Cognitive Walkthrough¶
Protocol — instantiates Cognitive Apprenticeship Modeling
Replays a completed piece of expert work move by move so the hidden decisions behind it — what was noticed, weighed, and ruled out — become inspectable.
A Cognitive Walkthrough takes a finished piece of expert work — an incident, a case, a decision trace, an artifact — and replays it step by step, pausing at each juncture to reconstruct what the performer perceived, which options were live, and why they committed to one. Its defining move is that it works backwards on a completed trace: everyone can already see the outcome, so the protocol's whole purpose is to recover the decision sequence that produced it — the part that is invisible in the result. That retrospective, replayed structure is what separates it from a live think-aloud (reasoning narrated as it happens) and from shadowing (watching without a structured stop-and-probe rhythm). A facilitator freezes the replay at each decision point and asks what did you see here, and what else could this have been? — turning a smooth after-the-fact story back into the branching, uncertain choice it actually was.
Example¶
After a long shift, a fire-behavior analyst sits with two trainees and a recording of the day's radio traffic and map updates to replay a single decision: why the burn boss pulled a crew off a ridge an hour before the wind shifted. They don't start from the shift — they start from the ridge. At the moment the crew was committed, the facilitator freezes the replay: what was on the map then? what was the humidity trend? what were the live options — hold, reinforce, disengage — and why did each look the way it did? Only after the trainees have reconstructed the call from the information available at that instant does the replay roll forward.
What looked, in hindsight, like one obvious decision turns out to have rested on three cues the trainees hadn't known to watch — a subtle change in the smoke column's lean, a spot fire's timing, and a gut-level unease the boss couldn't fully justify. The trainees leave not with "pull crews when the wind shifts" but with the cue set and the tempo of doubt that let an expert act before the shift was obvious.
How it works¶
- Anchor on a real trace. Work from an actual record — logs, a case file, a recording, an artifact — not a tidied summary, so the messy, genuine decision points survive.
- Freeze at each junction. Stop the replay at every point where the expert chose, and reconstruct only the information available then, before rolling forward.
- Surface the road not taken. At each stop, name the options that were live and why they were rejected — the rejected branches carry as much of the expertise as the chosen one.
- Direct attention to the cues. Point explicitly at what the expert noticed (and what novices skate past), turning passive watching into guided looking.
Tuning parameters¶
- Replay granularity — how finely you slice the trace. Finer stops surface more hidden micro-decisions but stretch a ten-minute event into an hour; coarse stops move fast but skip the subtle cues.
- Hindsight discipline — how strictly you forbid outcome knowledge at each stop. Strict reconstruction protects the reasoning from being judged by its result; loose replay is faster but teaches "should have known."
- Who narrates — the original performer walking their own trace, versus a facilitator reconstructing someone else's. Self-narration is richer on tacit cues; third-party replay is more honest about gaps.
- Case selection — replaying a clean success, a near-miss, or a failure. Near-misses and recoveries usually expose the most reasoning per minute.
When it helps, and when it misleads¶
Its strength is recovering the decision behind the outcome — the part that vanishes once you know how things turned out — and doing it on real work rather than a staged demonstration. It is especially good at surfacing the cues and rejected options that a clean retelling smooths away.
Its central hazard is hindsight bias: once the outcome is known, both facilitator and learner drift toward "the right call was obvious," judging the reasoning by its result instead of by the information available at the time.[1] The classic misuse follows directly — running the walkthrough to assign blame or ratify a verdict already reached, which teaches learners to narrate a defensible story rather than to reason under uncertainty. The discipline that guards against both is to reconstruct each juncture strictly from what was knowable then, and to judge the quality of the decision, not the luck of the outcome.
How it implements the components¶
visible_reasoning— the protocol's core output: it externalizes what the expert noticed, weighed, and ruled out at each junction, turning an opaque outcome into an inspectable decision sequence.guided_observation_frame— the freeze-and-probe rhythm directs novice attention to the specific cues and decision points that matter, rather than leaving them to watch the replay passively.
It does not stage a live performance to model from (expert_model, authentic_practice_task — see Coached Practice Session and the live Think-Aloud Demonstration), coach the learner's own attempts (coaching_feedback), or persist a durable record (apprenticeship_record — see Annotated Decision Record). A walkthrough exposes reasoning after the fact; it does not itself grade practice.
Related¶
- Instantiates: Cognitive Apprenticeship Modeling — the walkthrough supplies the "make the hidden reasoning visible" move at the heart of the archetype, working on completed cases.
- Consumes: Annotated Decision Record — a good written record of the original decision makes the replay far sharper, though a raw log will do.
- Sibling mechanisms: Annotated Decision Record · Case Conference · Think-Aloud Demonstration · Shadowing with Debrief · Worked Example with Reasoning · Coached Practice Session
Editorial Notes¶
Form Classification¶
Form family: Communication, Facilitation & Learning
Rationale: Replays a completed piece of expert work move by move so the hidden decisions behind it — what was noticed, weighed, and ruled out — become inspectable, making its operative form a designed message, facilitated interaction, ritual, or learning activity that changes shared understanding.
Independent corroboration: The frozen evidence defines Cognitive Walkthrough as 'Replays a completed piece of expert work move by move so the hidden decisions behind it — what was noticed, weighed, and ruled out — become inspectable', so its operative form is Communication, Facilitation & Learning.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Education & Pedagogy
Origin pattern: Historically ambiguous
Present-day reach: Multi-domain
Rationale: Cognitive apprenticeship established retrospective, stepwise replay of expert work to make hidden choices inspectable for learners.
Related originating lineages:
- Cognitive Science — Hindsight-bias controls and decision reconstruction supply the elicitation discipline.
- Human-Computer Interaction — A separate established usability method shares the name and step-through format but evaluates learnability rather than expert reasoning.
Review resolution: Both reviewers place the defined replay method in education but flag the name collision. The cognitive-apprenticeship literature documents replay and making expert thinking visible, while the HCI literature defines cognitive walkthrough as a distinct learnability inspection. The final record therefore preserves education as primary and records the term as historically ambiguous.
Attribution caveat: The term also names the canonical HCI usability-inspection method, while this mechanism uses an apprenticeship-oriented replay of completed expert work.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; medium confidence.
Sources consulted:
- Collins, Brown, and Newman: Cognitive Apprenticeship
- Wharton et al.: The Cognitive Walkthrough Method
Notes¶
The term also names a well-known usability-inspection method, in which evaluators step through an interface as a first-time user to find learnability problems. That technique shares the step-by-step spirit but targets an artifact's usability, not a performer's reasoning; the apprenticeship sense here replays expert cognition, not an interface.
References¶
[1] Fischhoff, B. “Hindsight ≠ Foresight: The Effect of Outcome Knowledge on Judgment under Uncertainty”. Journal of Experimental Psychology: Human Perception and Performance 1(3), 288–299 (1975). Shows that outcome knowledge alters retrospective likelihood judgments and the perceived relevance of information available before the outcome. registry ↩