Fostering Sustainable Learning via Embodied Intelligence: The E3-HOT Framework for Higher-Order Thinking in the AI Era

Synthesis: Zhu et al. (2026) address the risk that AI-assisted learning fosters cognitive outsourcing and learning detached from authentic contexts by proposing E3-HOT, a conceptual framework that leverages embodied intelligence to sustain learners' cognitive agency and higher-order thinking for sustainable learning, aligned with SDG 4. Three embodied pathways — situational embedding, embodied participation, and cognitive creation — map onto the higher-order targets of Analyze, Evaluate, and Create and are translated into a three-module E3 core (virtual–real integrated environment, embodied interaction, intelligent core) with bounded, teacher-governed AI support. Positioned as a design-and-evidence blueprint rather than an empirical study, it specifies multi-fidelity enactment options and auditable evidence artifacts for future validation.

Key Findings

• E3-HOT is a design-and-evidence blueprint, not an intervention report. Built through iterative conceptual synthesis (design-oriented scoping on titles, abstracts, and keywords), it frames two design questions (DQ1, DQ2) and specifies what should be observed and collected in future validation work, without claiming measured learning gains.
• Sustainability is the long-term maintainability of learners' cognitive agency under AI-intensive conditions, anchored to SDG 4. It additionally requires inclusive participation across diverse learner needs, feasibility through multi-fidelity enactment, and responsible evidence governance so Multimodal AI data collection stays bounded by pedagogical purpose rather than drifting toward surveillance.
• Three embodied pathways map to Analyze, Evaluate, and Create (revised Bloom's taxonomy, cognitive-process dimension): situational embedding → analytical thinking via concrete or simulated scenarios; embodied participation → evaluative thinking through bodily engagement and evidence-based judgment under explicit criteria; cognitive creation → creative thinking via AI-extended, human-determined co-creation.
• A three-module E3 core realizes the pathways: the Virtual–Real Integrated Environment supplies context; the Embodied Interaction module captures posture, motion, speech, and affective data; and the Intelligent Core acts as a Socratic guide (learner modeling/cognitive diagnosis, generative intelligence, intelligent tutoring/feedback, coordination and AI Regulation in Education) rather than a direct answer provider. A Profile/Learn orchestration layer parameterizes task requirements, constraints, and activity templates.
• AI support is bounded and teacher-governed. The intelligent core is constrained to prompts, critique, and strategy suggestions; students must articulate rationales and cite evidence before claims are accepted; system recommendations require teacher confirmation; and the system does not generate a final deliverable for direct submission.
• Multi-fidelity enactment and auditable evidence support equitable adoption. A substitution map preserves mechanism–objective alignment and evidence cues from high-tech (VR, wearables, motion capture, embedded Large Language Models (LLMs) coach) to low-tech options (scenario scripts with props, structured observation checklists, teacher-approved prompt cards). Evidence artifacts (Context Structure Map, Assumptions & Source Log, Criteria–Evidence Matrix, Embodied Trial Observation Sheet, Peer Critique Record, Iteration & Change Log, AI Assistance Disclosure Log, Teacher Gatekeeping Record) specify mandatory fields and rubric stubs for expert audit.
• An illustrative four-week university "human–AI co-creation design project" vignette walks through framing/exploration, synthesis/selection, prototyping/testing, and articulation/reflection, activating 12 facets across cognitive, conative, affective, sensorimotor, and knowledge dimensions — presented as a worked example, not an implemented study.
• Theory grounding draws on embodied intelligence and 4E cognition (embodied, embedded, enacted, extended), Constructivism learning theory, situated learning theory, and Self-Regulated Learning/metacognitive development, reframing sustainable learning as a shift from answer production to capacity preservation and growth.

What this means for practice

• Designers. Design activities so students repeatedly perform context-grounded analysis, evidence-based evaluation, and iterative creation, rather than using generative tools as a shortcut to a finished output.
• Designers. Constrain AI to prompts, critique, and strategy suggestions, and require students to articulate rationales and cite evidence before a claim is accepted.
• Designers. Use the multi-fidelity substitution map to preserve mechanism–objective alignment when immersive infrastructure is unavailable — scenario scripts with props, structured observation checklists, and teacher-approved prompt cards instead of VR, wearables, or motion capture.
• Administrators. Bound multimodal data collection by pedagogical purpose with teacher-configured logging, keep improvement evidence separate from data that could be repurposed for non-educational control, and integrate the approach within existing exercises rather than as added modules.
• Researchers. Collect the seven evidence artifacts with their mandatory fields and rubric stubs so the framework can be audited and validated.

Limitations

• E3-HOT is a conceptual design-and-evidence blueprint: no comparative empirical research across disciplines or classrooms has been conducted, and the four-week university "human–AI co-creation design project" is a worked example rather than an implemented study.
• The background review was a design-oriented scoping process on titles, abstracts, and keywords only, so the framework is a structured, auditable design baseline rather than an exhaustive synthesis.
• Validation remains planned rather than completed — small-scale design-based studies, quasi-experimental comparisons, and rubric-based expert audits of artifacts — so the paper claims no measured learning gains.

Citation

Zhu, H., Jiang, X., Zhang, X., Xu, H., Su, D., Chen, Z., & Zhu, X. (2026). Fostering Sustainable Learning via Embodied Intelligence: The E3-HOT Framework for Higher-Order Thinking in the AI Era. Sustainability, 18(7), 3469.