How this framework teaches
Eight named educational-psychology theories drive every chapter, every assignment, and every measurement instrument in the AI Engagement Model courses. This page names them, explains how they show up in the book structure, and links to the source research for readers who want to verify the design.
Companion to Rubric Theory. Rubric Theory covers the methodology of grading; this page covers the methodology of teaching.
Eight foundational theories
Each theory is named so adopters can cite the research when explaining the design, and so reviewers can verify that the pedagogy is empirically grounded rather than intuition-driven.
Cognitive Load Theory
Sweller 1988; 1994
Working memory has limited capacity. Learning fails when decorative content, poorly organized prose, or redundant text+image crowds out the cognitive work that actually builds schemas.
Applied here
Tier-specific scaffolding (Beginner extra supports, Expert minimal scaffolds); short word counts in foundational chapters; no decorative prose; every chapter passes a coherence audit before it ships.
Expertise Reversal Effect
Kalyuga, Ayres, Chandler & Sweller 2003
The instructional support that helps novices actively harms experts. Worked examples speed novice learning but frustrate experts; problem-first approaches benefit experts but overwhelm novices.
Applied here
Beginner paths anchor on Tutor mode for scaffolded understanding; Expert paths anchor on Problem Setter for meta-level interrogation. Same chapter, different tier, different rubric — never generic.
Cognitive Apprenticeship
Collins, Brown & Newman 1989
Skill develops through four phases: Modeling → Coaching → Scaffolding → Fading. Most courses model and coach but never fade — students leave able to follow guided exercises but unable to perform without them.
Applied here
Learn with AI (1-credit) = Modeling + Coaching. Maximize Your Life (3-credit) = Scaffolding. A future advanced cert track will fill the Fading phase with real-world client projects, no scaffolds.
Desirable Difficulties
Bjork 1994
Conditions that slow apparent learning during practice produce stronger long-term retention and transfer. Spacing, interleaving, and retrieval practice are desirable difficulties. Blocked re-reading feels productive but isn't.
Applied here
Chapters teach modes through interleaved tasks rather than blocking by mode. Every chapter's assignment includes a brief sub-task using a previously-taught mode — costs 5 minutes of student time, delivers substantial transfer gain.
Testing & Generation Effect
Roediger & Karpicke 2006; Slamecka & Graf 1978
Recalling information strengthens memory more than re-reading it. Producing an answer strengthens memory more than recognizing one. Generating something is more durable than receiving it.
Applied here
Every assignment is a generation — a real AI conversation plus an artifact, not a multiple-choice test. Self-check quizzes use retrieval practice. The perception-vs-behavior comparison is structurally a retrieval prompt.
Reflective Practitioner
Schön 1983
Professional skill develops through reflection-in-action (during the work) and reflection-on-action (after). Without explicit reflection structures, experience plateaus.
Applied here
The perception-vs-behavior gap is the central reflection engine of the course. Every chapter compares what students believed they did to what their transcript shows. Most students discover the gap is larger than they expected.
Metacognitive Awareness
Posner & Rothbart 2007; Veenman et al. 2006
Students who monitor their own thinking outperform those who don't, even at the same skill level. Metacognition is teachable, and explicit metacognitive instruction beats implicit modeling.
Applied here
The 8-mode framework is the metacognitive vocabulary. Naming what mode you are in is itself the intervention. Over a semester, students develop the habit of pausing mid-conversation to ask which mode they're in and whether it's the right one.
Multimedia Learning Principles
Mayer 2014
Six principles: coherence (exclude extraneous material), signaling (highlight what's important), redundancy (don't duplicate prose and imagery), pre-training (vocabulary before procedure), modality (combine visual and verbal), personalization (conversational tone).
Applied here
Every chapter passes a Mayer audit during editorial QA. Path diagrams carry weight instead of repeating prose. Second-person voice throughout. Vocabulary introduced before procedure in every module.
The per-chapter structure
Every chapter — foundational or applied, in either book — follows the same seven-step shape. Each step is a theoretical commitment, not a stylistic choice.
- 1
Hook with a specific gap or failure
Mayer signaling · Bjork productive failure
Concrete over abstract. Real example, not generic claim.
- 2
Pre-training: brief vocabulary
Mayer pre-training
New term names defined as glossary entries. No more than 3–4 new terms per chapter.
- 3
Worked example for tier
Sweller worked-example effect · Cognitive apprenticeship modeling
Beginner tier sees a complete worked transcript; Expert tier sees an open prompt without a model. Same task, different scaffold.
- 4
Student's own attempt
Generation effect
The assignment. A real AI conversation + artifact upload. Two versions: guided (scaffolded) and free (unscaffolded).
- 5
Feedback against the rubric
Schön reflection-on-action · Metacognitive awareness
Mode-by-mode breakdown vs target. Plain-language interpretation. Transcript vs self-report comparison made explicit.
- 6
Retrieval-practice question on prior chapters
Testing effect · Interleaved practice
One or two questions tying current chapter back to recent ones. Embedded in the chapter, not a separate quiz. Ungraded.
- 7
Forward-link to next chapter
Transfer cuing
One paragraph. Connect what they just did to what's coming.
Word target per chapter: roughly 2,500–3,500 for foundational chapters; up to 5,000 for applied chapters with deeper artifact criteria.
Book-level structural patterns
Interleaved practice across modules
Currently each module focuses on a different task. Bjork's research shows that interleaved practice across topics outperforms blocked practice for transfer. Every chapter's assignment therefore includes one brief sub-task using a previously-taught mode — a Module 7 writing assignment asks students to verify one citation, exercising the Verification mode they learned two modules earlier. Five extra minutes of student time; substantial transfer gain.
Cognitive apprenticeship across the book pair
Collins, Brown, and Newman's four-phase progression maps cleanly onto a two-book curriculum:
- Modeling + Coaching — the Learn with AI book. Show what good engagement looks like, then coach students through their first attempts.
- Scaffolding — the Maximize Your Life with AI book. Student leads; rubric still scores; coaching reduces.
- Fading — a future advanced certification track, with real-world projects and peer review. Currently a defined open scope item; ships when there's adopter pull.
Constructivist baseline
Module 1 captures the student's natural AI-use pattern beforeteaching the framework. This is not a stylistic choice; it's a Piagetian commitment to productive disequilibrium. Students must confront their pre-existing pattern before instruction lands. Framework content is deliberately kept out of the baseline task.
Spaced re-encounter across the book pair
The foundational chapters appear in both books. For Maximize-only students that is the first encounter; for students taking both courses it is a spaced re-encounter, which Bjork's research shows strengthens retention. The duplication is a feature, not a redundancy.
Two-version assignment design
Every chapter's assignment ships in two versions that the student completes back-to-back. This is where Sweller's worked-example effect and Bjork's transfer principle meet.
Version A — Guided
The student is walked step-by-step through the prescribed mode path for the chapter's task. Operationalizes Sweller's worked-example effect — the student is practicing inside the scaffold.
Measures: mastery. Can they execute the rubric when scaffolding is present?
Version B — Free
The student opens whatever AI tool they normally use, with no instructions. They have a real conversation about the task however feels natural. They paste the transcript for grading.
Measures: transfer. Does the guided practice actually change unscaffolded behavior?
Both versions also upload the artifact the conversation produced — essay, code, decision document, whatever the task requires. The comparison between the two transcripts (same student, same day, with and without scaffolding) is the within-student A/B that makes the perception-vs-behavior gap concrete.
Editorial QA checklist
Every chapter draft passes a theory-aligned review before shipping. The checklist is public so adopters can verify our chapters meet it, and so contributing authors have explicit criteria.
Mayer audit
Coherence, signaling, redundancy, pre-training, modality, personalization
Cognitive-load audit
Word count within target; ≤ 3–4 new terms; worked example for Beginner tier
Structure audit
All 7 chapter steps present in order
Interleaving audit
Assignment includes a sub-task using a prior chapter's mode
Reflection audit
Perception-vs-behavior comparison surfaced; reflection prompt at end
Theory-citation audit
Design choices cite the theory that motivates them
Tier audit
Beginner / Intermediate / Expert paths match the canonical tier rules
Full reference list
- Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In J. Metcalfe & A. Shimamura (Eds.), Metacognition: Knowing about knowing (pp. 185–205). MIT Press.
- Collins, A., Brown, J. S., & Newman, S. E. (1989). Cognitive apprenticeship: Teaching the crafts of reading, writing, and mathematics. In L. B. Resnick (Ed.), Knowing, learning, and instruction (pp. 453–494). Erlbaum.
- Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). The expertise reversal effect. Educational Psychologist, 38(1), 23–31.
- Mayer, R. E. (2014). The Cambridge Handbook of Multimedia Learning (2nd ed.). Cambridge University Press.
- Posner, M. I., & Rothbart, M. K. (2007). Educating the human brain. American Psychological Association.
- Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255.
- Schön, D. A. (1983). The Reflective Practitioner: How Professionals Think in Action. Basic Books.
- Slamecka, N. J., & Graf, P. (1978). The generation effect: Delineation of a phenomenon. Journal of Experimental Psychology: Human Learning and Memory, 4(6), 592–604.
- Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257–285.
- Sweller, J. (1994). Cognitive load theory, learning difficulty, and instructional design. Learning and Instruction, 4(4), 295–312.
- Veenman, M. V. J., Van Hout-Wolters, B. H. A. M., & Afflerbach, P. (2006). Metacognition and learning: Conceptual and methodological considerations. Metacognition and Learning, 1(1), 3–14.
This page is a living document. As the research behind the course grows and the books converge on the canonical chapter structure, we update both the theoretical foundations and the per-chapter QA practice. The full coordination-side pedagogical guide lives with the authoring team; this page is the public-facing summary.