Theoretical Perspectives on Teaching with Robots: From Interdisciplinary Prerequisites and Necessities in Today’s Classrooms to Five Different Types of Robots

Synthesis: Christ, Riedl, Schmid, Zürcher and Thilo (2026) argue that social robots can act as embodied pedagogical partners in post-pandemic European classrooms, where rising child and adolescent mental-health problems coincide with shortages of teachers, school psychologists and Special Education staff. The paper is conceptual and reports no new data: it synthesizes research on embodied cognition, touch and physical play, Social-Emotional Learning, and child–robot interaction into a 4E cognition heuristic (embodied, embedded, enacted, extended), then uses design-thinking workshops with a seven-expert team to derive five types of classroom robots anchored in the Swiss MindMatters mental-health program. The five types run from non-interactive demonstration to one-on-one empathic dialogue, each mapped onto competence domains and paired with ethical safeguards. The contribution is a generative design space for future evaluation, not evidence of effectiveness.

Key Findings

1. The authors tie the argument to converging pressures: the German COPSY study (Ravens-Sieberer et al., 2022) reports that the prevalence of psychological problems in children rose from 17.6% before the pandemic to over 30% during subsequent lockdown phases, and UNESCO (2025) projects a shortfall of over 4.8 million teachers in Europe and North America by 2030, of which secondary education would require 3.1 million.
2. Staff shortages reach psychological support: recommended ratios of one school psychologist per 500–1000 students are rarely met, with a German estimate as high as 1:9482 (Jimerson et al., 2009) against 1:800 to 1:1200 in Finland (OECD, 2025).
3. Current classroom technologies give only partial relief. Learning management systems, behavior apps such as ClassDojo, digital mental-health platforms, interactive whiteboards and peer-tutoring platforms help with communication and behavior, but bring screen fatigue, data-privacy and surveillance concerns (Manolev et al., 2019), uneven participation (Topping, 2005), and cost or training burdens.
4. 4E cognition is used as a heuristic rather than a tested model, with each dimension mapped onto robot design features: embodied cognition onto physical and sensorimotor engagement, embedded cognition onto integration into the classroom setting, enacted cognition onto real-time co-construction of meaning, and extended cognition onto robots as external cognitive and social partners.
5. A chapter on touch and physical play grounds the argument developmentally: Field (2010) on tactile experience in brain maturation, Carozza and Leong (2021) on synaptogenesis, Pellegrini and Smith (1998) on rough-and-tumble play for negotiation and conflict resolution, and Diamond (2013) on physical activity and executive functions.
6. Reviewed empirical work consistently reports that educational robots raise engagement and motivation. Ouyang and Xu (2024) find gains in conceptual understanding and Problem Solving in STEM, Huang (2024) finds robot-assisted learning most effective in small-group settings, Wang et al. (2024) conclude via Technological Pedagogical Content Knowledge (TPACK) that robots are facilitators and not replacements that work best when teachers mediate, and Schulz et al. (2020) report reduced stress and better vocabulary and pronunciation in language learning. Reviewers repeatedly call for longitudinal study of retention and sustained impact.
7. The pedagogical functions are tied to concrete AI technologies: natural-language processing for dialogue, affective computing for emotion recognition and expression, reinforcement learning for adaptive behavior, and behavior-prediction models for personalization, framed as hybrid intelligence in which human and machine capabilities co-evolve rather than as teacher substitution.
8. The typology came from a team of seven experts, six researchers in health sciences, robotics, computer science, mathematics and design plus one specialist in evidence-based mental-health modules for Swiss schools, who ran five workshops of two hours each using the inspiration and ideation phases of Brown’s (2008) Design Thinking model with think–pair–share and Metaplan methods. No real robots were used in the workshops.
9. Robot type a performs a non-interactive demonstration of generic social patterns, a pre-programmed automated theater of emotions shown to a group and then discussed, illustrating generalizable dynamics such as escalation or misunderstanding.
10. Robot type b is an interactive robot that reacts to touching with programmed human responses, enabling a participative physical theater with peer role-play in which pupils need not take unpleasant roles; cited work includes studies with autistic children and children with visual impairment.
11. Robot type c is a spoken-language interactive partner that shows empathic behavior and memorizes interactions with single students, creating a one-to-one setting that can reduce pupils’ fear of sharing secrets with real people.
12. Robot type d is an externally guided performative agent, steered by an invisible pedagogical intervention specialist and functioning like a puppet with greater degrees of freedom, intended to minimize social hierarchies and raise interest in the learning content.
13. Robot type e is a non-interactive robot programmed to mimic actions that recently occurred in the school setting, creating a stage for situated reflection; the authors suggest it may later be guided by AI that extracts behavioral patterns from videos, and contrast its context-specific abstraction with the generalized level of type a.
14. Competence domains and safeguards are mapped explicitly: types a and e support reflection and perspective-taking, type b supports embodied learning, boundary awareness and emotion regulation, type c supports self-disclosure and Trust in a protected one-to-one setting, and type d supports critical engagement with hierarchies and agency; the ethical section requires data minimization, on-device processing where feasible, retention limits such as immediate deletion or storage for 24–72 h, prohibitions on secondary use for discipline or ranking, framing robots as tools rather than confidants, planned end-of-intervention transitions, and disclosure of control modes with teacher override and incident review. Earlier iterations produced seven categories which were merged in a consensus process.

Study Design & Method

This is a Perspective paper, not an empirical study. The authors state that it offers a theoretical perspective rather than an empirical investigation, that no new data were created or analyzed, and that they use a narrative, concept-driven literature review rather than a systematic search. The typology is a design exercise: five documented workshops with an interdisciplinary expert team, situated in the Swiss implementation of the MindMatters mental-health program, which has been used in Swiss primary and secondary schools since 2003. The workshops targeted pupils aged eight to fifteen because of their verbal maturity and their exposure to bullying and stress. The typology is presented as a heuristic design space rather than an empirically validated classification, and the reported next step is teacher focus groups on feasibility, practicality, risks and ethical considerations.

What this means for practice

• Instructors. Choose the robot type by the competence you intend to teach rather than by available hardware: the five proposed types cover reflection and perspective-taking (a and e), embodied learning, boundary awareness and emotion regulation (b), self-disclosure in a protected one-to-one setting (c), and critical engagement with hierarchies and agency (d), with emotional learning, conflict mediation and exam or job-interview practice as the recurring scenarios.
• Instructors. Set the pedagogical scenario before the technology, design for the situation rather than deploying generically, and stay in the mediating role — the reviewed evidence (Wang et al., 2024, read through TPACK) concludes that robots are facilitators rather than replacements of teachers or therapists and work best when teachers mediate. The claimed advantage over multimedia is twofold: physical presence, and a machine abstraction that lets pupils rehearse scenarios impossible with real peers.
• Designers. Build the safeguards into the system, not the policy document: data minimization, on-device processing where feasible, retention limited to immediate deletion or 24–72 hours, no secondary use for discipline or ranking, disclosure of control modes with teacher override and incident review, and a planned end-of-intervention transition.
• Researchers. Measure retention beyond the novelty phase, which the authors name as unresolved; their own reported next step is teacher focus groups on feasibility, practicality, risks and ethical considerations rather than further conceptual elaboration.

Limitations

• The framework and its use cases are anchored in the Swiss context and in the logic of MindMatters, and practical realization through robots is highly sensitive to regional, cultural and institutional factors, including norms of physical interaction and emotional disclosure, data protection rules and support infrastructure.
• The typology is explicitly not empirically validated: it comes from a design-thinking exercise with an interdisciplinary expert team rather than an empirical study, and is presented as a heuristic design space rather than a tested classification.
• The authors do not claim long-term effectiveness or sustained impact, and they flag unresolved questions about durability of effects beyond novelty phases, the evolution of children's emotional relationships with robotic agents, and institutional Sustainability; the cited Robots in Education evidence is already short on longitudinal and retention data.

Connected Concepts

• Robots in Education — social robots as classroom pedagogical partners
• Embodied Learning — 4E cognition and physical interaction as a basis for learning
• Social-Emotional Learning — emotional regulation, empathy and conflict skills targeted by the use cases
• Situated Learning — cognition distributed across bodies, tools and classroom contexts
• Teaching — robots as facilitators that teachers must mediate, never replacements
• Design Thinking — the workshop process used to derive the five robot types
• Well-Being — school-based mental-health promotion as the paper’s purpose
• Ethics — privacy, surveillance, attachment and accountability safeguards
• Affective Computing — emotion recognition and expression underpinning adaptive robot behavior
• Special Education — touch-based and humanoid robot work with autistic and visually impaired children
• Theory Development in AI in Education — conceptual synthesis offered ahead of systematic evidence aggregation
• Pedagogies and Teaching Strategies — role types defined by pedagogical function rather than technology

Connected Articles

• Exploring the Implementation of the Pepper Social Robot in Formal Education: A Scoping Review — scoping review of social robot use in formal education
• RoboBuddy in the Classroom: Exploring LLM-Powered Social Robots for Storytelling in Learning and Integration Activities — LLM-driven social robots in classroom settings
• Knowledge-Based Design Requirements for Generative Social Robots in Higher Education — design approaches for generative social robots
• Play-Testing REMind: Evaluating an Educational Robot-Mediated Role-Play Game — robot-mediated role-play against bullying
• Multimodality and Social Interactions in AI-Enhanced Embodied Robot-Assisted Language Learning: A Meta-Analysis — robot assistance for language learning outcomes
• Co-Creating Buildable and Open Social Robot Study Companions with University Students — robots as study companions for learners
• Pathways to Learning: Exploring High School Students' Learning of AI-Powered Educational Robotics — pathways and open questions in educational robotics
• An Experimental Study Exploring Human–AI Complementarity in Early Social-Emotional Learning — human–AI complementarity in social-emotional learning
• A Multimodal Framework for Embodied Cognition in Oral Explanations — embodied and multimodal accounts of learning
• New AI-Driven Tools for Enhancing Campus Well-being: A Prevention and Intervention Approach — digital well-being support in educational settings

Citation

Christ, O., Riedl, R., Schmid, J., Zürcher, P., & Thilo, F. (2026). Theoretical Perspectives on Teaching with Robots: From Interdisciplinary Prerequisites and Necessities in Today's Classrooms to Five Different Types of Robots. AI in Education, 2(3), 28.