Niraj ChaurasiyaBuilding systems under uncertainty

EoL

Evidence of Learning

What observable evidence supports the claim that learning has occurred?

Evidence of Learning treats learning as a latent process that cannot be directly observed. Instead of asking only whether someone paid attention, completed a course, or felt that information made sense, the framework examines progressively stronger demonstrations of changed capability.

Purpose

To provide a practical vocabulary for reasoning about whether learning has occurred and what kind of evidence supports that conclusion.

Central principle

Exposure, attention, completion, and familiarity may accompany learning, but they are not independently sufficient evidence of learning.

Components

How the framework is organized

0101

Recall

Retrieve relevant information without relying entirely on the original source or prompt.

Can the learner recover the knowledge when the material is absent?
0202

Imitate

Reproduce a demonstrated process, pattern, or solution with guidance or a close example.

Can the learner follow and reproduce a known procedure?
0303

Perform Independently

Complete the task without direct step-by-step guidance.

Can the learner act without copying the original demonstration?
0404

Transfer

Apply knowledge or capability in a changed problem, environment, or representation.

Can the learner use the knowledge when the surface conditions change?
0505

Teach

Explain the idea, process, relationships, assumptions, and common mistakes to another person.

Can the learner make the underlying structure understandable?
0606

Create

Produce something new that depends on the learned knowledge or skill.

Can the learner use the knowledge as a building material?
0707

Reflection

Identify errors, limitations, uncertainty, changes in understanding, and remaining questions.

Can the learner examine the quality and boundaries of their own understanding?

Reasoning rules

Principles guiding its use

01

Learning is inferred from evidence rather than directly observed.

02

Different evidence supports different strengths of claim.

03

Recognition is weaker than independent recall.

04

Performance in one context does not automatically demonstrate transfer.

05

Teaching can expose gaps hidden during private familiarity.

06

Creation may demonstrate integration but does not guarantee correctness.

07

Reflection preserves uncertainty and makes revision possible.

Practical use

Where the framework can be applied

Applications

Evaluating whether a course or lecture produced learning.

Designing assessments beyond multiple-choice recognition.

Planning personal study sessions.

Evaluating technical tutorials.

Documenting learning through public engineering projects.

Designing educational recommendation signals.

Example 01

Learning SolidWorks

Context: A learner watches a tutorial explaining how to create a parametric link.

Application: Completion of the tutorial demonstrates exposure. Independent modeling, dimension changes, transfer to a new link, teaching the process, and reflecting on design intent provide progressively different evidence.

Example 02

Learning a programming concept

Context: A learner understands an example while reading it.

Application: EoL asks whether the learner can recall the concept, reproduce the pattern, solve independently, transfer it to a different problem, explain it, and build something with it.

Example 03

Watching educational shorts

Context: A user watches, replays, bookmarks, or marks a video Helpful.

Application: These behaviors may provide indirect evidence of attention or usefulness, but they do not independently demonstrate recall, performance, transfer, teaching, or creation.

Epistemic boundaries

What the framework does not solve

Limitations

Current limitations

The stages should not automatically be treated as a universal linear hierarchy.

Different forms of learning require different demonstrations.

Performance can be affected by anxiety, fatigue, context, and available tools.

Teaching fluently does not guarantee that every claim is correct.

Creation may contain misconceptions despite demonstrating integration.

The framework is conceptual and is not yet a validated measurement instrument.

Open questions

What remains unresolved

How should EoL differ across conceptual, procedural, perceptual, and motor learning?

How much evidence is sufficient before claiming learning occurred?

How should delayed retention be incorporated?

Should reflection be a separate dimension rather than a final stage?

How can evidence be evaluated without reducing learning to one score?

Version history

How the framework has changed

Expanded the framework to include Recall and Reflection.

  • Added Recall as an explicit evidence category.
  • Added Reflection to represent metacognition, limitations, and uncertainty.
  • Replaced the earlier Observation terminology.
  • Clarified the distinction between imitation and independent performance.

Initial framework for reasoning about observable evidence of learning.

  • Introduced Observation, Imitation, Independent Performance, Transfer, Teach, and Create.
  • Separated learning feelings from evidence of capability.