Designing Learning That Adapts

Abstract

Adaptive learning is most useful when it connects a clear goal and context to meaningful practice, observable evidence, timely feedback and a transparent next step. This paper turns that sequence into a practical framework for learning designers, educators, institutions and specialist AI tutoring builders.

The framework combines instructional design, a non-sensitive system architecture and an evaluation agenda. Wider evidence shows that immersive technology can support learning in some settings, but presence, enjoyment and learning do not always move together [2-6]. The source proposal contains no controlled outcome dataset. This technical white paper is not a completed outcome or efficacy study.

How to use this paper

The distinction matters because a proposal, a research finding and a design recommendation do different jobs. Source claims retain their exact page locators. External findings remain limited to the studies that produced them. Design recommendations are open to challenge. Evaluation steps describe future methods rather than completed activity.

Why adaptive learning matters

A fixed sequence may not respond well to different goals, prior knowledge, demonstrated performance, contexts or access needs. That does not mean every learner needs a unique technology stack. It means the route through an objective should respond to relevant evidence, learner choice and support.

Technology can help or hinder learning, depending on the problem, setting and intervention [6-7]. The practical question is not whether adaptation or immersion sounds engaging. It is whether a defined design improves learning or access against a fair alternative, for a known group, without creating larger costs or harms.

Research origin and history: Immersion3.org

Immersion3.org is the project site at the time of publication, 24 October 2026. The site is available and presents the project as inviting collaborative research.

Site availability does not establish current research activity, implementation, participants, activity cadence, efficacy or outcomes.

This paper uses the unpublished Immersion3: Technical and Innovation Proposal from September 2024 as its principal archival document because it consolidated design concepts, a proposed architecture and evaluation questions [1]. The proposal is not the sole research source, a complete research record or a controlled outcome study.

The wider evidence is deliberately balanced. Reviews report promising results for some immersive-learning designs, alongside neutral and detrimental outcomes and substantial differences in interventions and assessments [2,5]. Individual experiments show that greater presence or liking can coincide with no learning gain or lower measured learning [3-4]. These studies did not evaluate the source project.

The source proposal referred to matching content to learning styles. This paper does not carry that mechanism forward. Evidence does not support matching instruction to self-reported learning styles [8]. Goals, prior knowledge, demonstrated performance, access needs, preferences and learner choice remain relevant, but they should not be collapsed into a fixed style label.

The adaptive learning loop

A useful adaptive journey begins with the learner's goal and context, offers a meaningful explanation or scenario, invites a decision or task, gathers relevant evidence, provides feedback and selects a bounded next step. The cycle can operate in a classroom, an e-learning platform, a workplace simulation or a tutoring conversation.

Adaptation should answer an instructional need rather than personalise for its own sake. The learner and educator need to understand why a route changed and how to override it. The model is a proposed design, not a reconstruction of a deployed system or a measured result.

Practical design principles

Start with one competency or objective that matters in the learner's context. Establish the destination, relevant prior knowledge, constraints and access needs. Then choose an explanation, example or scenario that helps the learner act.

• Define the objective and what acceptable evidence would look like.
• Use a meaningful scenario or task rather than decorative interaction.
• Invite an observable decision, explanation, action or artefact.
• Assess evidence with a transparent rule and stated limits.
• Recommend continue, revisit, another explanation, supported practice, educator review or pause.
• Explain the recommendation and allow learner or educator override.

Keep adaptation bounded and reversible. A recommendation can organise practice, but it should not conceal the objective, remove meaningful choice or turn an uncertain signal into a high-stakes judgement.

Selecting and combining learning methods

Immersive delivery is one option, not the organising purpose. A useful programme may combine reading and explanations, hands-on projects, task-based learning, gaming-style practice, and conversation or Q&A. Each method should earn its place by serving the objective and the learner's context.

• Reading and explanations can introduce concepts, worked examples and accessible alternatives.
• Hands-on projects can produce artefacts that show planning, synthesis and application over time.
• Task-based learning can make a process observable in realistic steps.
• Gaming-style practice can use rules, feedback and repetition where they support the skill rather than distract from it.
• Conversation and Q&A can support explanation, reflection and checks for misunderstanding.

More methods are not automatically better. The learner should be able to continue, revisit, request an alternative explanation, move to supported practice, ask for educator review or pause.

From practice to demonstrated competency

Activity is not the same as competency. Evidence may include recall, explanation, task performance, a project artefact, dialogue-based reasoning or transfer to a new problem. Each evidence item needs a criterion, provenance, relevant attempt context, a transparent scoring rule, confidence limits and a review path.

Automated signals may recommend a next step. They should not independently certify high-stakes competence. Uncertain or consequential judgements need human review and another opportunity for the learner to demonstrate the objective.

Applying the framework across learning settings

The same adaptive loop can support physical classrooms, blended or hybrid learning, asynchronous e-learning and LMS delivery, workplace or vocational simulations, and AI-supported tutoring. These are proposed applications to design and test, not claims that any institution has adopted or validated the framework.

• Physical classrooms can use teacher-led branching cases, role-play, observation, feedback and differentiated next tasks.
• Blended and hybrid courses can connect in-person practice with online explanations, evidence capture and planned follow-up.
• Asynchronous e-learning and LMS routes can offer transparent branching, alternative explanations and educator escalation.
• Workplace and vocational simulations can rehearse consequential decisions with safe boundaries and human review.
• AI-supported tutoring can explain, question and recommend practice while keeping objectives, evidence and override visible.

Instructional and system architecture

A credible architecture separates interfaces and external I/O, session orchestration, instructional state, adaptive decisions, evidence, evaluation and governance. Learner and educator routes should remain usable when an optional immersive device or decision service is unavailable.

Orchestration selects content and schedules assessment. Domain state holds content metadata, competency definitions, learner context and purpose-limited evidence. A decision service returns a recommendation, confidence and explanation rather than an opaque verdict. Low-confidence or failed decisions return a fixed safe route and optional educator review instead of blocking learning.

Accessibility, inclusion, privacy, safety and responsible AI

An immersive route needs an equivalent non-VR route for the same objective [7,12]. Keyboard and assistive-technology access, captions or equivalent transcripts, readable text and reflow belong in accessibility acceptance criteria [12]. Where interaction animation is not essential, a reduced-motion option should be available [12]. Immersive delivery can introduce discomfort and access barriers [2-4,7]. Device availability and bandwidth can also constrain equitable access [7].

Privacy starts before data collection. A concrete implementation would need a defined purpose, minimum necessary data, lawful basis, access controls, retention and deletion rules, and lifecycle review [11]. AI-enabled adaptation also needs explanation, override, monitoring and incident review [10].

Questions for educators, institutions and researchers

A framework becomes useful research only when its assumptions can be challenged. Educators and institutions should ask what problem is being solved, what evidence matters and who can contest a decision. Directional hypotheses may fail, and null or adverse outcomes remain possible.

• Does a meaningful scenario improve delayed retention or transfer against an equivalent lesson?
• Does performance-based adaptation improve mastery, time or retention against fixed sequencing?
• Does combining methods add value beyond a simpler well-designed alternative?
• Which elements support authentic practice, and which add distraction, cognitive load, discomfort or exclusion?
• Are assessment signals reliable and valid enough to guide a next step?
• Do decisions perform consistently across groups, settings, devices, bandwidth conditions and access needs?
• Can learners and educators understand, override and contest recommendations?
• What minimum interaction data is needed to provide useful support?
• Can the architecture meet predefined reliability, completion, latency and accessibility criteria?
• What authoring, teaching and oversight effort keeps content accurate and current?

Proposed evaluation framework

A protocol should define population, setting, inclusion, recruitment and sample rationale before outcome inspection. It should specify the intervention and an equivalent comparator, assignment method, baseline, attrition, missing-data handling, instruments, timing, scoring and analysis across physical classrooms, blended or hybrid learning, asynchronous e-learning or LMS delivery, workplace or vocational simulations, and AI-supported tutoring [9].

Delayed retention and transfer are suitable proposed primary outcomes. Immediate learning, time to mastery, task completion, usability, cognitive load, presence, engagement, discomfort, accessibility, agency, educator workload, decision accuracy and system reliability can remain secondary. Presence and engagement do not substitute for learning.

Reporting should include estimates and uncertainty, protocol changes, null findings, adverse outcomes and distributional effects. Privacy, accessibility and governance review should precede recruitment. The final decision may be to revise, retain or stop the intervention.

Implementation trade-offs and limitations

The framework could introduce distraction, cognitive load, discomfort, unequal device or bandwidth access, content error, authoring effort, privacy risk, unfair decisions, over-adaptation, reduced learner agency, weak assessment validity and additional educator workload. A richer route may also cost more than an equivalent simpler route without adding learning value.

The source proposal has no bibliography, protocol, participant description, instruments, dataset, analysis, implementable algorithm or independent validation. Page 10 describes a completed pilot with 100 users, while page 17 presents a pilot of at least 100 users as a future target [1]. With no records to resolve that conflict, neither statement is evidence of a completed study or result.

The external literature supports careful testing rather than a conclusion about this framework. The architecture, competency model and evaluation recommendations still require implementation, expert review and empirical testing. Institutions should compare authoring effort, infrastructure, educator workload and accessibility against the simplest credible alternative.

Progression to specialist AI tutoring

The progression is a design direction rather than evidence of effectiveness. It does not imply that any current software, algorithm or tutoring result has been validated by the earlier source project.

Conclusion and research agenda

A useful adaptive-learning system keeps six things connected: the learner's goal and context, meaningful practice, observable evidence, timely feedback, bounded adaptation and demonstrated competency. Its quality depends less on novelty than on whether those connections are clear, accessible and open to review.

The next responsible step is small and auditable. Select one setting and objective, publish a protocol, create equivalent accessible routes, compare against a credible alternative, and report null and adverse findings. Replication should then test what transfers across classrooms, e-learning, workplace practice and AI-supported tutoring.

References and endnotes

1. DoviLearn®. Immersion3: Technical and Innovation Proposal. Unpublished proposal, September 2024. Pages 4, 5, 8-10, 12 and 17-18. No public URL.
2. Hamilton, D., McKechnie, J., Edgerton, E. and Wilson, C. Immersive virtual reality as a pedagogical tool in education: a systematic literature review of quantitative learning outcomes and experimental design. Journal of Computers in Education, 2021. Accessed 24 October 2026. https://doi.org/10.1007/s40692-020-00169-2 (opens in a new tab)
3. Makransky, G., Terkildsen, T. S. and Mayer, R. E. Adding immersive virtual reality to a science lab simulation causes more presence but less learning. Learning and Instruction, 2019. Accessed 24 October 2026. https://doi.org/10.1016/j.learninstruc.2017.12.007 (opens in a new tab)
4. Makransky, G., Andreasen, N. K., Baceviciute, S. and Mayer, R. E. Immersive virtual reality increases liking but not learning with a science simulation and generative learning strategies promote learning in immersive virtual reality. Journal of Educational Psychology, 2021. Accessed 24 October 2026. https://doi.org/10.1037/edu0000473 (opens in a new tab)
5. Merchant, Z., Goetz, E. T., Cifuentes, L., Keeney-Kennicutt, W. and Davis, T. J. Effectiveness of virtual reality-based instruction on students' learning outcomes in K-12 and higher education: A meta-analysis. Computers & Education, 2014. Accessed 24 October 2026. https://doi.org/10.1016/j.compedu.2013.07.033 (opens in a new tab)
6. Escueta, M., Nickow, A. J., Oreopoulos, P. and Quan, V. Upgrading Education with Technology: Insights from Experimental Research. Journal of Economic Literature, 2020. Accessed 24 October 2026. https://doi.org/10.1257/jel.20191507 (opens in a new tab)
7. UNESCO Global Education Monitoring Report. Technology in education: A tool on whose terms? 2023. Accessed 24 October 2026. https://www.unesco.org/gem-report/en/publication/technology (opens in a new tab)
8. Pashler, H., McDaniel, M., Rohrer, D. and Bjork, R. Learning Styles: Concepts and Evidence. Psychological Science in the Public Interest, 2008. Accessed 24 October 2026. https://doi.org/10.1111/j.1539-6053.2009.01038.x (opens in a new tab)
9. U.S. Department of Education, Institute of Education Sciences. What Works Clearinghouse Procedures and Standards Handbook, Version 5.0. 2022. Accessed 24 October 2026. https://ies.ed.gov/ncee/WWC/Docs/referenceresources/Final_WWC-HandbookVer5_0-0-508.pdf (opens in a new tab)
10. National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework 1.0. 2023. Accessed 24 October 2026. https://www.nist.gov/itl/ai-risk-management-framework (opens in a new tab)
11. UK Information Commissioner's Office. Data protection by design and by default. Updated 5 February 2026. Accessed 24 October 2026. https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/accountability-and-governance/guide-to-accountability-and-governance/data-protection-by-design-and-by-default/ (opens in a new tab)
12. World Wide Web Consortium. Web Content Accessibility Guidelines 2.2. W3C Recommendation, 2024. Accessed 24 October 2026. https://www.w3.org/TR/WCAG22/ (opens in a new tab)
13. Immersion3.org project site. Current first-party project context, checked 24 October 2026. Available over an unencrypted HTTP connection. http://immersion3.org/ (opens in a new tab)

Glossary

Immersion

A sense of being surrounded by or deeply involved in a mediated environment.

Adaptation

A bounded change to content, support or sequence in response to relevant evidence or learner choice.

Adaptive sequencing

Selecting a next learning opportunity using explicit objectives, rules and available evidence.

Assessment evidence

A recorded learner action or artefact interpreted against a stated criterion.

Competency

The ability to apply relevant knowledge or skill to a defined standard and context.

Learner model

A purpose-limited representation of relevant goals, prior knowledge, evidence and support needs.

Multimodal learning

Learning delivered through more than one suitable form, such as text, visual explanation, task, project or conversation.

Comparator

An alternative condition designed to make an evaluation question answerable.

Delayed retention

Evidence that knowledge or skill remains available after a defined period.

Transfer

Applying learning to a different but relevant problem or setting.

Human oversight

Meaningful review, intervention and accountability by an authorised person.

Publication information

Subtitle

A practical framework for immersive learning, competency and specialist AI tutoring

Version

v1

Published

24 October 2026

Lead Researcher

Vincent Chidozie Okeke

Research Assistant

Josephine Uchechukwu Okeke

Research Assistant

Chidi Jeremiah Okeke

Research publisher

DoviLearn® Research

Research imprint operator

Dovichi Labs Ltd

This technical white paper presents a proposed adaptive-learning framework and evaluation agenda. It is not a completed outcome or efficacy study.

Developed from the Immersion3 research project by DoviLearn® Research.

Published by DoviLearn® Research, the research imprint of Dovichi Labs Ltd.

Suggested citation: Vincent Chidozie Okeke (Lead Researcher), with Josephine Uchechukwu Okeke and Chidi Jeremiah Okeke (Research Assistants). Designing Learning That Adapts. DoviLearn® Research, 2026. https://doi.org/10.5281/zenodo.23669204