and self-awareness of their environment. Video games often use agents as companions to assist players in tasks and guide them through environments (40). These agents need specific behaviors while still assisting users, and unpredictable behaviors can disrupt the predictability of AI agents’ messages or actions (110). Bailey and Katchabaw (111) proposed a framework for designing psychosocial behavior agents based on emergent gameplay in video games, suggesting that each agent possess its own goals and interests, influenced by user interactions.

While these elements are commonly associated with video games and virtual environments, they suggest that SEPs could have their own objectives, such as achieving a specific number of steps, while helping their human counterparts. This insight contributes to the development of believable agents that could enhance behavior change through SEPs. The literature also underscores the effectiveness of group-based physical activities and the significance of social strategies tailored to different task types. However, social comparisons, inherent to these strategies, can have varying impacts on performance, depending on how well they align with individual capabilities and preferences. To effectively design these interactions, it is crucial for designers to balance social comparison by considering individuals' activity levels and unique preferences. For SEPs, this opens avenues to introduce optimal challenges and position them as ideal competitors in small contests. The performance or difficulty level of SEPs might be adjusted through their visual appearance to ensure believability. Although existing empirical studies provide insights into how SEPs could facilitate interactions and offer support, most research on social comparison and strategies involves human interactions, and the believability of behavior in virtual agents remains less explored, particularly in contexts related to physical activity support. To bridge these knowledge gaps, our participatory design study aims to pinpoint user expectations concerning SEP behavior, focusing on the following research question:

RQ2. What are users’ expectations for the behavior of simulated exercising peers in promoting motivation for physical activity interventions?

3.3 Communicating with agents

In contrast with the aspects of aesthetics and behavior, the communication of agents for the promotion of motivation in physical activity interventions has been vastly investigated. Indeed, the use of Conversational Agents (CAs) became a main persuasive technique. This allowed us to identify the following traits as the most important ones to be integrated in the communication of SEPs, with reference to the current scientific literature.

CAs in the context of behavior change interventions are mainly implemented to deliver personalized advice (47, 112–117), provide support for users’ goal setting and attainment through accurate feedback (26, 118), and support users’ self-reflection on their behavior through educational support (114, 119, 120).

This set of cases illustrates the importance for CAs to support context awareness, and personalization to fit messages to users’ current situation. In the last two decades, researchers have been able to exploit various types of sensors embedded in wearables and smartphones to help understand users context and activities, facilitating the creation of meaningful and time-sensitive feedback (118). However, the design and creation of these messages remains a complex task for researchers, as demonstrated by op den Akker et al. (121)’s study resulting in a multidimensional framework to build motivational messages emphasizing on: the timing, the intention, the content and their representation. In addition to these dimensions, the CAs main leveraging point remains their conversational skills, which may require the use of experts to create and curate messages that fit the purpose of the behavior change intervention (26).

Through prolonged conversations with the agents, users tend to create a bond, a construct that can increase the persuasiveness of the CAs’ messages (47, 122). Notably, researchers demonstrated that long-term relationship with agents can exist and be maintained if both users and agents are engaged in the conversation (43). However, the design of relation-enabled agents complexifies designers’ work as they need to support certain conversational skills such as humor, social dialog, empathy, self-disclosure, and persistent memory (22). These implementation complexities are also illustrated by situations where CAs designed to support and coach for physical activity failed to provide interesting messages, and ran out of messages after some time (123, 124). These situations have mostly led users’ disengagement with their CAs and had negative effects on the users’ experience.

One of the main factors exacerbating user experience is agents’ failure to meet users’ expectations. Users’ expectations on agents’ communicative skills are often based on agent’s visual appearance, underscoring the importance of designing believable agents (125). For example, human-like avatars can lead users to develop higher expectations on agents’ ability to exhibit human qualities (126). Adopting human-like avatars and characteristics, though technically demanding, can enhance users’ sense of social presence (127). This is because demonstrations of humanness shape how users relate to and perceive the support provided by agents (45, 128).

Additionally, higher humanness can influence users’ feelings of autonomy, as such agents are perceived to make fewer errors and provide more predictable responses (79). However, mimicking human behavior without clarifying the agent’s non-human nature raises ethical concerns about deception (129). A risk that can arise when users are exposed to CAs with a high degree of humanness, potentially leading to “subtle deception” (128). Such deception can cause users to distrust the CAs and lose motivation in interacting with them (130).

In summary, CAs have the potential to provide a sense of social presence and support for physical activity. However, designing effective CAs remains a significant challenge, as research has yet to establish well-defined guidelines. A key dilemma lies in meeting user expectations, particularly when interventions utilize human-like avatars that heighten the perception of humanness. This often leads to a mismatch between user expectations and CA capabilities, resulting in frustration. Addressing this issue requires a deeper understanding of the specific communication skills users expect from SEPs—a topic explored in our participatory design study, guided by the following research question:

RQ3. What are the necessary features that would make users feel connected with simulated peers?

4 Methods

Through our related work, we identified gaps in the literature regarding agents’ characteristics that promote motivation in physical activity. Based on our review, we have formulated the following three research questions:

RQ1. What are users' expectations in terms of visual appearance for a simulated exercising peer that can promote motivation in physical activity interventions?

RQ2. What are users' expectations for the behavior of simulated exercising peers in promoting motivation for physical activity interventions?

RQ3. What are the necessary features that would make users feel connected with simulated peers?

Given the limited literature, we adopted a participatory design approach to explore specific needs and expectations for agents promoting physical activity, a method proven effective in enhancing well-being and fostering healthier lifestyles (131–136). We focused on students, a group prone to dropping physical activities due to significant life changes, engaging them in our design study.

Participatory design has addressed physical inactivity across various demographics including adolescents (137), individuals with autism (138), and older adults (133, 136). It has been instrumental in integrating behavior change technology (133), evaluating mobile interventions (137), and designing new physical activity interventions (135). A recent study by Janols et al. (139) combined participatory design with SDT to develop a virtual coach for older adults, identifying three motivational profiles, underscoring the value of this approach in understanding user expectations for SEPs’ aesthetics, behavior, and communication capabilities.

Our review also highlighted that the interaction with intelligent agents is influenced by factors like appearance (140, 141), personality (142), cognitive abilities (143), and proactivity (144), which vary by context and user characteristics. Designing the interaction by involving an AI agent with the target users allows infusing the users’ latent knowledge, culture and emotions into the artefact (145). This possibly increases the final user acceptance and trust. To encourage the ideation process, participants were involved in different activities like “group walk”, “co-design”, “focus groups”, and “mutual evaluation” that we detail in a later section (cf., Section 4.3).

4.1 Recruitment

Participants were recruited from the University of Lausanne’s participant pool, targeting young adults willing to engage in at least 45 min of walking and who owned a smartphone.

Invitations were sent to the entire participant pool two weeks before the first session, including a link to an online screener to verify eligibility for our study; details of this screener are available via the OSF repository dedicated to this study.² Out of 111 registrants, 59 met the eligibility criteria, which required participants to be students available for at least one of the scheduled sessions. The screener also collected demographic information and assessed participants’ familiarity with physical activity tracking services.

In selecting participants, efforts were made to maximize gender diversity and accommodate varying availability during the experiment period, which consisted of two sessions across two different days. To enrich the discussions with diverse perspectives, we recruited students from various academic disciplines. The participant backgrounds included psychology ($n=4$), education science ($n=3$), computer science ($n=3$), criminology ($n=2$), medicine ($n=2$), politics ($n=1$), and Russian literature ($n=1$). Additional demographic details of the selected participants are provided in Table 1.

Table 1

Table 1. The participants demographics, their assigned group, and the session they were participating to.

4.2 Participants

We selected 59 people from the respondents with a matching profile after the screening process. Participants were contacted via e-mail and registered for the sessions according to their availability. We recruited a total of $N=16$ participants (eight females). We grouped the participants according to their registered session day. During the sessions, we balanced gender representation in each session $N_{\mathrm{session}\mathrm{one}}=9$ (five females) and $N_{\mathrm{session}\mathrm{two}}=7$ (three females). Participants within a session were separated into two groups—the first day being composed of a group of four (two females) and another of five participants (three females)—while the second was composed of a group of four (two females) and a group of three (one female). Among the selected participants, $50\mathrm{\%}$ mentioned already having used a physical activity tracking application. Information on participant demographics and the sessions they participated to are further provided in Table 1.

4.3 Procedure

We reviewed the literature on participatory design studies to create our own process (146–151). Before conducting the study, we performed a dry run involving two, non-author, researchers to test and refine our study protocol.

A detailed depiction of our process is shown in Figure 1. We organized the study into two sessions, each accommodating up to eight participants to facilitate management and adhere to social distancing protocols. Sessions were conducted on separate days to maximize attendance. Participants were divided into groups, engaging in identical activities concurrently. This setup ensured efficient management during the walk and simultaneous focus group discussions. All participants provided signed consent before sessions, which were conducted in French and audio-recorded. Compensation was set at USD $84$ (CHF 75) for full participation.

Figure 1

Figure 1. The timeline of activities for our participatory design study.

The sessions included five main activities: contextualization, co-creation: appearance, co-creation: behavior, co-creation: communication, and plenary presentation and discussion. The contextualization phase introduced participants to SEPs via a team-based step tracking app, familiarizing them with the concept and their integration into mobile platforms. Each co-creation activity was followed by small focus groups to facilitate reflective discussion and ensure inclusive participation. These discussions helped monitor and guide the ideation process, ensuring engagement and balanced contributions across the groups. A dedicated slideshow supported each activity, with slides designed to prompt reflection on specific research questions related to SEPs (cf., Table 2). Further details and rationale for the study procedure are discussed in the following sections.

Table 2

Table 2. The questions used to guide the design the SEPs for each stage of our participatory design study.

4.3.1 Contextualization

We conducted a 45-min walk by Leman's lake in Switzerland as a contextualization activity to spark ideation among participants, many of whom were unfamiliar with physical activity tracking apps and SEPs. Before the walk, introductions were made, and participants were given access to the Pacer app,³ chosen for its cross-platform availability, leaderboard, user profiles, and ease of use. All accounts were anonymized and subsequently deleted after the sessions. During the walk, our SEP, named “Eduardo”, was integrated into each group (cf., Figure 2), appearing as a user with an “AI” label on his profile picture (cf., Figure 3). Unknown to participants, Eduardo was present in both groups. Initially, Eduardo had no steps recorded but gradually began accumulating them. In the first 15 min, we encouraged participants to observe and discuss the scores and teams. By the second third of the walk, Eduardo had accumulated enough steps to appear on the team leaderboard and, in the final 15 min, increased his pace significantly, outperforming the participants. After the walk, we gathered feedback on the participants’ feelings and observations and clarified that Eduardo was not a real person but a SEP, explaining the underlying mechanism to prevent any deception.

Figure 2

Figure 2. The contextualization phase’s starting situation with Eduardo our SEP and the rest of the first group. Screenshot from: Pacer app, Pacer Health, Inc.

Figure 3

Figure 3. The profile of our SEP named Eduardo for the contextualization phase. Screenshot from: Pacer app, Pacer Health, Inc.

4.3.2 Co-Creation: appearance

For the ideation phase, participants were encouraged to co-create in groups using A3 blank paper sheets and various design tools such as pencils, colored pens, and post-its. The researcher facilitated the process by providing a series of guiding questions⁴ aimed at aiding the design process. In this initial co-creation session, participants focused on conceptualizing the appearance of the SEPs, defined broadly as their look and feel, to foster creativity without imposing restrictive expectations. Throughout the 30-min activity, the researcher and an assistant were on hand to offer additional information or prompt further reflection through targeted questions. Following the activity, a brief five-minute focus group discussion was held within each group to evaluate the designed SEPs and critically assess the proposed solutions.

4.3.3 Co-Creation: behavior

Using the same material, participants had to reflect on the overall behavior of the SEPs. Iterating on their initial design, the groups had to add components and information about the behavior adopted by the SEPs. A set of questions was also prepared for this phase, to encourage discussion on the performance behavior of the SEPs. Additionally, we asked questions about the purpose of the SEPs and whether there was a link to its behavior.

4.3.4 Co-Creation: communication

In the final design activity, participants were asked to consider methods of communication with their SEPs. Using the earlier walking activity as context, we explored their interest in interacting with the SEP, Eduardo. Specifically, we inquired whether participants wished to receive messages from Eduardo or had messages they wanted to convey to it. We provided prompts to guide discussions on the form of communication [e.g., text messages, kudos (152)], and its directionality. Participants were tasked with defining the purpose of these communications, such as offering encouragement or summarizing the activity. They were also instructed to incorporate these communication strategies into their earlier designs concerning the behavior and appearance of the SEPs. This activity lasted 30 min and concluded with a five-minute focus group discussion to review and refine their ideas.

4.3.5 Plenary presentation and discussion

In this last activity, the participants were invited to present their work. Each group presented their creation. The groups first introduced their SEPs’ appearance, and provided a rationale for it. Then, they did the same for the behavior and finished with the communication. The other group was prompted by the researchers to comment and challenge the proposed design. Additionally, the participants were asked to explain how they envisioned the integration of their SEPs in a mobile application.

4.4 Measurements

Voice recordings were made throughout the sessions to capture participant interactions. During the contextualization phase, researchers carried recording devices and moved towards participants during discussions, inviting them to speak closely to the recorders when asking prompting questions. In the design phase, each table was equipped with a recorder to document the participants’ thought and reflection processes. Focus groups were also thoroughly recorded, with devices strategically placed to ensure all participants were audible. In total, 8 h and 1 min of audio were recorded and subsequently transcribed across all sessions and activities. The transcription process was carried out by the two researchers, who also cross-checked each other’s work to ensure accuracy and consistency in the documentation.

4.5 Analysis

The participatory design was analyzed using the thematic analysis approach (153) after the transcription of the group discussions that were audio-recorded during the sessions. Transcripts were coded using the MaxQDA 2020.3 software. Two researchers worked together on the data analysis and identified $25$ different codes. The coders blindly coded $10\mathrm{\%}$ of the transcriptions and reached a Cohen Kappa of $0.79$, being considered as sufficient (154). Additionally, the coders reviewed the designs created during the sessions. A combination of the design reviews and the codes were then used to group the results according to SEPs: aesthetics, behavior, and social interaction. The codes resulting from our analysis are provided in Table 3.

Table 3

Table 3. The results of the coding analysis, presenting the first order, second order and themes that emerged in the participatory design sessions.

4.6 Ethical considerations

The Institutional Review Board (IRB) of the University of Lausanne approved our study. Both the recruitment of participants and the participatory design sessions were carried out in October 2021. We received permission to have in-presence participatory design sessions with a certificate control and researchers were asked to ensure that all the participants had a valid COVID-19 certificate. The selected participants all provided their informed consent prior to the start of the experiment. Participants were free to address their colleagues with their pseudonyms instead of their real names during the sessions. In all our datasets, participants names were replaced by anonymous identifiers (e.g., P1, P2, etc.).

5 Results

The participatory design sessions yielded diverse SEP concepts, each embodying unique aesthetic, behavioral, and communicative traits. Participants envisioned SEPs in two primary roles: companion or coach, with each role fostering distinct relationships and expectations. For example, companion SEPs and users are expected to mutually care for each other, whereas coach SEPs are envisioned as more proactive in supporting human users.

A critical insight from these sessions was the potential for deception, where users might mistakenly believe an AI agent is human. This risk is especially pronounced in virtual environments with multiple digital entities and could lead to user disengagement if the AI’s nature is misperceived. To address this, participants suggested specific design strategies for SEPs’ aesthetics, behaviors, and communication styles to prevent such misunderstandings.

In the subsequent subsections, we delve deeper into these findings, categorizing them by SEPs’ aesthetics, behavior, and communication. For brevity, only the final designs are discussed here. Detailed sketches and data, shared in compliance with the transparency criteria outlined by Niksirat et al. (194), are accessible via an OSF repository.⁵

To ensure clarity in our results, group names have been abbreviated. Groups will be identified by the letter G followed by their number. For more details on the groups’ composition, refer to Table 1.

5.1 SEPs’ visual appearance

While working on the aesthetics of SEPs, several participants were concerned by the risk of deception. In order to avoid any type of deception, the majority (G1, G2 and G3) used non-human representations for their SEPs. Participants argued that using animals (cf., Figures 4, 5) or cyborgs (cf., Figure 6) instead of a human being would reduce the risks of being deceived. Indeed, these groups (G1 to G3) stated that even with a strong signal (such as the “AI” label on the avatar during the Contextualization phase of this experiment, cf., Figure 3) a human avatar would be a possible source of deception because these symbols can be easily missed. Furthermore, these groups reported eerie feelings after competing against Eduardo, arguing that the SEP's avatar was too human-like, as illustrated by [G3, F]: “Actually, I don’t know, I feel like it is strange [about the SEP avatar]. Personally, I could not look at a complete picture of a person and tell myself it is an AI, it is frustrating, really. It’s frustrating as soon as you know [that it is an AI], […] it is a bit strange to have a photo of a person when you know there is an AI behind it.” Interestingly, the last group (G4) preferred to have an avatar representing a human in an ideal physical shape (cf., Figure 7) [G4, M]: “Yes a human clearly”, [G4, F]: “It would be more motivating if we see that it is shaped like us […] and that it evolves together with us.” This group explained that the visual appearance could not lead—by itself, to deception. Participants in G4 used the SEP’s body shape to represent the physical objective they want to attain—thus, modeling an ideal human self to use as motivational objective. However, these participants also stressed the need to have a distinctive visual cue in the proximity of the avatar, to indicate the SEP’s non-human nature (e.g., an “AI” indication on the side).

Figure 4

Figure 4. Group 1’s design representing a 2D animal (fox) and using a gauge as a performance comparison tool.

Figure 5

Figure 5. Group 2’s map display with the simulated exercising peer on the bottom-right, and on the side of the user’s position point. The mood from the simulate exercising peer is defined in the top, coloured, bar. Depending on the goal achievement rate, the simulated exercising peer would be more aggressive or remain kind.

Figure 6

Figure 6. Group 3’s designs of a 2D simulated exercising peer, going from human-like (right) to a cyborg (middle). This group finally selected the cyborg (middle) representation.

Figure 7

Figure 7. Group 4’s concept map of a human-like 3D simulated exercising peer taking the role of a virtual coach. The visual appearance of the simulated exercising peer should shape the current user on characteristics like height, and weight. The simulated exercising peer would represent an attractive version of themselves (e.g., more muscles) depicting the results obtained if they followed the advice. In this design, users can interact with multiple simulated exercising peers for different fields of expertise (e.g., running, and nutrition). Interactivity with the simulated exercising peer would either be through voice or text messages both possible during and before or after the physical effort. This simulated exercising peer would also adapt the proposed activities to the user’s capabilities and to the weather conditions.

All designs adopted visual elements that could enable self-identification with the avatar (e.g., non-gendered animal, human shape based on users’ attributes). Our participants explained that identification with their SEP would be key to motivate and engage users in the intervention. In addition, G1 and G2 argued that using animals would ease self-identification as animals are typically looked as gender-neutral beings [G1, F]: “Personally I would not have used a human […] I would have used an animal, because everyone knows what it is, and can identify with it.” Interestingly, all groups anticipated that the visual appearance of the SEPs would influence their behavior—a component that we explore in the next section.

Finally, we observed that two groups (G1, G2) designed features that would provide users with rewards. In their designs, the users could receive or buy new personalization accessories for their SEPs (e.g., glasses, t-shirts, collars, etc.) using points rewarded as they exercise and reached their goals. Thus, the rewards effect would be two-fold: encouraging the users through an external stimuli, and enabling the users to further personalize their SEPs visual appearance. This concept is illustrated by one of the participant’s input [G2, M] saying: “[…] there would be levels to attain, like by doing a certain number of meters per day. Or more like it would be that we do a given number of meters to go from level 3 to level 4 as it provides this sensation of accomplishment […] and with the levels, we would unlock new animals or accessories for the animals, and other things like this.”

Taken together, these results showed that participants emphasized the need for adaptability of SEPs’ visual appearance and raised concerns about potential deception. Across groups, participants noted that visual appearance could influence their relationship with SEPs, affecting their ability to identify with the agent. Such identification, they suggested, could strengthen connections and encourage engagement. However, they felt that assigning human-like avatar to AI agents like SEPs might evoke eerie feelings and suggested alternative design options to avoid discomfort.

Participants were also aware of contextual and individual preferences in visual design, recommending personalization features to enhance user satisfaction. Some groups also suggested using SEPs visual appearance as a motivational tool, where users could unlock new accessories as rewards for achieving goals, enhancing their engagement with the SEP.

5.2 SEPs’ behavior

Participants articulated the behaviors of their SEPs around two roles: companion (G1 to G3), and coach (G4). As a companion, SEPs would exercise together with their users. Companion SEPs would adjust their physical activity⁶ schedule to the habits of the human companion as if they were to train together. Other designs also proposed that the companion SEPs would be virtually represented next to the users’ avatar while they both walk (cf., Figure 5). In the role of coach, SEPs would not train with their users, but rather act as a human trainer (cf., Figure 7), as further illustrated by G4 during the final group review [G4, F]: “It would provide us feedback on our performance […] we thought that during the activity it would use audio cues as we can’t use our phone while doing sports in general.” As suggested by this quote, the SEPs in a coach role would provide suggestions or recommendations to help users reach their goal (e.g., reduce the pace when the SEP sees that the user is tired, or suggesting stretching exercises to do after the effort). In a companion role, a SEP was often represented as a target objective for the user to beat. As participants explained, a companion SEP should leverage social comparison to encourage users. Participants suggested the idea that when exercising with another partner (i.e., the SEP) they would feel encouraged to match or exceed their partner’s activity. Thus, designs from G1, G2, and G3 implicitly sustained the idea that companion SEPs would have to exercise providing an optimal challenge to their human users. In other words, companion SEPs should be capable of tracking and modeling users’ physical activity patterns (e.g., daily step count) to appropriately adjust the difficulty of competitions for their human peers. For example, G3 imagined a SEP that would adopt different behaviors during a week—i.e., setting a very high step count one day, and a lower count another day [G3, M]: “Like say, every day in a week of five days, you make it so that it [the SEP] is weaker than you, but from a 100th of step, a small margin. And some days you make it augment a little bit [the number of steps done by the SEP] to give you the will to try and outperform it otherwise you are constantly the best […] or it is better with three days out of four, like three days out of four on average we are stronger than it [the SEP], it stays a little bit under [in the number of steps] than us.” Interestingly, G1 and G2 mentioned that while companion SEPs would respect optimal challenge they should also be capable of adapting their behavior to their visual appearance, i.e., a lion would be faster than a turtle, exemplified by [G2, F]: “We can choose like, if we want something more in front of us [on the map] to challenge us then we take a certain animal […] certain animal will stay in front of us [on the map] or behind depending on the type of animal.”.

Group G4 was the only one to provide an example of SEP taking on a coaching role. Rather than acting as a competitor, the SEP would support users in setting and achieving their physical activity goals. Participants emphasized the importance of SEPs’ ability to model user performance (e.g., average steps, daily step count, etc.) to assist in setting goals and providing an optimal challenge (cf., Figure 7).

5.3 Communication with SEPs

All the groups designed SEPs that could communicate with their human counterpart. The participants envisioned the use of text (G1 to G3) or voice messages (G4) as a medium to interact with the SEPs. Participants in group G4 thought voice was much more adequate to support communication during the exercise, while the users were busy performing physical activity and therefore interacting with their mobile phone was impractical. All groups desired bidirectional communication with SEPs, [G3, M] mentioned: “It would be cool that we could also speak with it [the SEP].” Thus, the users could receive feedback during, or after the effort, but they could also provide feedback to the SEPs. The participants argued that communication from the users to the SEPs could help the SEPs understand the context, e.g., the users are sick and are unable to exercise. For example, G4 emphasized that [G4, M]: “[…] yeah it [the SEP] could propose activities [or exercises] that correspond to the weather. When it rains we might not go running, but we will do activities at home. […] notify us to do activities when we have not exercised in a while and we could say if we are able to proceed or not.” Additionally, the messages sent from the users to the SEPs would also help to regulate the goal with respect to the optimal challenge (e.g., the required effort is too high, thus the goal has to be reduced), as mentioned by [G2, F]: “[…] for when they want a bit more challenge. Like, the animal goes a bit further ahead [on the map] the user can send three types of messages [to their SEP], to continue at the same rhythm, to slow down or to go faster.” Interestingly, group G1 proposed the use of predefined messages when talking to a SEP in order to avoid “misunderstandings” as they identified it as a recurring problem when interacting with CAs.

Furthermore, the participants mentioned that the SEPs role would influence its communication style. Thus, companion SEPs would communicate using friendly informal messages, while the coach SEPs would use a more formal tone. For instance, a participant in G2 expressed a preference against overly formal or directive messages from a SEP acting as a companion. They highlighted the importance of empathy, contrasting it with a “cold”, apathetic entity, like a tree: [G2, M]: “And also it is important to note that it is more a companion than an expert per se, it gives me some exercising advice but without being like an inanimate tree, it should not give orders like ‘do 15 squats’.”

Unfortunately, participants were not able to provide examples of messages companion SEPs could send. However, they stated that messages should be encouraging, and supportive to let users feel cared for [G1, F]: “Every goal achievements it [the SEP] says something like ‘Well done’ […] during the training it keeps sending encouraging messages […] It should never be discouraging.” Two groups (G1, G2) have also imagined situations where the users would have to care for their SEPs, illustrating it with their interest in getting special accessories obtained as they achieved their goals (as reported in Section 5.1).

These results provide a complete overview of young adults on the use of SEPs as a solution to promote physical activity and support relatedness. The four groups proposed four designs and provided a significant amount of information on different aspects of SEPs’ design, notably:

• Participants stated that SEPs should avoid deception and should provide optimal challenges to their users.

• Participants identified two communication styles for SEPs: companion, and coach, each with their own behaviors, frequencies, and tone.

• Finally, participants wished social interaction could be bidirectional, and that relatedness could also be promoted by enabling users to care for their SEPs.

In Section 6, we discuss these results and the observations made by other scholars.

6 Discussion

The designs proposed by our participants highlighted that the risk of deception is a crucial factor for the design of SEPs’ aesthetics, behavior, and communication. For this reason, we structure the discussion in four subsections: deception, appearance, behavior and connectedness. We highlight the key differences in aesthethics, behavior and communication between each SEPs’ roles in Table 5. As we delve into the various components of SEPs, we elaborate on how these elements empower SEPs to encourage users based on the principles of SDT and report a summary in Table 4.

Table 4

Table 4. A summary of SEPs’ features that can promote the basic psychological needs defined by SDT.

Table 5

Table 5. A summary of the key differences in aesthetics, behavior, and communication with respect to SEPs’ roles.

6.1 Why we should avoid deception with SEPs design

Participants (3 out of 4 groups) expressed significant concern about possible deception when interacting with peers, fearing they might be unable to distinguish between a SEP and a real human. This concern is amplified as CAs’ capabilities improve (155). Prior research shows that deception within teams reduces relationship quality by decreasing trust and mutuality (130), which could profoundly affect human-agent relationships if agents adopt deceptive appearances, behaviors, or communication skills. Deception has broader implications, as humans often apply similar social factors to human-to-agent relationships as they do to human-to-human interactions (156, 157). Furthermore, agents’ human-like features can disinhibit emotions, fostering bonds like friendship or partnership (158–160). As emotional and social engagement with technology grows, exemplified by phenomena like personification (158, 160), transparency about the entity’s non-human nature becomes crucial. Otherwise, discovering that a teammate or challenger is an AI could leave users feeling tricked, unfairly treated, or tasked with meaningless objectives.

Designing AI agents with human-like features can lead to unintended consequences. For instance, Cowan et al. (161) found that users hesitated in their actions out of concern for hurting the agents’ feelings, potentially affecting intervention outcomes if users’ decisions are influenced by emotional connections with the AI. However, such effects may not persist after the intervention. While human-like features can foster prosocial behaviors (128), they may also provoke deception and feelings of eeriness, as described by Mori (46)’s Uncanny Valley phenomenon. This discomfort is compounded by the inherent tendency of AI development to lean toward deception, as exemplified by the Turing Test’s focus on agents mimicking humans (162). High human-like characteristics, such as anthropomorphic avatars, can lead users to overtrust AI, suspending disbelief and increasing the risk of deception (163–165). Moreover, designing agents that fit users’ idiosyncratic preferences, beliefs, expectations and assumptions about the agents’ capabilities remains challenging (166).

The considerations that are reported so far in this section refer to CAs, but these considerations would not cover the full set of possibilities enabled by SEPs. In addition to conversations, SEPs include features such as simulating to practicing a sport with the user or guiding the user via demonstration, which could also lead to deception if SEPs implement a near-human simulation (e.g., having its own pace and walking habits).

This study generated some recommendations in this regard to avoid this specific type of deception. First, SEPs should not use human avatars (cf., Section 6.2), and this is particularly important if the intervention pairs SEPs and human users in a social comparison context (e.g., in a competition). Second, the implemented communication modality should remain different between human-to-human and human-to-SEP. For instance, humans should only be able to interact with SEPs using predefined prompts. Third, SEPs’ behavior should remain coherent with their visual appearance (cf., Section 6.3). For example, a SEP adopting the visual appearance of a leopard should provide high performance (e.g., larger number of steps during a day or higher speed in a run) than one resembling a sloth.

6.2 Aesthetics: the benefits of SEPs’ visual appearance personalization

As noted in the Results section (cf., Section 5), all groups displayed diverse SEPs in visual appearance, highlighting varied user expectations for their digital activity partners. Consequently, it’s crucial to offer users the choice of SEPs’ visual appearance. All groups concurred on the importance of this personalization. Echoing video games literature, personalization can enhance the human-virtual character bond (90) and boost intrinsic motivation by promoting autonomy (85). Although our findings on personalization effects are preliminary, allowing such customization could enhance SEP-based interventions. Our results also support and expand on Birk and Mandryk (86)’s findings regarding game character personalization’s role in increasing engagement and interest in video games, suggesting similar benefits for SEP interventions.

This study provides insights into users’ expectations regarding the visual appearance of a SEP to enhance motivation in physical activity interventions. We found that aesthetic preferences vary depending on the SEP’s role, either as a coach or companion, suggesting that personalization features should match the expected role. For instance, interventions could offer avatar items suitable for coach or companion SEPs. When acting as a coach, some participants preferred SEPs that represent their ideal self, adding complexity to using anthropomorphic visuals which could risk deception. However, studies by Yee et al. (167) and Bessiére et al. (168) suggest that human-like features in avatars, such as height or attractiveness, are more influential than complete human-likeness in enhancing self-perception and well-being. Thus, we recommend incorporating human-like attributes in non-human avatars to avoid deception (cf., Section 6.1). In the companion role, participants favored zoomorphic SEPs for their ability to facilitate self-identification without being constrained by gender, suggesting these avatars effectively support diverse user identities. Past research supports the appeal of zoomorphic avatars among adults and young adults (28, 29) and highlights the benefits of self-representation in virtual environments for psycho-physiological well-being and user satisfaction (91, 169, 170). Offering diverse SEP visual personalization options could therefore enhance user satisfaction and positively impact behavior change interventions through increased self-identification and self-perception, as noted in video game research (167, 171).

Our findings align with existing literature on the effects of virtual entities’ visual appearance, highlighting the importance of self-identification and self-perception with avatars. Human-likeness, however, was not universally supported as an influencing factor due to the risks of deception identified by participants in certain scenarios. These results underscore the necessity for interventions involving SEPs to allow users to personalize their SEPs’ visual appearance. Literature also notes personalization as a valuable opportunity to foster BPNs, indicating its potential effect to also increase engagement in SEP-based interventions. Furthermore, personalization can help tailor SEPs to users’ expectations and preferences, however, require to be aligned with the roles SEPs should adopt to assist users in their behavior change journey. These findings suggest that research on avatar personalization, interpretation, and perception extends beyond video games and virtual worlds, potentially influencing the design of agents like SEPs for behavior change interventions.

We continue our discussions exploring SEPs’ roles and their implications for SEPs’ behavior and communication capabilities.

6.3 Behavior: the roles that SEPs can play and the implied behaviors they should adopt

Our results show two specific social roles that SEPs can play, each role having its corresponding expected behavior, i.e., coach or companion. Participants described coach-oriented SEPs’ as supportive digital entities designed to help users achieve defined goals. This is accomplished by influencing users’ thoughts, emotions and actions through conversations—a description aligning with the definition of virtual coaches found in the literature (120, 123, 172–174). Virtual coaches are typically used to provide physical exercise demonstrations (175), to help user set specific goals for their training (120, 123, 172, 173), and advise on health-related matters like nutrition (176). Coach SEPs and users are set to cooperate in order to achieve the users’ defined goals, where the SEP’s purpose is to support motivation and provide technical advice. In the companion role, the SEPs and users are in a competition, thus, becoming a representation of the users’ target to beat. In addition, a group also proposed coopetition as a social strategy engaging companion SEPs and users against other teams composed of humans and SEPs. In both the competition and coopetition strategies, the existence of a performance-contingent reward (i.e., a reward given to the best performing entity) between users, or between users and their SEPs, would likely lead to social comparison. This dynamic was defined in physical activity interventions as “the design that facilitates benchmarking individual’s fitness performance with that of others, and hence provides an opportunity for greater motivation in target behaviours” (99, p. 7), and was also considered as a strategy for physical activity promotion (98, 108). However, research has also demonstrated that social comparison could undermine an individual’s interest in their competitors if the competitors are over-performing them: in this case, the individuals will feel that their competitors are “unreachable” (106). The latter effects are in line with our results: our participants mentioned that social comparison could undermine their motivation in the case where the SEPs would be unbeatable.

Our findings suggest that all SEPs need to provide an optimal challenge. In the coach role, the SEPs would provide an adequate goal for the users to reach by measuring and predicting their physical capabilities. Coach SEPs could further leverage on the goal setting to foster users’ autonomy by letting the users negotiate the next goal. This resonates with most of the virtual coach literature that relies on the agent to advise the user in the goal setting process (120, 123, 172, 173)—emphasizing that coach SEPs should be designed with great attention to optimal challenge so as to set the goal and ensure the promotion of competence. In the companion role, the performance of the SEPs should be adapted to the users’ walking habits and daily number of steps. Thus, companion SEPs would become a fair competitor for the users. Furthermore, our findings suggest that companion SEPs should ensure believability by respecting the users’ expectations emerging from the SEPs’ visual appearance (41, 42, 89). Thus, we observe the SEPs’ believability requirement as an opportunity to let users choose a specific difficulty level or, in other words, the limit of the social comparison gap that could be created between the users and their SEPs. Overall, providing an optimal challenge would let SEPs promote users’ competence BPN as reported by SDT (34).

These results allow extracting two main design implications that fit RQ2: What are users' expectations for the behavior of simulated exercising peers in promoting motivation for physical activity interventions?: first, it is crucial to let users choose the specific role of the SEP, namely whether the SEP should behave as coach or companion; second, a SEP should always assess users’ capabilities to set tailored goals that make for an optimal challenge. The role of the SEP will also influence the social strategy that will be adopted between the users and the SEPs. For instance, companion SEPs are to be preferred in a competition or coopetition scenario, while a coach SEP would be the appropriate choice for cooperation.

The choice of the role also has an influence on the design of interactions between users and SEPs and we further analyze on this matter in the next section.

6.4 Connectedness: how SEPs’ roles can define their communication styles and their needs

In this study, participants expressed a desire to engage with SEPs during the contextualization phase to understand the entity’s identity and purpose. All groups included a chat feature in their designs, with most favoring textual communication, while one group suggested voice-based interaction for continuous support during exercise. SEPs’ messages were primarily designed to encourage and support users, aligning with research showing such techniques enhance engagement in physical activity interventions (177). However, overly frequent or predictable automated messages can lead to user disengagement (178), emphasizing the need for tailoring messages across dimensions like timing, intent, content, and representation (121). Furthermore, agents should adapt their interactions with the users’ routine (79), and motivation (179) to avoid being ignored.

Participants highlighted bidirectional communication, proposing a prepared prompt-based system to mitigate misunderstandings due to SEPs’ limitations in Natural Language Processing (NLP). Similar mechanisms, as noted by Ashktorab et al. (180), repair misunderstandings common in CAs with restricted NLP capabilities. Such approaches have also been used to minimize errors in agent responses, particularly for rule-based systems that rely on predefined triggers (172, 173). However, while effective, these solutions can limit users’ freedom during interactions, further highlighting distinctions between human-to-human communication and interactions with SEPs.

Participants suggested that SEPs’ communication styles should align with their roles. In the coach role, SEPs would use a formal tone, focusing on recommendations and instructions. As companions, SEPs would adopt a more informal, friendly style, with frequent messages, including small talk and motivational interactions. This contrast may reflect participants’ experiences with human counterparts in similar roles. This resonates with the literature on companionship that defines it as a social relationship characterized by intimacy, shared activities, and emotional connection (181–184), whereas coaches are often viewed as professional service providers (185), leading to expectations of a more structured interaction.

These findings align with previous research showing that CAs’ communication style similarity enhances users’ sense of presence (186). Informal communication by companion SEPs may increase perceived relatedness and companionship, while coach SEPs focus on optimizing user performance. However, neither our findings nor existing studies clarify how relatedness differs between these roles. While this study focused on the coach and companion roles, SEPs could also serve broader roles in contexts like family dynamics, fostering relatedness and integration in social settings (187).

Participants primarily designed SEPs to support users by nurturing their relationship with them. However, our findings suggest that companion SEPs could also receive support from users, reinforcing the bond rather than solely motivating through nurturance. These observations are consistent with prior research showing that humans form relationships and care for virtual entities (28, 29, 188). Additionally, participants proposed using rewards not for traditional self-rewards (e.g., badges or trophies) but to unlock visual accessories for SEPs, fostering care and strengthening the connection.

These findings resonate with studies on virtual pets, such as the iconic Tamagotchi,⁷ described as “creatures from another planet that needed human nurturance, both physical and emotional” (189, p. 506). Although Tamagotchis created strong bonds, they were often discarded when users lost interest (190). This risk also applies to SEPs, though it may be mitigated as SEPs focus on supporting users’ behavior change, rather than serving solely as entertainment tools.

Based on empirical results and participants’ designs, we identified key features to answer RQ3: What are the necessary features that would make users feel connected with simulated peers? First, bidirectional communication should be enabled, with predefined prompts to minimize SEPs’ misunderstandings. However, this may evolve as new NLP techniques, such as the mainstream adoption of ChatGPT, show unprecedented results in language and intent understanding (191). Second, our findings suggest that users seek believable agents, which requires clear role definitions for SEPs. Beyond visual appearance and behavior, communication capabilities should align with the social role SEPs assume in the intervention. This extends the theory of agent believability, showing that user expectations are influenced by the roles SEPs are designed to play. Finally, we advocate for integrating gamification, where rewards can personalize SEPs’ appearance. This approach would make goal achievement more meaningful, as users perceive rewards as virtual gifts for their SEPs, potentially strengthening bonds and promoting relatedness, similar to dynamics observed on digital platforms and in games (192, 193).

6.5 Summary of design implications

As discussed in the previous subsections, this participatory design study allowed for the identification of five key implications for the design of SEPs that we summarize in this list:

1. Implement distinctive visual cues: Ensure SEPs have clear visual markers (e.g., a unique color scheme, abstract or robotic features) that differentiate them from humans in order to avoid risks of deception. For instance, use non-human designs or other distinctly artificial characteristics;

2. Define and align SEPs’ roles: Clearly define the social role of each SEP (e.g., coach or companion) before starting the design process. Shape their behavior and communication style to match this role. For example, a coach SEP could use motivational language and structured feedback, while a companion SEP might use more casual, friendly interactions;

3. Enable personalization: Allow users to customize their SEP’s visual appearance to reflect their preferences and ideal selves. Provide a variety of customization options such as body type, attire, and accessories that align with the SEP’s role (e.g., sports’ expert gear for a coach SEP);

4. Personalize challenges: Design SEPs to assess users’ physical activity levels using wearable devices or app integrations. Based on this assessment, personalize the activity goals to provide an optimal challenge. For example, gradually increase step targets as users’ fitness improves; and

5. Facilitate bidirectional communication: Create communication features that allow users and SEPs to interact in a meaningful way. Integrate options for users to send and receive virtual gifts or messages, fostering a sense of mutual care. For instance, SEPs could send encouraging messages or virtual rewards after users achieve their goals, and users could thank their SEPs for support or adjust SEPs’ behavior.

6.6 Limitations of our approach

Our study was conducted with several limitations. Firstly, the design of SEPs, like other agent designs, may depend on various social factors and demographics, such as region, beliefs, culture, age, or gender of the participants. While we made efforts to select participants with diverse backgrounds and aimed for equal gender representation, our findings cannot be generalized beyond the young adult population in this region. Achieving broader generalization within this demographic would require additional participatory design sessions across various regions worldwide. Furthermore, our participatory design cohort consisted primarily of young adults, as this was the specific target group of our research. While involving teenagers or older adults might have led to distinct prototypes and perspectives on SEPs, our central objective was to explore the needs and expectations of young adults, and our study was not designed to address other age groups.

Secondly, we relied on a pre-existing mobile application to contextualize our participants, while we gained time using it, it imposed certain limitations. We could only represent the SEP through an avatar and a step counter indication displayed on a leaderboard. Moreover, participants could see their teammates (session participants) on the leaderboard, possibly influencing social comparisons. Some smartphones already stored step counts during the day, and our mobile application synchronized these counts, providing certain participants with an advantage. While the social comparison didn’t hinder motivation for activity performance, competing against such a high-performance teammate prompted remarks by some participants. To address such emotional states, we limited our questions to the SEP’s performance compared to theirs, allowing them to focus solely on the design of the SEP.

Thirdly, the leaderboard structure inherently placed participants in a competitive position, not letting them experience cooperation or coopetition. This situation might have impacted the SEP’s design concerning the chosen social strategy. In an attempt to alleviate this limitation, we emphasized the behavior of the SEP rather than other group members to the participants.

7 Conclusion

We have involved a total of 16 young adults in participatory design sessions to gain insights on the design of new AI agents to support physical activity and foster relatedness. By analyzing scientific literature and adopting a user-centered approach, we have distilled a comprehensive list of design implications for SEPs, encompassing three main components: aesthetics, behavior, and communication. These design implications contribute to the field of HCI, offering insights into how to address the potential challenges of deception that can arise when users struggle to distinguish between human peers and AI agents. Participants highlighted that attention to SEPs’ visual appearance could facilitate self-identification, and emphasized its importance in defining SEPs’ behavior and communication styles to ensure believable and effective agents.

We have also identified SEPs’ potential to foster motivation for physical activity by effectively addressing SDT’s BPNs. Of particular significance, we elaborate on the implementation of optimal challenge with SEPs as an additional means to create fair objectives to support users in their physical activity behavior change journey.

Finally, the ongoing advancements in AI agency, exemplified by the emergence of Large Language Models (LLMs), hold the promise of a bright future for the development of SEPs with enhanced capabilities to support individuals in their daily physical activity endeavors. We hope that the presented study will contribute to the design of technological interventions that could empower users and allow them to have a healthier lifestyle.

Looking ahead, future work will delve deeper into the practical implementation of these design implications. This will involve incorporating SEPs into longitudinal experiments to explore how users interact with these agents and interpret the integration of their aesthetics, behavior, and communication skills. Additionally, future research should address the potential ethical issues associated with agents that possess human-like appearances and implement LLMs’ advanced conversational skills, which could expose users to Uncanny Valley or deception effects. By continuing to refine and test SEPs, we aim to enhance their effectiveness and ensure they provide meaningful support in promoting healthier lifestyles.

Data availability statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://doi.org/10.17605/OSF.IO/9T3CM.

Ethics statement

The studies involving humans were approved by Commission d’éthique de la recherche de la HEC (CER-HEC). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

AS: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing; MC: Funding acquisition, Methodology, Supervision, Writing – original draft, Writing – review & editing, Conceptualization; MC: Conceptualization, Funding acquisition, Methodology, Supervision, Writing – original draft, Writing – review & editing.

Funding

The author(s) declare that financial support was received for the research and publication of this article. This work was supported by a grant from the Ra&D Fund of the School of Management Fribourg (PARTISIM), which covered a portion of the main author's expenses and facilitated the acquisition of necessary materials. Participant compensation was jointly funded by PARTISIM and additional financial support provided by the HEC Faculty of the University of Lausanne.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declare that no Generative AI was used in the creation of this manuscript.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

1. ^Young adulthood is set as a period spanning from 18 to 25 years old (2, 6). Most researchers define young adulthood as a transitional period where individuals “participate in self-exploration to cultivate a personal identity and belief system, all the while gaining independence and autonomy” (6, p.3) such as is the case of university students (3).

2. ^See https://doi.org/10.17605/OSF.IO/9T3CM, last accessed November 2024.

3. ^See https://www.mypacer.com/, last accessed November 2024.

4. ^See https://doi.org/10.17605/OSF.IO/9T3CM, last accessed November 2024.

5. ^See https://doi.org/10.17605/OSF.IO/9T3CM, last accessed November 2024.

6. ^In our context this means the SEP would record steps like if it was a regular user.

7. ^See https://tamagotchi-official.com/us/, last accessed November 2024.

References

1. World Health Organization. Global Status Report on Physical Activity 2022 (Tech. rep.). World Health Organization (2022).

Google Scholar

2. Gavin J, Keough M, Abravanel M, Moudrakovski T, Mcbrearty M. Motivations for participation in physical activity across the lifespan. Int J Wellbeing. (2014) 4:46–61. doi: 10.5502/ijw.v4i1.3

Crossref Full Text | Google Scholar

3. Gibson A-M, Shaw J, Hewitt A, Easton C, Robertson S, Gibson N. A longitudinal examination of students’ health behaviours during their first year at university. J Further Higher Educ. (2018) 42:36–45. doi: 10.1080/0309877X.2016.1188902

Crossref Full Text | Google Scholar

4. Molanorouzi K, Khoo S, Morris T. Motives for adult participation in physical activity: type of activity, age, and gender. BMC Public Health. (2015) 15:66. doi: 10.1186/s12889-015-1429-7

PubMed Abstract | Crossref Full Text | Google Scholar

5. Ullrich-French S, Cox AE, Bumpus MF. Physical activity motivation and behavior across the transition to university. Sport Exerc Perform Psychol. (2013) 2:90–101. doi: 10.1037/a0030632

Crossref Full Text | Google Scholar

6. Higley E. Defining young adulthood. DNP Qualifying Manuscripts. 17 (2019). Available at: https://repository.usfca.edu/dnp_qualifying/17 (Accessed November 09, 2024).

Google Scholar

7. Allison KR, Dwyer JJM, Goldenberg E, Fein A, Yoshida KK, Boutilier M. Male adolescents’ reasons for participating in physical activity, barriers to participation, and suggestions for increasing participation. Adolescence. (2005) 40:155–70.15861623

PubMed Abstract | Google Scholar

8. Gómez-López M, Gallegos AG, Extremera AB. Perceived barriers by university students in the practice of physical activities. J Sports Sci Med. (2010) 9:374–81.

Google Scholar

9. Wing Kwan MY, Bray SR, Martin Ginis KA. Predicting physical activity of first-year university students: an application of the theory of planned behavior. J Am Coll Health. (2009) 58:45–55. doi: 10.3200/JACH.58.1.45-55

PubMed Abstract | Crossref Full Text | Google Scholar

10. Consolvo S, Everitt K, Smith I, Landay JA. Design requirements for technologies that encourage physical activity. In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, CHI ’06. New York, NY, USA: Association for Computing Machinery (2006). p. 457–66. doi: 10.1145/1124772.1124840

Crossref Full Text | Google Scholar

11. Agapie E, Chinh B, Pina LR, Oviedo D, Welsh MC, Hsieh G, et al.. Crowdsourcing exercise plans aligned with expert guidelines and everyday constraints. In: Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems, CHI ’18. New York, NY, USA: Association for Computing Machinery (2018). p. 1–13. doi: 10.1145/3173574.3173898

Crossref Full Text | Google Scholar

12. Al-Eisa E, Al-Rushud A, Alghadir A, Anwer S, Al-Harbi B, Al-Sughaier N, et al.. Effect of motivation by “instagram” on adherence to physical activity among female college students. Biomed Res Int. (2016) 2016:e1546013. doi: 10.1155/2016/1546013

Crossref Full Text | Google Scholar

13. Khalil A, Abdallah S. Harnessing social dynamics through persuasive technology to promote healthier lifestyle. Comput Human Behav. (2013) 29:2674–81. doi: 10.1016/j.chb.2013.07.008

Crossref Full Text | Google Scholar

14. Molina MD, Zhan ES, Agnihotri D, Abdullah S, Deka P. Motivation to use fitness application for improving physical activity among hispanic users: the pivotal role of interactivity and relatedness. In: Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, CHI ’23. New York, NY, USA: Association for Computing Machinery (2023). p. 1–13. doi: 10.1145/3544548.3581200

Crossref Full Text | Google Scholar

15. Stragier J, Mechant P, De Marez L, Cardon G. Computer-mediated social support for physical activity: a content analysis. Health Educ Behav. (2017) 45:124–31. doi: 10.1177/1090198117703055

PubMed Abstract | Crossref Full Text | Google Scholar

16. Hamer O, Larkin D, Relph N, Dey P. Fear as a barrier to physical activity in young adults with obesity: a qualitative study. Qual Res Sport Exerc Health. (2021) 15:18–34. doi: 10.1080/2159676X.2021.2012243

Crossref Full Text | Google Scholar

17. Mollee JS, Klein MCA. The effectiveness of upward and downward social comparison of physical activity in an online intervention. In: 2016 15th International Conference on Ubiquitous Computing and Communications and 2016 International Symposium on Cyberspace and Security (IUCC-CSS) (2016). p. 109–15. doi: 10.1109/IUCC-CSS.2016.023

Crossref Full Text | Google Scholar

18. La Guardia JG, Ryan RM, Couchman CE, Deci EL. Within-person variation in security of attachment: a self-determination theory perspective on attachment, need fulfillment, and well-being. J Pers Soc Psychol. (2000) 79:367–84. doi: 10.1037/0022-3514.79.3.367

PubMed Abstract | Crossref Full Text | Google Scholar

19. Orji R, Oyibo K, Lomotey RK, Orji FA. Socially-driven persuasive health intervention design: competition, social comparison, and cooperation. Health Informatics J. (2018) 25:1451–84. doi: 10.1177/1460458218766570

PubMed Abstract | Crossref Full Text | Google Scholar

20. Hahn L, Rathbun SL, Schmidt MD, Johnsen K, Annesi JJ, Ahn SJG. Using virtual agents and activity monitors to autonomously track and assess self-determined physical activity among young children: a 6-week feasibility field study. Cyberpsychol Behav Soc Netw. (2020) 23:471–8. doi: 10.1089/cyber.2019.0491

PubMed Abstract | Crossref Full Text | Google Scholar

21. Hahn L, Schmidt MD, Rathbun SL, Johnsen K, Annesi JJ, Ahn SJG. Using virtual agents to increase physical activity in young children with the virtual fitness buddy ecosystem: Study protocol for a cluster randomized trial. Contemp Clin Trials. (2020) 99:106181. doi: 10.1016/j.cct.2020.106181

PubMed Abstract | Crossref Full Text | Google Scholar

22. Bickmore TW, Caruso L, Clough-Gorr K, Heeren T. ‘It’s just like you talk to a friend’ relational agents for older adults. Interact Comput. (2005) 17:711–35. doi: 10.1016/j.intcom.2005.09.002

Crossref Full Text | Google Scholar

23. Salman S, Richards D, Dras M. Identifying which relational cues users find helpful to allow tailoring of e-coach dialogues. Multimodal Technol Interact. (2023) 7:93. doi: 10.3390/mti7100093

Crossref Full Text | Google Scholar

24. Morgan O. Approaches to increase physical activity: reviewing the evidence for exercise-referral schemes. Public Health. (2005) 119:361–70. doi: 10.1016/j.puhe.2004.06.008

PubMed Abstract | Crossref Full Text | Google Scholar

25. Fasola J, Mataric MJ. Using socially assistive human–robot interaction to motivate physical exercise for older adults. Proc IEEE. (2012) 100:2512–26. doi: 10.1109/JPROC.2012.2200539

Crossref Full Text | Google Scholar

26. Zhang J, Oh YJ, Lange P, Yu Z, Fukuoka Y. Artificial intelligence chatbot behavior change model for designing artificial intelligence chatbots to promote physical activity and a healthy diet: viewpoint. J Med Internet Res. (2020) 22:e22845. doi: 10.2196/22845

PubMed Abstract | Crossref Full Text | Google Scholar

27. Ahn SJG, Johnsen K, Robertson T, Moore J, Brown S, Marable A, et al.. Using virtual pets to promote physical activity in children: an application of the youth physical activity promotion model. J Health Commun. (2015) 20:807–15. doi: 10.1080/10810730.2015.1018597

PubMed Abstract | Crossref Full Text | Google Scholar

28. Kniestedt I, Gómez Maureira MA. Little fitness dragon: a gamified activity tracker. In: Wallner G, Kriglstein S, Hlavacs H, Malaka R, Lugmayr A, Yang H-S, editors. Entertainment Computing - ICEC 2016. Lecture Notes in Computer Science. Cham: Springer International Publishing (2016). p. 205–10. doi: 10.1007/978-3-319-46100-7_18

Crossref Full Text | Google Scholar

29. Tong X, Gromala D, Shaw C, Jin W. Encouraging physical activity with a game-based mobile application: FitPet. In: 2015 IEEE Games Entertainment Media Conference (GEM). Toronto, ON: IEEE (2015). p. 1–2. doi: 10.1109/GEM.2015.7377251

Crossref Full Text | Google Scholar

30. DeSmet A, De Bourdeaudhuij I, Chastin S, Crombez G, Maddison R, Cardon G. Adults’ preferences for behavior change techniques and engagement features in a mobile app to Promote 24-hour movement behaviors: cross-sectional survey study. JMIR Mhealth Uhealth. (2019) 7:e15707. doi: 10.2196/15707

PubMed Abstract | Crossref Full Text | Google Scholar

31. Organization Of Queerinai, Ovalle A, Subramonian A, Singh A, Voelcker C, Sutherland DJ, et al.. Queer in AI: A case study in community-led participatory AI. In: Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency, FAccT ’23. New York, NY, USA: Association for Computing Machinery (2023). p. 1882–95. doi: 10.1145/3593013.3594134

Crossref Full Text | Google Scholar

32. Smith RC, Bossen C, Kanstrup AM. Participatory design in an era of participation. CoDesign. (2017) 13:65–9. doi: 10.1080/15710882.2017.1310466

Crossref Full Text | Google Scholar

33. Spinuzzi C. The methodology of participatory design. Tech Commun. (2005) 52:163–74. Available at: https://www.ingentaconnect.com/content/stc/tc/2005/00000052/00000002/art00005

Google Scholar

34. Ryan RM, Deci EL. Self-Determination Theory: Basic Psychological Needs in Motivation, Development, and Wellness. New York, NY: The Guilford Press. (2017). doi: 10.1521/978.14625/28806

Crossref Full Text | Google Scholar

35. Deci EL, Ryan RM. The “what” and “why” of goal pursuits: human needs and the self-determination of behavior. Psychol Inq. (2000) 11:227–68. doi: 10.1207/S15327965PLI1104_01

Crossref Full Text | Google Scholar

36. Ballou N, Deterding S, Tyack A, Mekler ED, Calvo RA, Peters D, et al.. Self-determination theory in HCI: shaping a research agenda. In: Extended Abstracts of the 2022 CHI Conference on Human Factors in Computing Systems (CHI EA '22). New Orleans, LA: ACM (2022). p. 1–6. doi: 10.1145/3491101.3503702

Crossref Full Text | Google Scholar

37. Carpentier J, Mageau GA. When change-oriented feedback enhances motivation, well-being and performance: a look at autonomy-supportive feedback in sport. Psychol Sport Exerc. (2013) 14:423–35. doi: 10.1016/j.psychsport.2013.01.003

Crossref Full Text | Google Scholar

38. Curran T, Hill AP, Niemiec CP. A conditional process model of children’s behavioral engagement and behavioral disaffection in sport based on self-determination theory. J Sport Exerc Psychol. (2013) 35:30–43. doi: 10.1123/jsep.35.1.30

PubMed Abstract | Crossref Full Text | Google Scholar

39. Procter S, Mutrie N, Davis A, Audrey S. Views and experiences of behaviour change techniques to encourage walking to work: a qualitative study. BMC Public Health. (2014) 14:868. doi: 10.1186/1471-2458-14-868

PubMed Abstract | Crossref Full Text | Google Scholar

40. Emmerich K, Ring P, Masuch M. I’m glad you are on my side: how to design compelling game companions. In: Proceedings of the 2018 Annual Symposium on Computer-Human Interaction in Play (CHI PLAY '18). New York, NY, USA: Association for Computing Machinery (2018). p. 141–52. doi: 10.1145/3242671.3242709

Crossref Full Text | Google Scholar

41. Lee MS, Heeter C. What do you mean by believable characters? the effect of character rating and hostility on the perception of character believability. J Gaming Virtual Worlds. (2012) 4:81–97. doi: 10.1386/jgvw.4.1.81_1

Crossref Full Text | Google Scholar

42. Warpefelt H, Verhagen H. A model of non-player character believability. J Gaming Virtual Worlds. (2017) 9:39–53. doi: 10.1386/jgvw.9.1.39_1

Crossref Full Text | Google Scholar

43. Bickmore TW, Picard RW. Establishing and maintaining long-term human–computer relationships. ACM Trans Comput Hum Inter. (2005) 12:293–327. doi: 10.1145/1067860.1067867

Crossref Full Text | Google Scholar

44. Kamali ME, Angelini L, Caon M, Carrino F, Röcke C, Guye S, et al.. Virtual coaches for older adults’ wellbeing: a systematic review. IEEE Access. (2020) 8:101884–902. doi: 10.1109/ACCESS.2020.2996404

Crossref Full Text | Google Scholar

45. Rapp A, Boldi A, Curti L, Perrucci A, Simeoni R. How do people ascribe humanness to chatbots? an analysis of real-world human-agent interactions and a theoretical model of humanness. Int J Hum Comput Interact. (2023) 40(19):6027–50. doi: 10.1080/10447318.2023.2247596

Crossref Full Text | Google Scholar

46. Mori M. The uncanny valley. Energy. (1970) 7:33–5.

Google Scholar

47. Sillice MA, Morokoff PJ, Ferszt G, Bickmore T, Bock BC, Lantini R, et al.. Using relational agents to promote exercise and sun protection: assessment of participants’ experiences with two interventions. J Med Internet Res. (2018) 20:e7640. doi: 10.2196/jmir.7640

Crossref Full Text | Google Scholar

48. Lippke S, Ziegelmann JP. Theory-based health behavior change: developing, testing, and applying theories for evidence-based interventions. Appl Psychol. (2008) 57:698–716. doi: 10.1111/j.1464-0597.2008.00339.x

Crossref Full Text | Google Scholar

49. Hoj TH, Covey EL, Jones AC, Haines AC, Hall PC, Crookston BT, et al.. How do apps work? an analysis of physical activity app users’ perceptions of behavior change mechanisms. JMIR Mhealth Uhealth. (2017) 5:e7206. doi: 10.2196/mhealth.7206

Crossref Full Text | Google Scholar

50. Rhodes RE, McEwan D, Rebar AL. Theories of physical activity behaviour change: a history and synthesis of approaches. Psychol Sport Exerc. (2019) 42:100–9. doi: 10.1016/j.psychsport.2018.11.010

Crossref Full Text | Google Scholar

51. Brug J, Oenema A, Ferreira I. Theory, evidence and intervention mapping to improve behavior nutrition and physical activity interventions. Int J Behav Nutr Phys Act. (2005) 2:2. doi: 10.1186/1479-5868-2-2

PubMed Abstract | Crossref Full Text | Google Scholar

52. Bluethmann SM, Bartholomew LK, Murphy CC, Vernon SW. Use of theory in behavior change interventions: an analysis of programs to increase physical activity in posttreatment breast cancer survivors. Health Educ Behav. (2016) 44:245–53. doi: 10.1177/1090198116647712

PubMed Abstract | Crossref Full Text | Google Scholar

53. Bødker S. When second wave HCI meets third wave challenges. In: Proceedings of the 4th Nordic Conference on Human–Computer Interaction: Changing Roles, NordiCHI ’06. New York, NY, USA: Association for Computing Machinery (2006). p. 1–8. doi: 10.1145/1182475.1182476

Crossref Full Text | Google Scholar

54. Stephanidis C, Salvendy G, Antona M, Chen JYC, Dong J, Duffy VG, et al.. Seven HCI grand challenges. Int J Hum Comput Interact. (2019) 35:1229–69. doi: 10.1080/10447318.2019.1619259

Crossref Full Text | Google Scholar

55. Ntoumanis N, Thørgersen-Ntoumani C, Quested E, Chatzisarantis N. Theoretical approaches to physical activity promotion. In: Oxford Research Encyclopedia of Psychology. (2018). p. 1–34. doi: 10.1093/acrefore/9780190236557.013.212

Crossref Full Text | Google Scholar

56. Rhodes RE, Pfaeffli LA. Mediators of physical activity behaviour change among adult non-clinical populations: a review update. Int J Behav Nutr Phys Act. (2010) 7:37. doi: 10.1186/1479-5868-7-37

PubMed Abstract | Crossref Full Text | Google Scholar

57. Bandura A. Health promotion from the perspective of social cognitive theory. Psychol Health. (1998) 13:623–49. doi: 10.1080/08870449808407422

Crossref Full Text | Google Scholar

58. Prochaska JO, Velicer WF. The transtheoretical model of health behavior change. Am J Health Promot. (1997) 12:38–48. doi: 10.4278/0890-1171-12.1.38

PubMed Abstract | Crossref Full Text | Google Scholar

59. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Process. (1991) 50:179–211. doi: 10.1016/0749-5978(91)90020-T

Crossref Full Text | Google Scholar

60. McAuley E, Blissmer B. Self-efficacy determinants and consequences of physical activity. Exerc Sport Sci Rev. (2000) 28:85–8.10902091

PubMed Abstract | Google Scholar

61. Williams DM, Anderson ES, Winett RA. A review of the outcome expectancy construct in physical activity research. Ann Behav Med. (2005) 29:70–9. doi: 10.1207/s15324796abm2901_10

PubMed Abstract | Crossref Full Text | Google Scholar

62. Young MD, Plotnikoff RC, Collins CE, Callister R, Morgan PJ. Social cognitive theory and physical activity: a systematic review and meta-analysis. Obes Rev. (2014) 15:983–95. doi: 10.1111/obr.12225

PubMed Abstract | Crossref Full Text | Google Scholar

63. Webb TL, Sheeran P. Does changing behavioral intentions engender behavior change? a meta-analysis of the experimental evidence. Psychol Bull. (2006) 132:249–68. doi: 10.1037/0033-2909.132.2.249

PubMed Abstract | Crossref Full Text | Google Scholar

64. Chatzisarantis NLD, Hagger MS. Effects of a brief intervention based on the theory of planned behavior on leisure-time physical activity participation. J Sport Exerc Psychol. (2005) 27:470–87. doi: 10.1123/jsep.27.4.470

Crossref Full Text | Google Scholar

65. Armitage CJ. Can the theory of planned behavior predict the maintenance of physical activity? Health Psychol. (2005) 24:235–45. doi: 10.1037/0278-6133.24.3.235

PubMed Abstract | Crossref Full Text | Google Scholar

66. Nigg CR, Geller KS, Motl RW, Horwath CC, Wertin KK, Dishman RK. A research agenda to examine the efficacy and relevance of the transtheoretical model for physical activity behavior. Psychol Sport Exerc. (2011) 12:7–12. doi: 10.1016/j.psychsport.2010.04.004

PubMed Abstract | Crossref Full Text | Google Scholar

67. Riemsma RP, Pattenden J, Bridle C, Sowden AJ, Mather L, Watt IS, et al.. A systematic review of the effectiveness of interventions based on a stages-of-change approach to promote individual behaviour change in health care settings. Health Technol Assess (Rockv). (2002) 6. doi: 10.3310/hta6240

Crossref Full Text | Google Scholar

68. Adams J, White M. Why don’t stage-based activity promotion interventions work? Health Educ Res. (2005) 20:237–43. doi: 10.1093/her/cyg105

PubMed Abstract | Crossref Full Text | Google Scholar

69. Thøgersen-Ntoumani C. An ecological model of predictors of stages of change for physical activity in Greek older adults. Scand J Med Sci Sports. (2009) 19:286–96. doi: 10.1111/j.1600-0838.2007.00751.x

PubMed Abstract | Crossref Full Text | Google Scholar

70. Aldenaini N, Oyebode O, Orji R, Sampalli S. Mobile phone-based persuasive technology for physical activity and sedentary behavior: a systematic review. Front Comput Sci. (2020) 2:19. doi: 10.3389/fcomp.2020.00019

Crossref Full Text | Google Scholar

71. Teixeira PJ, Carraça EV, Markland D, Silva MN, Ryan RM. Exercise, physical activity, and self-determination theory: a systematic review. Int J Behav Nutr Phys Act. (2012) 9:78. doi: 10.1186/1479-5868-9-78

PubMed Abstract | Crossref Full Text | Google Scholar

72. Newman MW, Lauterbach D, Munson SA, Resnick P, Morris ME. It’s not that i don’t have problems, i’m just not putting them on facebook: challenges and opportunities in using online social networks for health. In: Proceedings of the ACM 2011 Conference on Computer Supported Cooperative Work (CSCW '11). New York, NY: Association for Computing Machinery (2011). p. 341–50. doi: 10.1145/1958824.1958876

Crossref Full Text | Google Scholar

73. Deci EL, Intrinsic Motivation. Boston, MA: Springer US (1975).

Google Scholar

74. Csikszentmihalyi M, Flow: The Psychology of Optimal Experience (Global Learning Communities). New York: Harper Perennial Edn (2008).

Google Scholar

75. Hassenzahl M, Eckoldt K, Diefenbach S, Laschke M, Lenz E, Kim J. Designing moments of meaning and pleasure. Int J Des. (2013) 7:21–31.

Google Scholar

76. Peters D, Calvo RA, Ryan RM. Designing for motivation, engagement and wellbeing in digital experience. Front Psychol. (2018) 9:797. doi: 10.3389/fpsyg.2018.00797

PubMed Abstract | Crossref Full Text | Google Scholar

77. Villalobos-Zúñiga G, Cherubini M. Apps that motivate: a taxonomy of app features based on self-determination theory. Int J Hum Comput Stud. (2020) 140:102449. doi: 10.1016/j.ijhcs.2020.102449

Crossref Full Text | Google Scholar

78. Jansen A, Van Mechelen M, Slegers K. Personas and behavioral theories: a case study using self-determination theory to construct overweight personas. In: Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, CHI ’17. New York, NY, USA: Association for Computing Machinery (2017). p. 2127–36. doi: 10.1145/3025453.3026003

Crossref Full Text | Google Scholar

79. Yang X, Aurisicchio M. Designing conversational agents: a self-determination theory approach. In: Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems, CHI ’21. New York, NY, USA: Association for Computing Machinery (2021). p. 1–16. doi: 10.1145/3411764.3445445

Crossref Full Text | Google Scholar

80. Feil-Seifer D, Mataric M. Defining socially assistive robotics. In: 9th International Conference on Rehabilitation Robotics, 2005. ICORR 2005 (2005). p. 465–8. doi: 10.1109/ICORR.2005.1501143

Crossref Full Text | Google Scholar

81. Fong T, Nourbakhsh I, Dautenhahn K. A survey of socially interactive robots. Rob Auton Syst. (2003) 42:143–66. doi: 10.1016/S0921-8890(02)00372-X

Crossref Full Text | Google Scholar

82. Fasola J, Matarić MJ. Robot motivator: increasing user enjoyment and performance on a physical/cognitive task. In: 2010 IEEE 9th International Conference on Development and Learning (2010). p. 274–9. doi: 10.1109/DEVLRN.2010.5578830

Crossref Full Text | Google Scholar

83. Fasola J, Matarić MJ. A socially assistive robot exercise coach for the elderly. J Hum Robot Interact. (2013) 2:3–32. doi: 10.5898/JHRI.2.2.Fasola

Crossref Full Text | Google Scholar

84. Lin JJ, Mamykina L, Lindtner S, Delajoux G, Strub HB. Fish’n’Steps: encouraging physical activity with an interactive computer game. In: Dourish P, Friday A, editors. UbiComp 2006: Ubiquitous Computing. Lecture Notes in Computer Science. Berlin, Heidelberg: Springer (2006). Vol. 4206, p. 261–78. doi: 10.1007/11853565_16

Crossref Full Text | Google Scholar

85. Birk MV, Atkins C, Bowey JT, Mandryk RL. Fostering intrinsic motivation through avatar identification in digital games. In: Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems, CHI ’16. New York, NY, USA: Association for Computing Machinery (2016). p. 2982–95. doi: 10.1145/2858036.2858062

Crossref Full Text | Google Scholar

86. Birk MV, Mandryk RL. Combating attrition in digital self-improvement programs using avatar customization. In: Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems, CHI ’18. New York, NY, USA: Association for Computing Machinery (2018). p. 1–15. doi: 10.1145/3173574.3174234

Crossref Full Text | Google Scholar

87. Lankoski P, Björk S. Gameplay design patterns for believable non-player characters. In: DiGRA ’07 - Proceedings of the 2007 DiGRA International Conference: Situated Play (2007). p. 416–23. doi: 10.26503/dl.v2007i1.262

Crossref Full Text | Google Scholar

88. Peña J, Kim E. Increasing exergame physical activity through self and opponent avatar appearance. Comput Human Behav. (2014) 41:262–7. doi: 10.1016/j.chb.2014.09.038

Crossref Full Text | Google Scholar

89. Warpefelt H, Johansson M, Verhagen H. Analyzing the believability of game character behavior using the game agent matrix. In: Proceedings of DiGRA 2013 Conference (2013). p. 1–11.

Google Scholar

90. Pimentel D, Kalyanaraman S. Your own worst enemy: implications of the customization, and destruction, of non-player characters. In: Proceedings of the Annual Symposium on Computer-Human Interaction in Play (CHI PLAY '20). Virtual Event Canada: ACM (2020). p. 93–106. doi: 10.1145/3410404.3414269

Crossref Full Text | Google Scholar

91. Trepte S, Reinecke L. Avatar creation and video game enjoyment: effects of life-satisfaction, game competitiveness, and identification with the avatar. J Med Psychol Theor Methods Appl. (2010) 22:171–84. doi: 10.1027/1864-1105/a000022

Crossref Full Text | Google Scholar

92. Van Looy J, Courtois C, De Vocht M, De Marez L. Player identification in online games: validation of a scale for measuring identification in MMOGs. Media Psychol. (2012) 15:197–221. doi: 10.1080/15213269.2012.674917

Crossref Full Text | Google Scholar

93. Hao W, Yu-Chun R, Sheng-Yi H, Chun-Tsai S. Effects of game design features on player-avatar relationships and motivation for buying decorative virtual items. In: Proceedings of DiGRA 2019 Conference: Game, Play and the Emerging Ludo-Mix (DiGRA) (2019). p. 1–22. doi: 10.26503/dl.v2019i1.1107

Crossref Full Text | Google Scholar

94. Kinnafick F-E, Thøgersen-Ntoumani C, Shepherd SO, Wilson OJ, Wagenmakers AJ, Shaw CS. In it together: a qualitative evaluation of participant experiences of a 10-week, group-based, workplace HIIT program for insufficiently active adults. J Sport Exerc Psychol. (2018) 40:10–9. doi: 10.1123/jsep.2017-0306

PubMed Abstract | Crossref Full Text | Google Scholar

95. Lovell GP, Gordon JAR, Mueller MB, Mulgrew K, Sharman R. Satisfaction of basic psychological needs, self-determined exercise motivation, and psychological well-being in mothers exercising in group-based versus individual-based contexts. Health Care Women Int. (2015) 37:568–82. doi: 10.1080/07399332.2015.1078333

PubMed Abstract | Crossref Full Text | Google Scholar

96. Sanz-Remacha M, Aibar A, Sevil-Serrano J, García-González L. Evaluation of a 20-month physical activity intervention to improve motivational and affective outcomes among disadvantaged adult women. Qual Health Res. (2021) 31:1392–403. doi: 10.1177/1049732321997136

PubMed Abstract | Crossref Full Text | Google Scholar

97. Estabrooks PA, Harden SM, Burke SM. Group dynamics in physical activity promotion: what works? Soc Personal Psychol Compass. (2012) 6:18–40. doi: 10.1111/j.1751-9004.2011.00409.x

Crossref Full Text | Google Scholar

98. Almutari N, Orji R. How effective are social influence strategies in persuasive apps for promoting physical activity? a systematic review. In: Adjunct Publication of the 27th Conference on User Modeling, Adaptation and Personalization, UMAP’19 Adjunct. New York, NY, USA: Association for Computing Machinery (2019). p. 167–72. doi: 10.1145/3314183.3323855

Crossref Full Text | Google Scholar

99. Yoganathan D, Kajanan S. Persuasive technology for smartphone fitness apps. In: PACIS 2013 Proceedings (2013). p. 185.

Google Scholar

100. Wolf T, Jahn S, Hammerschmidt M, Weiger WH. Competition versus cooperation: how technology-facilitated social interdependence initiates the self-improvement chain. Int J Res Mark. (2021) 38(2):472–91. doi: 10.1016/j.ijresmar.2020.06.001

Crossref Full Text | Google Scholar

101. Johnson DW, Johnson RT, Cooperation and Competition: Theory and Research. Cooperation and Competition: Theory and Research. Edina, MN, US: Interaction Book Company (1989).

Google Scholar

102. Tauer JM, Harackiewicz JM. The effects of cooperation and competition on intrinsic motivation and performance. J Pers Soc Psychol. (2004) 86:849–61. doi: 10.1037/0022-3514.86.6.849

PubMed Abstract | Crossref Full Text | Google Scholar

103. Bouncken RB, Gast J, Kraus S, Bogers M. Coopetition: a systematic review, synthesis, and future research directions. Rev Manage Sci. (2015) 9:577–601. doi: 10.1007/s11846-015-0168-6

Crossref Full Text | Google Scholar

104. Chin K-S, Chan BL, Lam P-K. Identifying and prioritizing critical success factors for coopetition strategy. Ind Manage Data Syst. (2008) 108:437–54. doi: 10.1108/02635570810868326

Crossref Full Text | Google Scholar

105. Klein MCA, Manzoor A, Mollee JS. Active2Gether: a personalized m-health intervention to encourage physical activity. Sensors. (2017) 17:1436. doi: 10.3390/s17061436

PubMed Abstract | Crossref Full Text | Google Scholar

106. Festinger L. A theory of social comparison processes. Hum Relat. (1954) 7:117–40. doi: 10.1177/001872675400700202

Crossref Full Text | Google Scholar

107. Zhu J, Dallal DH, Gray RC, Villareale J, Ontañón S, Forman EM, et al. Personalization paradox in behavior change apps: lessons from a social comparison-based personalized app for physical activity. Proc ACM Hum-Comput Interact. (2021) 5:116. doi: 10.1145/3449190

Crossref Full Text | Google Scholar

108. Zuckerman O, Gal-Oz A. Deconstructing gamification: evaluating the effectiveness of continuous measurement, virtual rewards, and social comparison for promoting physical activity. Pers Ubiquitous Comput. (2014) 18:1705–19. doi: 10.1007/s00779-014-0783-2

Crossref Full Text | Google Scholar

109. Neighbors C, Raymond Knee C. Self-determination and the consequences of social comparison. J Res Pers. (2003) 37:529–46. doi: 10.1016/S0092-6566(03)00047-3

Crossref Full Text | Google Scholar

110. Guckelsberger C, Salge C, Colton S. Intrinsically motivated general companion NPCs via coupled empowerment maximisation. In: 2016 IEEE Conference on Computational Intelligence and Games (CIG) (2016). p. 1–8. doi: 10.1109/CIG.2016.7860406

Crossref Full Text | Google Scholar

111. Bailey C, Katchabaw M. An emergent framework for realistic psychosocial behaviour in non player characters. In: Proceedings of the 2008 Conference on Future Play: Research, Play, Share. Toronto Ontario Canada: ACM (2008). p. 17–24. doi: 10.1145/1496984.1496988

Crossref Full Text | Google Scholar

112. Beinema T, op den Akker H, van Velsen L, Hermens H. Tailoring coaching strategies to users’ motivation in a multi-agent health coaching application. Comput Human Behav. (2021) 121:106787. doi: 10.1016/j.chb.2021.106787

Crossref Full Text | Google Scholar

113. Gabrielli S, Marie K, Corte CD. SLOWBot (chatbot) lifestyle assistant. In: Proceedings of the 12th EAI International Conference on Pervasive Computing Technologies for Healthcare, PervasiveHealth ’18. New York, NY, USA: Association for Computing Machinery (2018). p. 367–70. doi: 10.1145/3240925.3240953

Crossref Full Text | Google Scholar

114. Kocielnik R, Xiao L, Avrahami D, Hsieh G. Reflection companion: a conversational system for engaging users in reflection on physical activity. Proc ACM Interact Mob Wear Ubiquitous Technol. (2018) 2:70:1–70:26. doi: 10.1145/3214273

Crossref Full Text | Google Scholar

115. Kowatsch T, Volland D, Shih I, Rüegger D, Künzler F, Barata F, et al.. Design and evaluation of a mobile chat app for the open source behavioral health intervention platform MobileCoach. In: Maedche A, vom Brocke J, Hevner A, editors. Designing the Digital Transformation. Lecture Notes in Computer Science. Cham: Springer International Publishing (2017). Vol. 10243, p. 485–9. doi: 10.1007/978-3-319-59144-5_36

Crossref Full Text | Google Scholar

116. Kramer J-N, Künzler F, Mishra V, Smith SN, Kotz D, Scholz U, et al.. Which components of a smartphone walking app help users to reach personalized step goals? results from an optimization trial. Ann Behav Med. (2020) 54:518–28. doi: 10.1093/abm/kaaa002

PubMed Abstract | Crossref Full Text | Google Scholar

117. Stein N, Brooks K. A fully automated conversational artificial intelligence for weight loss: longitudinal observational study among overweight and obese adults. JMIR Diab. (2017) 2:e8590. doi: 10.2196/diabetes.8590

Crossref Full Text | Google Scholar

118. Nahum-Shani I, Smith SN, Spring BJ, Collins LM, Witkiewitz K, Tewari A, et al.. Just-in-time adaptive interventions (JITAIs) in mobile health: key components and design principles for ongoing health behavior support. Ann Behav Med Publ Soc Behav Med. (2018) 52:446–62. doi: 10.1007/s12160-016-9830-8

PubMed Abstract | Crossref Full Text | Google Scholar

119. Fadhil A, Wang Y, Reiterer H. Assistive conversational agent for health coaching: a validation study. Methods Inf Med. (2019) 58:9–23. doi: 10.1055/s-0039-1688757

PubMed Abstract | Crossref Full Text | Google Scholar

120. Maher CA, Davis CR, Curtis RG, Short CE, Murphy KJ. A physical activity and diet program delivered by artificially intelligent virtual health coach: proof-of-concept study. JMIR Mhealth Uhealth. (2020) 8:e17558. doi: 10.2196/17558

PubMed Abstract | Crossref Full Text | Google Scholar

121. op den Akker H, Cabrita M, Jones VM, Hermens HJ. Tailored motivational message generation. J Biomed Inform. (2015) 55:104–15. doi: 10.1016/j.jbi.2015.03.005

PubMed Abstract | Crossref Full Text | Google Scholar

122. Schulman D, Bickmore T. Persuading users through counseling dialogue with a conversational agent. In: Proceedings of the 4th International Conference on Persuasive Technology, Persuasive ’09. New York, NY, USA: Association for Computing Machinery (2009). p. 1–8. doi: 10.1145/1541948.1541983

Crossref Full Text | Google Scholar

123. Mollee JS, Middelweerd A, Klein MCA. Evaluation of a personalized coaching system for physical activity: user appreciation and adherence. In: Proceedings of the 11th EAI International Conference on Pervasive Computing Technologies for Healthcare, PervasiveHealth ’17. New York, NY, USA: Association for Computing Machinery (2017). p. 315–24. doi: 10.1145/3154862.3154933

Crossref Full Text | Google Scholar

124. Schäfer H, Bachner J, Pretscher S, Groh G, Demetriou Y. Study on motivating physical activity in children with personalized gamified feedback. In: Adjunct Publication of the 26th Conference on User Modeling, Adaptation and Personalization, UMAP ’18. New York, NY, USA: Association for Computing Machinery (2018). p. 221–6. doi: 10.1145/3213586.3225227

Crossref Full Text | Google Scholar

125. Umarov I, Mozgovoy M. Believable and effective AI agents in virtual worlds: current state and future perspectives. Int J Gaming Comput Mediated Simul (IJGCMS). (2012) 4:37–59. doi: 10.4018/jgcms.2012040103

Crossref Full Text | Google Scholar

126. Brendel AB, Hildebrandt F, Dennis AR, Riquel J. The paradoxical role of humanness in aggression toward conversational agents. J Manage Inf Syst. (2023) 40:883–913. doi: 10.1080/07421222.2023.2229127

Crossref Full Text | Google Scholar

127. Groom V, Srinivasan V, Bethel CL, Murphy R, Dole L, Nass C. Responses to robot social roles and social role framing. In: 2011 International Conference on Collaboration Technologies and Systems (CTS) (2011). p. 194–203. doi: 10.1109/CTS.2011.5928687

Crossref Full Text | Google Scholar

128. Rapp A, Curti L, Boldi A. The human side of human-chatbot interaction: a systematic literature review of ten years of research on text-based chatbots. Int J Hum Comput Stud. (2021) 151:102630. doi: 10.1016/j.ijhcs.2021.102630

Crossref Full Text | Google Scholar

129. Schuetzler RM, Grimes GM, Giboney JS. The effect of conversational agent skill on user behavior during deception. Comput Human Behav. (2019) 97:250–9. doi: 10.1016/j.chb.2019.03.033

Crossref Full Text | Google Scholar

130. Fuller CM, Marett K, Twitchell DP. An examination of deception in virtual teams: effects of deception on task performance, mutuality, and trust. IEEE Trans Prof Commun. (2012) 55:20–35. doi: 10.1109/TPC.2011.2172731

Crossref Full Text | Google Scholar

131. Kanstrup AM, Bertelsen P. Bringing new voices to design of exercise technology: participatory design with vulnerable young adults. In: Proceedings of the 14th Participatory Design Conference: Full Papers - Volume 1, PDC ’16. New York, NY, USA: Association for Computing Machinery (2016). p. 121–30. doi: 10.1145/2940299.2940305

Crossref Full Text | Google Scholar

132. Kanstrup AM, Bertelsen P. Design for healthy horizons in a local community: digital relations in a neighbourhood with health challenges. In: Proceedings of the 9th International Conference on Communities & Technologies - Transforming Communities, C&T ’19. New York, NY, USA: Association for Computing Machinery (2019). p. 41–50. doi: 10.1145/3328320.3328370

Crossref Full Text | Google Scholar

133. LaMonica HM, Davenport TA, Roberts AE, Hickie IB. Understanding technology preferences and requirements for health information technologies designed to improve and maintain the mental health and well-being of older adults: participatory design study. JMIR Aging. (2021) 4:e21461. doi: 10.2196/21461

PubMed Abstract | Crossref Full Text | Google Scholar

134. Martin A, Caon M, Adorni F, Andreoni G, Ascolese A, Atkinson S, et al.. A mobile phone intervention to improve obesity-related health behaviors of adolescents across Europe: iterative co-design and feasibility study. JMIR Mhealth Uhealth. (2020) 8:e14118. doi: 10.2196/14118

PubMed Abstract | Crossref Full Text | Google Scholar

135. Revenäs Å, Opava CH, Åsenlöf P. Lead users’ ideas on core features to support physical activity in rheumatoid arthritis: a first step in the development of an internet service using participatory design. BMC Med Inform Decis Mak. (2014) 14:21. doi: 10.1186/1472-6947-14-21

Crossref Full Text | Google Scholar

136. Wiklund Axelsson S, Wikberg-Nilsson Å, Melander Wikman A. Sustainable lifestyle change—participatory design of support together with persons with obesity in the third age. Int J Environ Res Public Health. (2016) 13:1248. doi: 10.3390/ijerph13121248

PubMed Abstract | Crossref Full Text | Google Scholar

137. Martins J, Rodrigues A, Marques A, Cale L, da Costa FC. Adolescents’ experiences and perspectives on physical activity and friend influences over time. Res Q Exerc Sport. (2020) 92(3):399–410. doi: 10.1080/02701367.2020.1739607

PubMed Abstract | Crossref Full Text | Google Scholar

138. Kim S, Li M. Awareness, understanding, and action: a conceptual framework of user experiences and expectations about indoor air quality visualizations. In: Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, CHI ’20. New York, NY, USA: Association for Computing Machinery (2020). p. 1–12. doi: 10.1145/3313831.3376521

Crossref Full Text | Google Scholar

139. Janols R, Sandlund M, Lindgren H, Pettersson B. Older adults as designers of behavior change strategies to increase physical activity—report of a participatory design process. Front Public Health. (2022) 10:988470. doi: 10.3389/fpubh.2022.988470

PubMed Abstract | Crossref Full Text | Google Scholar

140. MacDorman KF, Green RD, Ho C-C, Koch CT. Too real for comfort? uncanny responses to computer generated faces. Comput Human Behav. (2009) 25:695–710. doi: 10.1016/j.chb.2008.12.026

PubMed Abstract | Crossref Full Text | Google Scholar

141. McDonnell R, Breidt M, Bülthoff HH. Render me real? investigating the effect of render style on the perception of animated virtual humans. ACM Trans Graph. (2012) 31(4):91. doi: 10.1145/2185520.2185587

Crossref Full Text | Google Scholar

142. Zhou MX, Mark G, Li J, Yang H. Trusting virtual agents: the effect of personality. ACM Trans Interact Intell Syst. (2019) 9(2–3):10. doi: 10.1145/3232077

Crossref Full Text | Google Scholar

143. Stein J-P, Appel M, Jost A, Ohler P. Matter over mind? how the acceptance of digital entities depends on their appearance, mental prowess, and the interaction between both. Int J Hum Comput Stud. (2020) 142:102463. doi: 10.1016/j.ijhcs.2020.102463

Crossref Full Text | Google Scholar

144. Meurisch C, Mihale-Wilson CA, Hawlitschek A, Giger F, Müller F, Hinz O, et al.. Exploring user expectations of proactive AI systems. Proc ACM Interact Mob Wear Ubiquitous Technol. (2020) 4(4):146. doi: 10.1145/3432193

Crossref Full Text | Google Scholar

145. Sanders EB-N. From user-centered to participatory design approaches. In: Frascara J, editor. Design and the Social Sciences. London: CRC Press (2002). p. 1–8.

Google Scholar

146. Björling EA, Rose E. Participatory research principles in human-centered design: engaging teens in the co-design of a social robot. Multimodal Technol Interact. (2019) 3:8. doi: 10.3390/mti3010008

Crossref Full Text | Google Scholar

147. Browne JT. Wizard of oz prototyping for machine learning experiences. In: Extended Abstracts of the 2019 CHI Conference on Human Factors in Computing Systems, CHI EA ’19. New York, NY, USA: Association for Computing Machinery (2019). p. 1–6. doi: 10.1145/3290607.3312877

Crossref Full Text | Google Scholar

148. Jung MF, Martelaro N, Hoster H, Nass C. Participatory materials: having a reflective conversation with an artifact in the making. In: Proceedings of the 2014 Conference on Designing Interactive Systems, DIS ’14. New York, NY, USA: Association for Computing Machinery (2014). p. 25–34. doi: 10.1145/2598510.2598591

Crossref Full Text | Google Scholar

149. Lee S, Kim S, Lee S. “What does your agent look like?”: a drawing study to understand users’ perceived persona of conversational agent. In: Extended Abstracts of the 2019 CHI Conference on Human Factors in Computing Systems, CHI EA ’19. New York, NY, USA: Association for Computing Machinery (2019). p. 1–6. doi: 10.1145/3290607.3312796

Crossref Full Text | Google Scholar

150. Strömberg H, Pettersson I, Andersson J, Rydström A, Dey D, Klingegård M, et al.. Designing for social experiences with and within autonomous vehicles – exploring methodological directions. Des Sci. (2018) 4:e13. doi: 10.1017/dsj.2018.9

Crossref Full Text | Google Scholar

151. Winkle K, Senft E, Lemaignan S. LEADOR: a method for end-to-end participatory design of autonomous social robots. Front Robot AI. (2021) 8:704119. doi: 10.3389/frobt.2021.704119

PubMed Abstract | Crossref Full Text | Google Scholar

152. Couture J. Reflections from the ‘Strava-sphere’: Kudos, community, and (self-)surveillance on a social network for athletes. Qual Res Sport Exerc Health. (2020) 13:184–200. doi: 10.1080/2159676X.2020.1836514

Crossref Full Text | Google Scholar

153. Braun V, Clarke V. Using thematic analysis in psychology. Qual Res Psychol. (2006) 3:77–101. doi: 10.1191/1478088706qp063oa

Crossref Full Text | Google Scholar

154. McHugh ML. Interrater reliability: the kappa statistic. Biochem Med (Zagreb). (2012) 22:276–82. doi: 10.11613/BM.2012.031

PubMed Abstract | Crossref Full Text | Google Scholar

155. Pradhan A, Lazar A. Hey google, do you have a personality? Designing personality and personas for conversational agents. In: CUI 2021 - 3rd Conference on Conversational User Interfaces. Bilbao (online) Spain: ACM (2021). p. 1–4. doi: 10.1145/3469595.3469607

Crossref Full Text | Google Scholar

156. Nass C, Steuer J, Tauber ER. Computers are social actors. In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, CHI ’94. New York, NY, USA: Association for Computing Machinery (1994). p. 72–8. doi: 10.1145/191666.191703

Crossref Full Text | Google Scholar

157. Reeves B, Nass CI, The Media Equation: How People Treat Computers, Television, and New Media like Real People and Places. New York, NY, US: Cambridge University Press (1996).

Google Scholar

158. Lopatovska I, Williams H. Personification of the Amazon Alexa: BFF or a mindless companion. In: Proceedings of the 2018 Conference on Human Information Interaction & Retrieval, CHIIR ’18. New York, NY, USA: Association for Computing Machinery (2018). p. 265–8. doi: 10.1145/3176349.3176868

Crossref Full Text | Google Scholar

159. Pradhan A, Findlater L, Lazar A. “Phantom friend” or “just a box with information”: personification and ontological categorization of smart speaker-based voice assistants by older adults. Proc ACM Hum Comput Interact. (2019) 3:214. doi: 10.1145/3359316

Crossref Full Text | Google Scholar

160. Purington A, Taft JG, Sannon S, Bazarova NN, Taylor SH. “Alexa is my new BFF”: social roles, user satisfaction, and personification of the Amazon echo. In: Proceedings of the 2017 CHI Conference Extended Abstracts on Human Factors in Computing Systems, CHI EA ’17. New York, NY, USA: Association for Computing Machinery (2017). p. 2853–9. doi: 10.1145/3027063.3053246

Crossref Full Text | Google Scholar

161. Cowan BR, Pantidi N, Coyle D, Morrissey K, Clarke P, Al-Shehri S, et al.. “What can i help you with?”: infrequent users’ experiences of intelligent personal assistants. In: Proceedings of the 19th International Conference on Human–Computer Interaction with Mobile Devices and Services, MobileHCI ’17. New York, NY, USA: Association for Computing Machinery (2017). p. 1–12. doi: 10.1145/3098279.3098539

Crossref Full Text | Google Scholar

162. Turing AM. Computing machinery and intelligence. Mind. (1950) LIX:433–60. doi: 10.1093/mind/LIX.236.433

Crossref Full Text | Google Scholar

163. Castelfranchi C. Artificial liars: why computers will (necessarily) deceive us and each other. Ethics Inf Technol. (2000) 2:113–9. doi: 10.1023/A:1010025403776

Crossref Full Text | Google Scholar

164. Heckman CE, Wobbrock JO. Put your best face forward: anthropomorphic agents, e-commerce consumers, and the law. In: Proceedings of the Fourth International Conference on Autonomous Agents, AGENTS ’00. New York, NY, USA: Association for Computing Machinery (2000). p. 435–42. doi: 10.1145/336595.337562

Crossref Full Text | Google Scholar

165. Schuetzler RM, Grimes GM, Scott Giboney J. The impact of chatbot conversational skill on engagement and perceived humanness. J Manage Inf Syst. (2020) 37:875–900. doi: 10.1080/07421222.2020.1790204

Crossref Full Text | Google Scholar

166. Liu B, Sundar SS. Should machines express sympathy and empathy? experiments with a health advice chatbot. Cyberpsychol Behav Soc Netw. (2018) 21:625–36. doi: 10.1089/cyber.2018.0110

PubMed Abstract | Crossref Full Text | Google Scholar

167. Yee N, Bailenson JN, Ducheneaut N. The proteus effect: implications of transformed digital self-representation on online and offline behavior. Communic Res. (2009) 36:285–312. doi: 10.1177/0093650208330254

Crossref Full Text | Google Scholar

168. Bessière K, Seay AF, Kiesler S. The ideal elf: identity exploration in World of Warcraft. Cyberpsychol Behav. (2007) 10:530–5. doi: 10.1089/cpb.2007.9994

PubMed Abstract | Crossref Full Text | Google Scholar

169. Li BJ, Ratan R, Lwin MO. Virtual game changers: how avatars and virtual coaches influence exergame outcomes through enactive and vicarious learning. Behav Inf Technol. (2022) 41:1529–43. doi: 10.1080/0144929X.2021.1884290

Crossref Full Text | Google Scholar

170. Lim S, Reeves B. Being in the game: effects of avatar choice and point of view on psychophysiological responses during play. Media Psychol. (2009) 12:348–70. doi: 10.1080/15213260903287242

Crossref Full Text | Google Scholar

171. Li K, Nguyen HV, Cheng T, Teng C-I. How do avatar characteristics affect avatar friendliness and online gamer loyalty? perspective of the theory of embodied cognition. Internet Res. (2018) 28:1103–21. doi: 10.1108/IntR-06-2017-0246

Crossref Full Text | Google Scholar

172. Ellis T, Latham NK, DeAngelis TR, Thomas CA, Saint-Hilaire M, Bickmore TW. Feasibility of a virtual exercise coach to promote walking in community-dwelling persons with Parkinson disease. Am J Phys Med Rehabil. (2013) 92:472–85. doi: 10.1097/PHM.0b013e31828cd466

PubMed Abstract | Crossref Full Text | Google Scholar

173. Watson A, Bickmore T, Cange A, Kulshreshtha A, Kvedar J. An internet-based virtual coach to promote physical activity adherence in overweight adults: randomized controlled trial. J Med Internet Res. (2012) 14:e1629. doi: 10.2196/jmir.1629

Crossref Full Text | Google Scholar

174. Weimann TG, Schlieter H, Brendel AB. Virtual coaches. Bus Inf Syst Eng. (2022) 64:515–28. doi: 10.1007/s12599-022-00757-9

Crossref Full Text | Google Scholar

175. Ding D, Liu H-Y, Cooper R, Cooper RA, Smailagic A, Siewiorek D. Virtual coach technology for supporting self-care. Phys Med Rehabil Clin N Am. (2010) 21:179–94. doi: 10.1016/j.pmr.2009.07.012

PubMed Abstract | Crossref Full Text | Google Scholar

176. Tropea P, Schlieter H, Sterpi I, Judica E, Gand K, Caprino M, et al.. Rehabilitation, the great absentee of virtual coaching in medical care: scoping review. J Med Internet Res. (2019) 21:e12805. doi: 10.2196/12805

PubMed Abstract | Crossref Full Text | Google Scholar

177. Martin Seth S, Feldman David I, Blumenthal Roger S, Jones Steven R, Post Wendy S, McKibben Rebeccah A, et al.. mActive: a randomized clinical trial of an automated mHealth intervention for physical activity promotion. J Am Heart Assoc. (2015) 4:e002239. doi: 10.1161/JAHA.115.002239

PubMed Abstract | Crossref Full Text | Google Scholar

178. Wang JB, Cadmus-Bertram LA, Natarajan L, White MM, Madanat H, Nichols JF, et al.. Wearable sensor/device (fitbit one) and SMS text-messaging prompts to increase physical activity in overweight and obese adults: a randomized controlled trial. Telemed J E-Health Off J Am Telemed Assoc. (2015) 21:782–92. doi: 10.1089/tmj.2014.0176

PubMed Abstract | Crossref Full Text | Google Scholar

179. Schembre SM, Liao Y, Robertson MC, Dunton GF, Kerr J, Haffey ME, et al.. Just-in-time feedback in diet and physical activity interventions: systematic review and practical design framework. J Med Internet Res. (2018) 20:e106. doi: 10.2196/jmir.8701

PubMed Abstract | Crossref Full Text | Google Scholar

180. Ashktorab Z, Jain M, Liao QV, Weisz JD. Resilient chatbots: repair strategy preferences for conversational breakdowns. In: Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems (CHI '19). New York, NY: Association for Computing Machinery (2019). p. 1–12. doi: 10.1145/3290605.3300484

Crossref Full Text | Google Scholar

181. Hatfield EC, Pillemer JT, O’Brien MU, Le Y-CL. The endurance of love: passionate and companionate love in newlywed and long-term marriages. Interpers Int J Pers Relatsh. (2008) 2:35–64. doi: 10.5964/ijpr.v2i1.17

Crossref Full Text | Google Scholar

182. Loader BD, Muncer S, Burrows R, Pleace N, Nettleton S. Medicine on the line? computer-mediated social support and advice for people with diabetes. Int J Soc Welf. (2002) 11:53–65. doi: 10.1111/1468-2397.00196

Crossref Full Text | Google Scholar

183. Rook KS. Social support versus companionship: effects on life stress, loneliness, and evaluations by others. J Pers Soc Psychol. (1987) 52:1132–47. doi: 10.1037/0022-3514.52.6.1132

PubMed Abstract | Crossref Full Text | Google Scholar

184. Rook KS. Social relationships as a source of companionship: implications for older adults’ psychological well-being. In: Social Support: An Interactional View. Wiley Series on Personality Processes. Oxford, England: John Wiley & Sons (1990). p. 219–50.

Google Scholar

185. Wolever RQ, Simmons LA, Sforzo GA, Dill D, Kaye M, Bechard EM, et al.. A systematic review of the literature on health and wellness coaching: defining a key behavioral intervention in healthcare. Glob Adv Health Med. (2013) 2:38–57. doi: 10.7453/gahmj.2013.042

PubMed Abstract | Crossref Full Text | Google Scholar

186. Li M, Mao J. Hedonic or utilitarian? exploring the impact of communication style alignment on user’s perception of virtual health advisory services. Int J Inf Manage. (2015) 35:229–43. doi: 10.1016/j.ijinfomgt.2014.12.004

Crossref Full Text | Google Scholar

187. Luria M, Zheng R, Huffman B, Huang S, Zimmerman J, Forlizzi J. Social boundaries for personal agents in the interpersonal space of the home. In: Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, CHI ’20. New York, NY, USA: Association for Computing Machinery (2020). p. 1–12. doi: 10.1145/3313831.3376311

Crossref Full Text | Google Scholar

188. Lu J, Lopes P. Integrating living organisms in devices to implement care-based interactions. In: Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology, UIST ’22. New York, NY, USA: Association for Computing Machinery (2022). p. 1–13. doi: 10.1145/3526113.3545629

Crossref Full Text | Google Scholar

189. Turkle S. Authenticity in the age of digital companions. Interact Stud. (2007) 8:501–17. doi: 10.1075/is.8.3.11tur

Crossref Full Text | Google Scholar

190. Bloch L-R, Lemish D. Disposable love: the rise and fall of a virtual pet. New Med Soc. (1999) 1:283–303. doi: 10.1177/14614449922225591

Crossref Full Text | Google Scholar

191. van Dis EAM, Bollen J, Zuidema W, van Rooij R, Bockting CL. ChatGPT: five priorities for research. Nature. (2023) 614:224–6. doi: 10.1038/d41586-023-00288-7

PubMed Abstract | Crossref Full Text | Google Scholar

192. Li Y, Peng Y. What drives gift-giving intention in live streaming? the perspectives of emotional attachment and flow experience. Int J Hum Comput Interact. (2021) 37:1317–29. doi: 10.1080/10447318.2021.1885224

Crossref Full Text | Google Scholar

193. Wang W, Hang H. Exploring the eudaimonic game experience through purchasing functional and nonfunctional items in MMORPGs. Psychol Mark. (2021) 38:1847–62. doi: 10.1002/mar.21503

Crossref Full Text | Google Scholar

194. Niksirat KS, Goswami L, Rao PSB, Tyler J, Silacci A, Aliyu S, et al. Changes in research ethics, openness, and transparency in empirical studies between CHI 2017 and CHI 2022. In: Schmidt A, Väänänen P, Goyal T, Kristensson PO, Peters A, Mueller S, et al., editors. CHI '23: Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. New York, NY: Association for Computing Machinery (2023). p. 1–23. doi: 10.1145/3544548.3580848

Crossref Full Text | Google Scholar