1 Introduction
2 Conceptual Foundations
2.1 Definition
2.2 Related System Classes and Differences
2.3 Research Framework
3 Related Work and Opportunities for Future Research
3.1 (A) Virtual Coaching System–Frontend
3.2 (A) Virtual Coaching System–Backend
3.3 (B) User (Coachee)
3.4 (C) User–Coach Interaction
3.5 (D) Human Coach
3.6 (E) Context
Theory | Explanation | Exemplary research questions |
---|---|---|
Cognitive Load Theory states three additive factors that hamper learning: intrinsic (due to the natural complexity of the learning material), extraneous (due to inappropriate instructions), and germane cognitive load (effective efforts of the learner to understand the material). One main assumption of the theory is that the human working memory capacity is limited. | How should VCs take the users’ cognitive load into account? Under which conditions do VCs reduce cognitive load and improve coaching outcomes? | |
Social Cognitive Theory (Bandura 1986) | The Social Cognitive Theory assumes an interaction between personal, environmental, and behavioral factors that influence each other. Different theoretical constructs can be assigned to the three factors (e.g., self-efficacy and self-regulation to personal and observational learning to environmental factors). | How can peer coaching be integrated into VCs? Do peer coaching elements increase self-efficacy? |
The Transtheoretical Model is a stage model and assumes that people are passing different stages of change (SOC) during the behavior change process. In addition to the stages of change, there are also other theoretical constructs: processes of change (that promote the passage through the SOC’s), decisional balance (evaluating pros and cons of changing), self-efficacy, and temptation (opposite of self-efficacy). | How can the VC guide the user through the different SOCs? Does tailoring the coaching interventions to the SOC’s improve long-term behavior change? | |
Theory of Planned Behavior postulates that changes in behavior are influenced by one’s attitude towards the behavior, subjective norms, and perceived behavioral control. The behavioral intention mediates the three influencing variables. | How can the VC influence the coachee’s attitudes, subjective norms, and perceived behavioral control by using persuasive techniques? | |
Self Determination Theory attempts to explain that motivation for a certain behavior is influenced by the individual’s competence, autonomy, and relatedness. | How can the VC support competence, autonomy, and relatedness? | |
The COM-B model assumes that the three factors (capability, opportunity, and motivation) mutually influence the behavior. The model is embedded into the core of the so-called “Behavior Change Wheel” (a higher-level framework), which points out several policy strategies and intervention functions (e.g., education, persuasion, incentives) for sustainable behavior change. | How can the intervention functions named by the behavior change wheel be effectively implemented in digital systems? | |
The CASA paradigm suggests that computers can influence cognition, affection, and behavior the same way as other people can. Humans apply social rules to computers and, thus, socially respond to certain anthropomorphic cues. | Will the VC’s anthropomorphic design lead to better coaching outcomes? What design elements will trigger certain social responses with respect to the coaching scenario? | |
Theory of Uncanny Valley (Mori et al. 2012) | The Theory of Uncanny Valley posits the idea that increasing the “humanness” by implementing an anthropomorphic design can increase the acceptance (affinity) at first, the acceptance can turn negative if the system appears “too human” but do not behave like a real human. | What are the limitations of an anthropomorphic design in VCs? |
4 Conclusion
Research aspect | Exemplary research questions |
---|---|
A: Virtual Coach (VC) | RQ1: How can the VC be deployed on multiple devices (e.g., smartphone, smartwatch, VR/AR glasses) for combined application scenarios? RQ2: How can a “multi coach” approach be designed and implemented? RQ3: How can learning abilities of the system be implemented to personalize the coaching process? RQ4: Which XAI methods are particularly suitable for VCs to enhance user trust? RQ5: What are technical mechanisms to ensure the safety of the user? RQ6: How should generic virtual coaching solutions be designed? |
B: User | RQ7: How can the constructs of psychological theories that explain learning and behavior change be effectively mapped to software systems (see Table 2)? RQ8: How can the VC support habit formation? |
C: User—VC Interaction | RQ9: What are the long-term effects of certain social cues? RQ10: How can the VC be designed to effectively promote persuasiveness and user engagement? RQ11: How can a long-term human–VC relationship be established and maintained? RQ12: What are the consequences of an omnipresence of the coach? RQ13: How can ethical aspects be addressed when designing VCs? |
D: Human Coach | RQ14: How should the interface for the human coach be designed? RQ15: How can human expert knowledge be efficiently integrated into the VC? |
E: Context | RQ16: What are unobtrusive approaches to capture and predict the context? RQ17: What are application areas of VCs? |
Overarching research questions | RQ18: Do VCs influence coaching-oriented business models (creating a coaching economy)? RQ19: Do VCs change established industries or other domains (e.g., healthcare)? RQ20: Will VCs help reduce inequalities (e.g., in education)? |