---
title: Separating The Experience Contract Curriculum And Instructional Design
description: Separating the Experience Contract, Curriculum, and Instructional Design
created: 2026-10-23
updated: 2026-10-23
authors: ThinkingCap
topics: [Research Development]
status: published
canonical: https://console.thinkingcap.com/guest/Research-Development/Skills/Separating-the-Experience-Contract--Curriculum--and-Instructional-Design
---

Separating the Experience Contract, Curriculum, and Instructional Design

**A research-informed white paper and implementation guide**  
Version 1.0 · 23 October 2026

## Abstract

Educational experiences combine promises to learners, decisions about what must be learned, and decisions about how learning is supported. When these decisions are compressed into a single course outline or content-production process, changing an activity can quietly change the evidence required for success, and adding content can silently expand the promise made to learners. The Experience Envelope is proposed here as an architecture that makes those decisions explicit through three connected layers: the **Experience Contract**, the **Curriculum**, and **Instructional Design**.

The Experience Contract defines the purpose, participation conditions, commitments, boundaries, and recognized success of an experience. The Curriculum defines the capabilities, knowledge, progression, and evidence standards necessary to fulfil that purpose. Instructional Design defines the activities, materials, feedback, supports, and delivery arrangements through which learners develop and demonstrate those capabilities. The layers are distinct in responsibility and connected through traceability.

The framework draws on research on alignment, motivation, self-regulation, cognitive load, formative assessment, retrieval practice, and active learning. Those research traditions support particular design principles; they do not establish that this named architecture improves outcomes. This paper supplies comparison matrices, diagrams, example pathways, a change-impact process, an AI-assisted authoring workflow, and an evaluation agenda for testing that proposition.

## Executive position

An Experience Envelope is the documented set of commitments, learning requirements, and design options that makes a learning experience coherent and adaptable.

Its central proposition is:

> Make the learner-facing promise explicit, define the learning and evidence needed to fulfil it, and permit instructional variation within those boundaries.

The three layers answer different questions:

1. **Experience Contract:** What are people entering into, under what conditions, and what does the provider undertake to make possible?
2. **Curriculum:** What capabilities and knowledge must learners develop, and what evidence will count?
3. **Instructional Design:** What will learners and educators actually do to support and assess that development?

This separation is a proposed governance and authoring approach. It does not require three teams, three software systems, or a rigid sequence. One educator may own all three layers while recording their decisions separately.

The strongest defensible research claim is that the framework is **consistent with selected findings and principles from learning research**. Claims of proven superiority, guaranteed engagement, universal transfer, or automatic personalization would require additional evidence.

## Contents

- [1. The problem the framework addresses](#1-the-problem-the-framework-addresses)
- [2. The three-layer architecture](#2-the-three-layer-architecture)
- [3. Layer 1: Experience Contract](#3-layer-1-experience-contract)
- [4. Layer 2: Curriculum](#4-layer-2-curriculum)
- [5. Layer 3: Instructional Design](#5-layer-3-instructional-design)
- [6. Comparison and boundary matrices](#6-comparison-and-boundary-matrices)
- [7. A complete worked example](#7-a-complete-worked-example)
- [8. Example learner pathways](#8-example-learner-pathways)
- [9. Change-impact flow and versioning](#9-change-impact-flow-and-versioning)
- [10. AI-assisted authoring workflow](#10-ai-assisted-authoring-workflow)
- [11. Evidence, limitations, and evaluation](#11-evidence-limitations-and-evaluation)
- [12. Implementation and release checklist](#12-implementation-and-release-checklist)
- [13. Conclusion](#13-conclusion)
- [14. Academic bibliography and source notes](#14-academic-bibliography-and-source-notes)
- [Appendix A: Reusable specification](#appendix-a-reusable-specification)
- [Appendix B: MDX publishing notes](#appendix-b-mdx-publishing-notes)

## 1. The problem the framework addresses

A course may have polished lessons but an unclear purpose. It may promise flexible participation while requiring attendance at an undisclosed live event. It may teach one kind of performance and assess another. These are different defects, and they call for different repairs.

Consider a program advertised as helping participants interpret evidence and make a justified recommendation. Its curriculum includes uncertainty and source evaluation, but its final assessment asks learners only to recall terminology. Adding a more engaging video would leave the central mismatch intact. The required capability, assessment evidence, and instructional activity need to be reconciled.

Constructive alignment connects intended performances, learning activities, and assessment. Biggs places the learner's meaning-making activities at the centre of that alignment. The Experience Envelope adopts alignment as a design obligation and adds an explicit place to record participation promises and operating constraints. That additional layer is this paper's proposal. [Biggs, 1996](https://doi.org/10.1007/BF00138871).

Three recurring failure patterns motivate the separation:

| Failure pattern | Observable symptom | First place to investigate |
| --- | --- | --- |
| Promise drift | A short introductory experience becomes a prerequisite-heavy qualification | Experience Contract |
| Learning drift | A growing resource library has no clear capability progression | Curriculum |
| Activity drift | Attractive tasks consume time without preparing learners for the required performance | Instructional Design |
| Evidence drift | Completion or clicks are treated as proof of competence | Curriculum evidence standards, then assessment design |
| Participation drift | Published flexibility conflicts with the actual schedule or technology | Experience Contract, then delivery design |

These categories are diagnostic starting points. A defect may cross all three layers.

## 2. The three-layer architecture

The envelope is the whole arrangement. It is not a fourth layer above the other three.

```mermaid
flowchart TB
  subgraph EE["Experience Envelope: the complete arrangement"]
    EC["Layer 1: Experience Contract\nPurpose, participation, commitments, boundaries"]
    CU["Layer 2: Curriculum\nCapabilities, knowledge, progression, evidence standards"]
    ID["Layer 3: Instructional Design\nActivities, materials, feedback, support, delivery"]
    EC -->|"sets commitments and constraints"| CU
    CU -->|"specifies learning and evidence requirements"| ID
    ID -.->|"reveals feasibility and learning evidence"| CU
    CU -.->|"exposes scope or promise conflicts"| EC
  end
  LE["Learner experience in practice"]
  ID --> LE
  LE -.->|"evaluation informs all three layers"| EC
```

**Figure 1.** The layers distinguish decision responsibilities. The feedback arrows make revision possible; they do not authorize silent changes to commitments.

Read the diagram as a dependency structure, rather than a compulsory waterfall. Teams can prototype an activity early to discover whether a promise is feasible. They must then update the relevant specification explicitly.

### 2.1 Three forms of coherence

**Contract coherence** means that the offered experience, participation conditions, and recognized success fit together. **Curricular coherence** means that outcomes, prerequisite relationships, content, and evidence standards support a defensible learning progression. **Instructional coherence** means that activities and supports prepare learners for the performances required.

The working test is bidirectional: every mandatory activity should have a reason, and every required capability should have an adequate opportunity for development and demonstration.

### 2.2 Invariants and permitted variation

An invariant is a requirement that remains stable within a published version. A permitted variation is a choice that may change without breaking that requirement.

| Layer | Example invariant | Example permitted variation |
| --- | --- | --- |
| Experience Contract | Learners have an asynchronous completion route | The schedule of optional live support |
| Curriculum | Learners must justify a recommendation using evidence and uncertainty | The sector used for the case study |
| Instructional Design | Learners receive feedback before the final submission | Written, audio, or annotated feedback, with accessible equivalents |

Variation still requires checking. An audio submission may preserve reasoning evidence while making a required writing capability unobservable. The curriculum determines whether that substitution is acceptable.

## 3. Layer 1: Experience Contract

The Experience Contract is a plain-language specification of what participation means and what the provider commits to delivering. Here, “contract” names a design artifact; whether any statement creates a legal obligation depends on its context and is outside this paper's scope.

This layer contains purpose, audience, entry assumptions, workload expectations, participation modes, available choices, support commitments, assessment arrangements, and exclusions. It also states how changes will be communicated.

### 3.1 Minimum contents

| Field | Question answered | Example commitment |
| --- | --- | --- |
| Purpose | What practical value is intended? | Develop a defensible evidence-based recommendation |
| Audience and entry | For whom is this feasible? | Adults comfortable reading short workplace reports; no statistics prerequisite |
| Scope | What does this experience cover? | Interpretation of supplied evidence, excluding statistical modelling |
| Participation | How can learners take part? | Fully asynchronous route; optional live clinic |
| Workload and timing | What effort and deadlines should learners plan for? | Estimated 6–8 hours across two weeks |
| Support | What assistance is actually available? | One draft review within two business days |
| Choice and access | Which decisions can learners make? | Choose one of three cases; accessible text alternatives provided |
| Recognized success | What will completion mean? | Meet the published recommendation rubric |
| Assessment conditions | What tools and help are permitted? | AI may support practice; final reasoning must be explainable by the learner |
| Boundaries and changes | What is excluded, and how are changes handled? | No professional licence; material changes are announced with transition arrangements |

These are illustrative commitments, not research-derived thresholds. An organization should publish only what it has the capacity to provide.

### 3.2 Why this layer matters

Self-determination theory identifies autonomy, competence, and relatedness as important psychological needs. This supports attending to meaningful choice, achievable challenge, and supportive relationships; it does not imply that unrestricted choice or a written contract will cause motivation. [Ryan & Deci, 2000](https://doi.org/10.1037/0003-066X.55.1.68).

Self-regulated learning includes processes such as setting goals, monitoring progress, and reflecting on performance. Explicit expectations can provide material for those processes, but learners may also need direct guidance in using them. [Zimmerman, 2002](https://doi.org/10.1207/S15430421TIP4102_2).

The proposed implication is to make choices and expectations usable: show the route, explain what success looks like, and provide a way to seek help. Do not assume that merely publishing information makes it understandable or actionable.

### 3.3 Boundaries of responsibility

The contract states that a learner will have an opportunity to demonstrate a capability and identifies the credential or completion claim. The curriculum defines the capability in detail and sets the evidence standard. Instructional Design creates the actual assessment and support arrangements. These statements should reference one another instead of being independently rewritten.

## 4. Layer 2: Curriculum

The Curriculum is the organized specification of what learners need to know and be able to do, how those requirements relate, and what evidence will justify a judgment of achievement.

It includes concepts, facts, procedures, practices, dispositions where assessable, prerequisite relationships, progression, transfer contexts, and assessment criteria. A content list alone does not supply all of that information.

### 4.1 Start with observable performances

“Understand evidence” leaves too much undecided. “Compare two sources, identify a relevant limitation, and justify a recommendation under uncertainty” gives authors and assessors a performance to work toward.

| Capability ID | Required performance | Necessary knowledge | Evidence standard |
| --- | --- | --- | --- |
| C1 | Judge the relevance and credibility of sources | Provenance, methods, recency where relevant, conflicts of interest | Explain why each selected source is appropriate and identify a material limitation |
| C2 | Interpret a claim without overstating it | Baselines, denominators, uncertainty, association and causation | Interpret a supplied chart and distinguish an observation from a causal claim |
| C3 | Recommend an action with a defensible rationale | Trade-offs, alternatives, assumptions, applicability | Connect evidence to the action, compare an alternative, and state what could change the decision |

These are the proposed outcomes for the worked example. They are not a universal taxonomy.

### 4.2 Map dependencies rather than assuming one route

```mermaid
flowchart LR
  K1["Provenance and source methods"] --> C1["C1: Evaluate sources"]
  K2["Baselines and uncertainty"] --> C2["C2: Interpret claims"]
  C1 --> C3["C3: Justify a recommendation"]
  C2 --> C3
  K3["Trade-offs and alternatives"] --> C3
  C3 --> T["Apply reasoning to a new case"]
```

**Figure 2.** This dependency map is a design hypothesis. Diagnostic and performance evidence should test whether the assumed relationships hold for the intended learners.

The curriculum identifies dependencies; Instructional Design determines how to teach and check them. A learner who already meets C1 may need less source-evaluation practice while still needing to demonstrate C1 in the final performance.

### 4.3 Assessment belongs across two layers

Curriculum owns the **claim and warrant**: the capability being judged, the quality criteria, and what evidence is sufficient. Instructional Design owns the **elicitation**: the task, instructions, examples, delivery, timing, and feedback that produce interpretable evidence.

For C3, “connect a recommendation to evidence and qualify uncertainty” is curricular. “Submit a 700-word decision brief using the supplied case” is an instructional and assessment-design choice. Word count becomes a curriculum requirement only when that particular communication constraint is part of the target capability.

## 5. Layer 3: Instructional Design

Instructional Design translates the curriculum into a feasible set of experiences under the contract's conditions. It specifies what learners do, what educators or systems do, which resources are used, how feedback works, and how performance is assessed.

Merrill's synthesis proposes problem-centred learning supported by activation, demonstration, application, and integration. It provides a useful design lens rather than proof that one sequence is optimal in every setting. [Merrill, 2002](https://doi.org/10.1007/BF02505024).

### 5.1 A design sequence for the worked example

| Design move | Learner action | Purpose | Evidence collected |
| --- | --- | --- | --- |
| Activate | Describe a recent decision and the evidence used | Surface prior knowledge and assumptions | Short diagnostic response |
| Demonstrate | Annotate an expert's reasoning in a worked case | Make judgment processes visible | Learner explanation of a key decision |
| Practise | Compare two sources and interpret a chart | Build C1 and C2 before integration | Source notes and chart interpretation |
| Receive and use feedback | Revise a flawed recommendation | Address a specific performance gap | Before-and-after explanation |
| Integrate | Produce a recommendation for a new case | Combine C1, C2, and C3 | Decision brief and rationale |
| Reflect and revisit | Explain how the reasoning transfers and retrieve key distinctions later | Support monitoring and retention | Reflection and delayed check |

Cognitive load theory supports attending to working-memory demands and to the learner's existing knowledge. In this design, segmenting the task and providing worked examples are proposed ways to manage those demands. Support should be adjusted as expertise develops, rather than retained at the same level for everyone. [Sweller et al., 2019](https://doi.org/10.1007/s10648-019-09465-5).

### 5.2 Feedback must change the next action

Formative assessment uses evidence about learning to inform subsequent teaching and learning. A score by itself may not provide that function. [Black & Wiliam, 1998](https://doi.org/10.1080/0969595980050102). Feedback design can also support learners in judging and regulating their own work. [Nicol & Macfarlane-Dick, 2006](https://doi.org/10.1080/03075070600572090).

```mermaid
flowchart LR
  A["Attempt a task"] --> E["Examine evidence against criteria"]
  E --> F["Identify a specific gap"]
  F --> N["Choose and explain a next action"]
  N --> R["Revise or attempt a new case"]
  R --> E
```

**Figure 3.** The feedback loop is an implementation proposal informed by formative-assessment research. A revision opportunity makes it possible to observe whether feedback was used.

### 5.3 Retention and transfer require separate evidence

In experiments using prose passages, retrieval practice improved delayed retention relative to repeated study, while immediate testing could favour restudy. This motivates a delayed retrieval opportunity; it does not establish that a short quiz measures transfer to complex workplace decisions. [Roediger & Karpicke, 2006](https://doi.org/10.1111/j.1467-9280.2006.01693.x).

The proposed design therefore uses a short delayed check for key distinctions and a new case for application. It treats immediate satisfaction, delayed retention, and transfer performance as different outcomes.

## 6. Comparison and boundary matrices

### 6.1 The three layers at a glance

| Dimension | Experience Contract | Curriculum | Instructional Design |
| --- | --- | --- | --- |
| Governing question | What is promised and under what conditions? | What learning and evidence are required? | How will learning and evidence be supported? |
| Primary object | Commitments and boundaries | Capability and knowledge structure | Tasks, interactions, resources, and supports |
| Typical artifact | Learner-facing agreement and operating specification | Outcome map, dependency map, evidence criteria | Learning sequence, resources, assessment tasks |
| Success test | The delivered experience honours published commitments | Required capabilities are coherent and assessable | Activities prepare learners and elicit usable evidence |
| Typical owner | Accountable program lead with learner input | Curriculum or subject lead | Educator or instructional designer |
| Common error | Promising flexibility that delivery cannot support | Confusing topic coverage with competence | Confusing activity completion with learning |
| Revision trigger | Changed audience, credential, workload, participation, or support | Changed capabilities, prerequisites, or evidence standards | Learning difficulty, inaccessible tasks, weak feedback, or delivery constraints |

Ownership describes accountability, not professional exclusivity. Learners, accessibility specialists, assessors, and operational staff may contribute throughout.

### 6.2 Classifying ambiguous decisions

| Decision | Primary layer | Required cross-layer check |
| --- | --- | --- |
| Completion requires evidence of independent reasoning | Curriculum | Contract discloses conditions; task can elicit reasoning |
| Learners may complete the program without attending live | Experience Contract | All required teaching and assessment have an asynchronous route |
| Replace a video with an annotated worked example | Instructional Design | Capability coverage, access, and workload still fit |
| Accept oral or written submissions | Instructional Design, if both elicit the required capability | Curriculum does not specifically require written communication; contract remains accurate |
| Add a compulsory prerequisite | Curriculum for the learning requirement | Contract entry assumptions and affected learners need review |
| Require paid software | Instructional Design proposes the tool | Contract costs and access conditions need explicit review |
| Change the rubric's minimum standard | Curriculum | Contract's completion claim and transition arrangements need review |

### 6.3 Comparison with familiar artifacts and approaches

| Artifact or approach | Usual emphasis | How the envelope can use it | Limitation of the comparison |
| --- | --- | --- | --- |
| Syllabus or course outline | Scope, schedule, policies, assessment | Links contract, curriculum, and delivery information in one learner-facing view | Syllabi vary widely and may already contain all three layers |
| Constructive alignment | Intended performances, activities, assessment | Provides an alignment principle across curricular and instructional decisions | The three-layer vocabulary is not Biggs's model |
| Resource library | Discoverable information and media | Supplies assets for selected learning tasks | Assets alone do not specify a learning progression |
| Lesson plan | A particular instructional episode | Implements part of Layer 3 | A lesson need not restate the entire program contract |
| Learning management system | Distribution, participation, assessment administration | Hosts published pathways and collects evidence | Software configuration cannot establish educational validity |
| Experience Envelope | Explicit commitments plus learning requirements plus implementation | Makes decision ownership and change consequences inspectable | Its added value requires empirical evaluation |

This is a comparison of purposes, not a ranking of effectiveness.

## 7. A complete worked example

### 7.1 Scenario: evidence-informed workplace decisions

A provider offers a two-week experience for adults who must make decisions from short reports. All specifications and thresholds in this example are illustrative.

**Contract EC-01.** Participants receive an asynchronous route, an estimated 6–8 hours of work, three case choices, one draft review within two business days, and accessible text versions of required media. Completion recognizes performance against a published rubric. The experience does not confer professional accreditation. Learners can request clarification or an access adjustment through a named support channel.

**Curriculum CU-01.** The required capabilities are C1 source evaluation, C2 claim interpretation, and C3 justified recommendation. All pathways must produce evidence for all three. A learner may bypass introductory practice after a diagnostic review but must still satisfy the final evidence standard.

**Instructional design ID-01.** Participants complete a diagnostic, study an annotated example where needed, practise on a shared case, revise after feedback, and submit a decision brief on a new case. An oral explanation with an accessible transcript route is available because the target is reasoning rather than writing fluency. A short follow-up check revisits key distinctions seven days later; it is used for evaluation rather than the completion award.

### 7.2 Traceability matrix

| Contract commitment | Curriculum requirement | Instructional implementation | Release evidence |
| --- | --- | --- | --- |
| Practical decision capability | C1, C2, C3 | New-case recommendation and explanation | Task-to-criterion mapping reviewed |
| No statistics prerequisite | Teach relevant foundations before C2 | Baseline and denominator mini-example | Diagnostic and novice walkthrough |
| Asynchronous completion | Same evidence criteria across routes | Recorded demonstration, discussion alternative, asynchronous submission | Full route completed without live attendance |
| Draft feedback | Opportunity to identify and address gaps | One rubric-linked review and revision | Staffing capacity and response process confirmed |
| Accessible media | Capability requirements do not depend on an incidental media format | Text alternatives and accessible task documents | Accessibility review and learner testing |
| Bounded workload | Prioritize necessary capability development | Core route plus clearly optional extension | Pilot time records, including struggling learners |

### 7.3 Example assessment rubric

Use four descriptive levels for each criterion: **absent**, **developing**, **meets**, and **extends**. In this example, completion requires at least “meets” on every criterion. The assessor records the reason for each judgment; a total score cannot compensate for a missing essential capability.

| Criterion | “Meets” descriptor | Capability |
| --- | --- | --- |
| Source judgment | Explains the relevance of two sources and identifies a limitation that affects their use | C1 |
| Claim interpretation | Interprets the supplied comparison accurately and does not turn association into an unsupported causal conclusion | C2 |
| Recommendation | Connects the recommended action to evidence and considers a plausible alternative | C3 |
| Uncertainty | States a consequential assumption or uncertainty and explains what additional evidence could change the decision | C2, C3 |

Before release, assessors should apply these descriptors to sample submissions and resolve disagreements. The rubric is an authored instrument that needs testing; its presence alone does not establish validity or reliable scoring.

## 8. Example learner pathways

An envelope permits different routes when they preserve the required learning evidence and published participation conditions. Pathways respond to demonstrated readiness, access needs, and goals. They should not assign people to fixed “learning styles.”

```mermaid
flowchart TB
  S["Start: review contract and complete diagnostic"] --> D{"What support is needed?"}
  D -->|"Foundations needed"| N["Guided route: examples and supported practice"]
  D -->|"Some relevant experience"| M["Core route: targeted practice and feedback"]
  D -->|"Strong prior evidence"| X["Challenge route: early performance attempt"]
  X --> G{"Required evidence sufficient?"}
  G -->|"No"| M
  G -->|"Yes"| A["Common evidence review against C1, C2, C3"]
  N --> A
  M --> A
  A --> R{"All essential criteria met?"}
  R -->|"No"| F["Targeted feedback and supported reassessment"]
  F --> A
  R -->|"Yes"| C["Recognize achievement and offer transfer extension"]
```

**Figure 4.** Different preparation routes converge on common evidence standards. Reassessment follows the published limits and support commitments; the loop does not imply unlimited attempts.

| Learner situation | Proposed pathway | What can vary | What stays stable |
| --- | --- | --- | --- |
| New to evidence interpretation | Worked example, guided practice, feedback, new case | Amount of scaffolding and practice | C1–C3 and essential criteria |
| Experienced practitioner | Diagnostic review, targeted practice, new case | Introductory material and practice volume | Required evidence and assessment integrity |
| Irregular schedule | Entirely asynchronous route | Timing within the published window | Support entitlement and quality criteria |
| Uses assistive technology | Accessible formats and an agreed submission route | Interface, media, and response format where appropriate | The capability being assessed |
| Wants deeper application | Core route plus optional sector-specific extension | Case complexity and additional feedback if available | Completion requirements do not silently expand |

A diagnostic is fallible. Learners need a clear way to change routes, and staff should check whether bypass decisions create later difficulties. Faster completion is not sufficient evidence that a pathway is educationally equivalent.

## 9. Change-impact flow and versioning

The point of separation is controlled adaptation. Every change begins with its reason and likely effect, then follows dependencies to the affected commitments, capabilities, and tasks.

```mermaid
flowchart TB
  P["Propose change and record reason"] --> L["Identify affected layer and linked artifacts"]
  L --> C{"Published commitment affected?"}
  C -->|"Yes"| CR["Review contract, capacity, communication, and transition"]
  C -->|"No"| Q{"Capability or evidence standard affected?"}
  CR --> Q
  Q -->|"Yes"| QR["Review curriculum, assessments, and comparability"]
  Q -->|"No"| I["Review instructional implementation"]
  QR --> I
  I --> T["Test affected routes, access, workload, and traceability"]
  T --> G{"Review criteria satisfied?"}
  G -->|"No"| P
  G -->|"Yes"| V["Approve version and publish change record"]
  V --> M["Monitor learner effects and keep a rollback route"]
```

**Figure 5.** A change can originate in any layer. Approval depends on its consequences, not the file in which it first appears.

### 9.1 Change-impact matrix

| Proposed change | Contract impact | Curriculum impact | Instructional impact | Required action |
| --- | --- | --- | --- | --- |
| Correct a misleading chart label | Check whether published claims were affected | Recheck interpretation and scoring consequences | Replace resource and linked feedback | Correct promptly; notify affected learners if material |
| Replace a video with accessible text | Usually limited; check published access and mode commitments | Confirm the same necessary concepts remain | Redesign prompts and timing | Test comprehension, route completion, and workload |
| Add statistical modelling | Scope and entry assumptions likely change | New capability and prerequisite relationships | New teaching and assessment | Reapprove scope before introducing it to learners |
| Remove draft feedback | Support commitment changes | Check effects on evidence opportunities | Redesign revision process | Review and communicate transition; do not silently withdraw support |
| Replace written brief with oral defence | Check participation and scheduling | Check whether writing was an assessed capability | New task and assessor guidance | Establish comparable evidence and accessible participation |
| Change completion standard | Completion claim may change | Evidence threshold changes | Rubric and examples change | Publish a new standard and explicit cohort transition |

### 9.2 A practical change record

Assign stable IDs to commitments, outcomes, criteria, activities, assessments, and sources. Record the version each published experience uses. A release entry should identify the change, rationale, affected IDs, reviewer, effective date, communication, transition treatment, and rollback decision.

For example: replacing an expert video with annotated text affects ID-DEM-01 and its C2 mapping. If testing shows that the text route requires substantially more time, EC-01's workload estimate must also be reviewed. A nominally instructional change can therefore reach the contract.

Version numbers are local conventions. Their value comes from an intelligible record and reproducible published requirements, not from a particular numbering scheme.

## 10. AI-assisted authoring workflow

AI can assist with drafting, alternatives, formatting, and consistency checks. Responsibility for commitments, disciplinary accuracy, evidence standards, and release remains with accountable people. UNESCO's guidance provides a human-centred policy basis for retaining human agency and evaluating pedagogical appropriateness; it is guidance rather than experimental evidence for this workflow. [Miao & Holmes, 2023](https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research).

```mermaid
flowchart LR
  B["Human brief and source packet"] --> C["Draft Experience Contract"]
  C --> H1["Human review: promise and feasibility"]
  H1 --> U["Draft Curriculum and evidence map"]
  U --> H2["Human review: accuracy and standards"]
  H2 --> I["Draft Instructional Design variants"]
  I --> K["Check citations, alignment, access, and syntax"]
  K --> P["Pilot with learners and assessors"]
  P --> H3["Human release decision"]
  H3 --> R["Publish version and retain provenance"]
```

**Figure 6.** The review points are proposed controls. They reduce opportunities for unnoticed errors but cannot guarantee correctness.

### 10.1 Authoring stages and deliverables

| Stage | AI-assisted work | Human responsibility | Output |
| --- | --- | --- | --- |
| Establish evidence | Extract candidate claims from supplied sources; flag gaps | Open sources, judge relevance, check claim strength | Claim-to-source ledger |
| Define contract | Draft plain-language commitments and identify ambiguities | Confirm audience, resources, exclusions, and feasibility | Approved contract version |
| Specify curriculum | Suggest outcomes, prerequisite links, and evidence criteria | Check disciplinary accuracy, necessity, and assessability | Capability and evidence map |
| Design instruction | Generate worked examples, tasks, feedback, and route alternatives | Judge suitability, accessibility, workload, and evidence quality | Linked activity and assessment specifications |
| Review across layers | Flag orphan tasks, unsupported claims, conflicting conditions | Resolve each issue and record decisions | Review record and revised artifacts |
| Pilot and publish | Summarize observed problems and prepare release files | Interpret learner evidence and approve release | Versioned envelope and change record |

### 10.2 A bounded authoring prompt

```text
Use the approved contract and curriculum below as constraints.
Draft two instructional routes for capabilities C1, C2, and C3.
For every activity, identify the capability, learner action,
required resource, feedback, and observable evidence.
Keep the asynchronous route complete and estimate workload.
Do not add prerequisites, tools, paid resources, completion
criteria, or support promises without flagging a proposed change.
Use only the supplied source packet for research claims.
Give source identifiers and distinguish findings from inferences.
If support is absent, mark the claim as needing verification.
List unresolved assumptions for human review.
```

The prompt helps constrain a draft; it does not enforce the constraints. Review the output against the approved specifications.

### 10.3 Source and assessment safeguards

Open and inspect every source relied upon. A plausible title or valid DOI does not establish that a source supports a particular claim. Keep publication metadata, the supported claim, relevant location, access level, and the author's inference in the source ledger. Never treat AI-generated citations as verified evidence.

For assessment materials, check model-generated answers against the rubric and discipline. Review whether examples inadvertently reveal the final answer. Use approved data arrangements for learner submissions and sensitive institutional material. Record substantive AI assistance and human review in the authoring history.

If learners may use AI, state the permitted uses before assessment and design evidence accordingly. A follow-up explanation can help examine reasoning, but it should be accessible, aligned with the capability, and covered by the published participation conditions.

## 11. Evidence, limitations, and evaluation

### 11.1 Claim-to-evidence matrix

| Research-supported proposition | Source and evidence type | Proposed application | Limit on the inference |
| --- | --- | --- | --- |
| Outcomes, activities, and assessment should be aligned | Biggs (1996), conceptual account and illustration | Trace capabilities to tasks and criteria | Does not test this three-layer architecture |
| Learning depends on prior knowledge, context, culture, and motivation | NASEM (2018), consensus research synthesis | Check entry assumptions and contextual suitability | Broad synthesis does not prescribe this specification format |
| Autonomy, competence, and relatedness matter to motivation | Ryan and Deci (2000), theoretical and research synthesis | Make meaningful choice and support explicit | Does not prove that a contract raises motivation |
| Learners regulate goals, performance, and reflection | Zimmerman (2002), theoretical overview | Provide progress criteria and reflective prompts | A published pathway alone does not teach self-regulation |
| Problem-centred instruction can integrate activation, demonstration, application, and integration | Merrill (2002), synthesis of design theories | Use a worked case followed by application | Not an experimental comparison of the proposed routes |
| Prior knowledge and working-memory demands affect instructional suitability | Sweller et al. (2019), theory and evidence review | Adjust scaffolding and remove unnecessary task demands | Exact design choices need contextual testing |
| Assessment evidence can guide subsequent learning | Black and Wiliam (1998), research review | Give actionable feedback and revision opportunities | Effects depend on implementation and context |
| Feedback can support learner self-regulation | Nicol and Macfarlane-Dick (2006), conceptual synthesis | Ask learners to interpret criteria and choose next actions | Does not validate every feedback technique |
| Retrieval can improve delayed retention | Roediger and Karpicke (2006), experiments | Include delayed retrieval of important distinctions | Retention of prose is not the same outcome as professional transfer |
| Active learning improved outcomes on average in the studied undergraduate STEM courses | Freeman et al. (2014), meta-analysis | Include meaningful learner reasoning and practice | The population and interventions differ from many other settings |

The context-sensitive view of learning summarized above is developed in [How People Learn II](https://doi.org/10.17226/24783). The active-learning finding comes from [Freeman et al.](https://doi.org/10.1073/pnas.1319030111). All applications in the third column are design inferences made in this paper.

### 11.2 An empirical comparison chart, with a narrow scope

Freeman and colleagues synthesized 225 studies comparing active learning and traditional lecturing in undergraduate STEM. Their reported average failure rates were 21.8% for active learning and 33.8% for traditional lecturing. The chart below reproduces those aggregate descriptive rates, not an estimated effect of the Experience Envelope. [Freeman et al., 2014](https://www.pnas.org/doi/10.1073/pnas.1319030111).

```text
Reported average failure rate in the meta-analysis
Scale: one full block is approximately two percentage points

Active learning       ███████████        21.8%
Traditional lecturing █████████████████  33.8%

Absolute difference between these reported averages: 12.0 percentage points
```

These aggregate rates do not isolate the contribution of contract clarity, layer separation, or this paper's workflow.

### 11.3 What remains unproven

No study cited here evaluates an intervention named the Experience Envelope with these three layers. The architecture is an original synthesis for this paper. It may improve clarity while adding documentation costs. Separating artifacts may also create coordination problems if ownership is unclear or linked requirements fall out of sync.

Alternative explanations matter. An apparent benefit might result from more authoring time, better staffing, improved feedback, or a stronger curriculum rather than layer separation. A fair evaluation must distinguish those possibilities as far as feasible.

The literature used is a purposive selection of foundational sources, not a systematic review or a claim to cover all relevant scholarship through 2026. The examples are hypothetical, and no learner-outcome data were collected for this paper. Accessibility and cultural suitability require direct review with intended learners.

### 11.4 Evaluation plan

| Question | Measure | Proposed method | Interpretation caution |
| --- | --- | --- | --- |
| Do learners understand the offer? | Accuracy of learner explanations of requirements and support | Short scenario questions before participation | Self-reported clarity alone is insufficient |
| Does the design deliver required learning? | Rubric performance on a new case | Independently scored performance tasks with assessor calibration | Baseline differences and scoring bias can distort comparison |
| Is learning retained? | Delayed recall and application | Follow-up after a defined interval | Nonresponse may bias results |
| Do routes preserve the standard? | Criterion-level performance and missing evidence | Compare routes while accounting for diagnostic readiness | Self-selection limits causal interpretation |
| Are commitments fulfilled? | Feedback response, route availability, access issues, time spent | Service records and learner walkthroughs | Average performance can conceal subgroup problems |
| Does layer separation improve maintenance? | Authoring effort, missed dependencies, correction time | Compare documented change tasks with an existing process | Novelty and team experience affect results |

A staged pilot should first test feasibility and comprehension. A subsequent comparative study could use matched offerings or a randomized design where feasible, with common outcome measures, adequate sample planning, and explicit reporting of uncertainty. Predefine meaningful success criteria locally; this paper supplies no validated universal thresholds.

Collect only information needed for evaluation. Report withdrawals and missing data, examine route and access differences where appropriate, and use qualitative feedback to explain quantitative patterns. Completion rate and learner satisfaction are useful operational signals but do not establish competence.

## 12. Implementation and release checklist

Begin with one existing experience and its real constraints. Extract commitments from advertisements, registration pages, staff instructions, and assessment policies. Reconcile contradictions before redesigning activities. Map the capabilities and evidence already required, then inspect the existing tasks against that map.

Use one versioned specification if that is sufficient. Separate responsibilities do not require duplicate prose across multiple documents.

| Release gate | Evidence needed | Accountable reviewer |
| --- | --- | --- |
| Contract feasible | Published commitments match staffing, timing, technology, and support capacity | Program lead |
| Curriculum coherent | Outcomes, dependencies, scope, and evidence criteria reviewed | Curriculum or subject lead |
| Instruction aligned | Every essential capability is practised and assessed; mandatory tasks have a rationale | Instructional lead |
| Assessment interpretable | Sample work scored, disagreements examined, and criteria clarified | Assessment lead |
| Participation workable | Full asynchronous and accessible routes tested where promised | Delivery and accessibility reviewers |
| Evidence accurately represented | Claims checked against sources; inferences labelled | Evidence reviewer |
| Publication usable | Links, tables, diagrams, and MDX build checked in the publishing environment | Publishing owner |
| Change process ready | Version, effective date, communication, transition, and rollback recorded | Release owner |

Before release, confirm:

- Every public promise has an owner and a feasible delivery mechanism.
- Each essential capability has a clear evidence standard and an assessment that can elicit it.
- Each mandatory activity supports a capability or a disclosed participation requirement.
- Learners have appropriate opportunities to practise before consequential assessment.
- Every promised pathway can be completed under its stated conditions.
- Workload estimates and support response times have been tested or are clearly identified as estimates.
- Sources support the claims made, and proposed decisions are distinguishable from findings.
- Material changes have an explicit treatment for learners already enrolled.

## 13. Conclusion

The Experience Envelope separates three decisions that educational work needs to coordinate: the commitments made to participants, the learning and evidence required, and the experiences designed to support that learning. Its three layers are the Experience Contract, Curriculum, and Instructional Design.

The framework's proposed contribution is an inspectable relationship among those decisions. It allows a team to explain why an activity exists, what may vary across pathways, and which consequences must be reviewed when something changes.

The research supports selected principles used in this architecture. Demonstrating that the architecture itself improves learning, equity, clarity, or authoring efficiency remains an empirical task. The responsible position is to use it as a research-informed design proposal, publish its assumptions, and evaluate its effects.

## 14. Academic bibliography and source notes

Bibliographic details and the basis for the claims were checked against publisher pages, original-paper copies, institutional sources, or bibliographic records on 23 October 2026. Access notes distinguish abstract-level checks from full-text access. These entries are a selected bibliography rather than a systematic review.

1. **Biggs, J. (1996).** Enhancing teaching through constructive alignment. *Higher Education, 32*(3), 347–364. [DOI and publisher record](https://doi.org/10.1007/BF00138871). Supports the alignment principle. Verification: publisher metadata and abstract; claims here stay within that account.

2. **National Academies of Sciences, Engineering, and Medicine. (2018).** *How people learn II: Learners, contexts, and cultures.* The National Academies Press. [DOI](https://doi.org/10.17226/24783). Supports attention to prior knowledge, motivation, culture, and learning context. Verification: institutional publication record and [Chapter 7 text](https://nap.nationalacademies.org/skim.php?act=nap&chap=135-162&record_id=24783), especially its concluding implications.

3. **Ryan, R. M., & Deci, E. L. (2000).** Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. *American Psychologist, 55*(1), 68–78. [DOI](https://doi.org/10.1037/0003-066X.55.1.68). Supports the motivation discussion. Verification: [original article PDF hosted by the theory's research organization](https://selfdeterminationtheory.org/SDT/documents/2000_RyanDeci_SDT.pdf), including its account of autonomy, competence, and relatedness.

4. **Zimmerman, B. J. (2002).** Becoming a self-regulated learner: An overview. *Theory Into Practice, 41*(2), 64–70. [DOI and publisher record](https://doi.org/10.1207/S15430421TIP4102_2). Supports the account of goal setting, monitoring, and reflection. Verification: publisher search record and an [original-paper copy](https://www.normanjackson.co.uk/uploads/1/0/8/4/10842717/zimmerman_-_2002_-_becoming_a_self-regulated_learner_an_overview__1_.pdf). The publisher and seven-page paper support pages 64–70; one ERIC record lists a different end page.

5. **Merrill, M. D. (2002).** First principles of instruction. *Educational Technology Research and Development, 50*(3), 43–59. [DOI and publisher record](https://doi.org/10.1007/BF02505024). Supports the problem-centred design lens and five associated principles. Verification: publisher metadata and abstract. This is a synthesis of design theories, not a trial of the envelope.

6. **Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019).** Cognitive architecture and instructional design: 20 years later. *Educational Psychology Review, 31*, 261–292. [DOI and open-access article](https://doi.org/10.1007/s10648-019-09465-5). Supports the discussion of working-memory demands, prior knowledge, and instructional guidance. Verification: publisher metadata and open article, particularly the introduction and treatment of instructional effects.

7. **Black, P., & Wiliam, D. (1998).** Assessment and classroom learning. *Assessment in Education: Principles, Policy & Practice, 5*(1), 7–74. [DOI](https://doi.org/10.1080/0969595980050102). Supports the formative-assessment discussion. Verification: [original article PDF hosted by the University of California, Riverside](https://assess.ucr.edu/sites/default/files/2019-02/blackwiliam_1998.pdf), including its definition and review of formative assessment. The original issue year is 1998; later online dates do not change it.

8. **Nicol, D. J., & Macfarlane-Dick, D. (2006).** Formative assessment and self-regulated learning: A model and seven principles of good feedback practice. *Studies in Higher Education, 31*(2), 199–218. [DOI and publisher record](https://doi.org/10.1080/03075070600572090). Supports feedback as a means of supporting learner self-regulation. Verification: publisher abstract and metadata. The journal issue is 2006 despite a later online-publication date.

9. **Roediger, H. L., III, & Karpicke, J. D. (2006).** Test-enhanced learning: Taking memory tests improves long-term retention. *Psychological Science, 17*(3), 249–255. [DOI](https://doi.org/10.1111/j.1467-9280.2006.01693.x). Supports the bounded delayed-retention claim. Verification: [PubMed metadata and original-study abstract](https://pubmed.ncbi.nlm.nih.gov/16507066/), describing two experiments with prose passages and immediate and delayed tests.

10. **Freeman, S., Eddy, S. L., McDonough, M., Smith, M. K., Okoroafor, N., Jordt, H., & Wenderoth, M. P. (2014).** Active learning increases student performance in science, engineering, and mathematics. *Proceedings of the National Academy of Sciences, 111*(23), 8410–8415. [DOI and original article](https://doi.org/10.1073/pnas.1319030111). Supports the scoped undergraduate STEM comparison. Verification: publisher article and [PubMed record](https://pubmed.ncbi.nlm.nih.gov/24821756/); aggregate failure rates and effect measures checked in the paper. These are not Experience Envelope outcome data.

### Additional policy source

11. **Miao, F., & Holmes, W. (2023).** *Guidance for generative AI in education and research.* UNESCO. [Official publication page](https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research). Supports the human-centred framing of AI use. Verification: UNESCO authorship and publication metadata and the guidance's discussion of human agency and pedagogically appropriate interaction. This is policy guidance, distinguished here from academic empirical evidence.

## Appendix A: Reusable specification

Copy and complete this structure for a specific experience. Its fields are a proposed authoring schema.

```yaml
experience:
  id: EE-001
  title: Evidence-informed decisions
  version: '1.0'
  accountable_owner: Program lead
  effective_date: To be set before release

  experience_contract:
    id: EC-01
    purpose: Make a defensible recommendation from supplied evidence
    audience: Adult workplace learners
    entry_assumptions: Can read short reports; no statistics prerequisite
    scope: Source evaluation, claim interpretation, and recommendation
    exclusions: Statistical modelling and professional accreditation
    participation: Complete asynchronous route; optional live clinic
    workload: Estimated 6-8 hours across two weeks
    support: One draft review within two business days
    access: Accessible text alternatives and an adjustment contact
    choices: Three cases and equivalent response formats where suitable
    completion_claim: Meets every essential criterion for C1, C2, and C3
    assessment_conditions: Published permitted assistance and tool use
    changes: Notify material changes and record cohort transition

  curriculum:
    id: CU-01
    capabilities:
      - id: C1
        performance: Evaluate source relevance and credibility
        evidence: Source judgment with a consequential limitation
      - id: C2
        performance: Interpret claims and uncertainty accurately
        evidence: Chart interpretation without unsupported causal claims
      - id: C3
        performance: Justify a recommendation
        evidence: Rationale, alternative, and condition for revision
    dependencies:
      - C1 supports C3
      - C2 supports C3
    essential_standard: Meets every published criterion

  instructional_design:
    id: ID-01
    routes:
      - Guided
      - Core
      - Challenge with diagnostic review
    activities:
      - Diagnostic
      - Annotated worked example
      - Supported practice
      - Draft feedback and revision
    assessment: New-case recommendation and explanation
    evaluation: Delayed retrieval and application check
    traceability: Map each task and criterion to capability IDs

  release:
    evidence_review: Required
    learner_walkthrough: Required
    assessment_calibration: Required
    access_review: Required
    publishing_check: Required
    change_record: Required
```

## Appendix B: MDX publishing notes

This file uses standard MDX with an exported metadata object, Markdown headings and links, GitHub-flavoured Markdown tables, and fenced Mermaid source. It contains no external component imports, remote scripts, or runtime data requests.

Compile with an MDX-compatible toolchain and enable `remark-gfm` to render the tables as tables. Configure the host to render code fences labelled `mermaid` using Mermaid. Without that integration, the diagrams remain readable source blocks. Their adjacent captions and surrounding prose explain their meaning independently of the graphical rendering.

The metadata is a JavaScript export rather than YAML front matter, avoiding a front-matter parsing dependency. A publishing system with a required front-matter schema can translate the metadata during integration. Heading-anchor conventions are host-dependent; the contents links follow common GitHub-style slugs and should be checked in the target site.

Syntax validation establishes that the source can compile under the tested configuration. It does not certify accessibility, responsive layout, or integration with every MDX host. Mermaid rendering and table appearance should be checked again in the final publishing environment.