Cognitive Interview¶

Method — instantiates Tacit Knowledge Elicitation

Walks an expert back through one specific remembered episode in fine-grained detail, reinstating the original context to recover the cues, reasoning, and rule-breaks they can no longer summarize.

Ask an expert how they generally do something and you get an average — a smoothed generalization that has lost every particular. Cognitive Interview refuses the general question and anchors instead to one concrete episode: a single case the practitioner remembers, walked through moment by moment. Its defining move is context reinstatement — deliberately rebuilding the setting, sequence, and sensory detail of that specific event so that memory releases the cues and micro-decisions it can't produce cold. By keeping the whole interview inside one lived case rather than ranging across "how you usually work," it recovers the branching points, the uncertainty, and the moment the ordinary rule stopped fitting — the texture that compresses away the instant an expert answers in the abstract.

Example¶

An airline captain is asked, in a cognitive interview, about a specific go-around — the day she aborted a landing at low altitude rather than continuing. The interviewer doesn't ask "when do you go around?" (which would yield the manual's answer). She asks the captain to rebuild the approach second by second: the weather, the callouts, what she saw out the window, what the first officer said. Reinstated in the moment, the captain recovers cues she'd never volunteer to a general question — the runway lights "looked wrong, too bunched," a sink rate that "felt a half-second late to arrest," an instinct that arrived before the instruments confirmed it.

The interview then presses on the decision itself: What else could that have been? What would have made you continue instead? What was the first sign the normal picture wasn't holding? Those probes recover the branch she rejected (continuing, which the numbers still permitted) and the exact threshold at which she overrode the standard call. The output is a densely specified single case — cues, reasoning, and the rule-break — that no summary of her general practice would ever contain. What it is not is a map of her whole job; this interview went deep on one decision, not broad across the role.

How it works¶

Anchor to one real episode. Pick a specific remembered case — ideally recent and consequential — and stay inside it. Generalized questions ("how do you usually…") are banned precisely because they average away the detail.
Reinstate the context first. Rebuild the setting, sequence, and sensory frame before probing, so recall runs off the original experience rather than a reconstruction.
Probe why, not just what. Use counterfactual prompts — What else could this have been? What would have changed your call? What did you deliberately ignore? — to recover rejected alternatives and hidden uncertainty.
Chase the exception. Ask specifically for the first sign the ordinary rule didn't fit, which is where the tacit expertise concentrates.

Tuning parameters¶

Episode selection — which case to reconstruct. A dramatic or hard case yields more tacit content but risks over-weighting the atypical; a routine case calibrates the everyday but surfaces less.
Reinstatement depth — how thoroughly the original context is rebuilt before probing. Deeper reinstatement recovers more but lengthens the interview and can seed detail the interviewer implies.
Counterfactual pressure — how hard the interviewer pushes on alternatives and thresholds. More pressure surfaces rejected branches but risks the expert confabulating a tidier rationale than they actually had.
Recency window — how soon after the event. Sooner is sharper; later loses fidelity but gains the practitioner's reflective read of what mattered.

When it helps, and when it misleads¶

Its strength is depth on a single decision: by staying inside one reinstated episode it recovers the specific cues, rejected alternatives, and rule-break thresholds that any "how do you generally…" question dissolves. It is the sharpest tool for the why and what-if behind one consequential judgment.

Its failure modes are the failure modes of memory. Retrospective recall is reconstructive — the expert can, in perfect good faith, supply a cleaner rationale than they actually held at the time, and a suggestive interviewer can seed detail that the practitioner then "remembers."^[1] Depth on one case also buys narrowness: a single vivid episode may not represent the practitioner's typical practice, so building a curriculum on one interview over-fits the anecdote. The discipline that keeps it honest is to reinstate context before probing (so recall leads and the interviewer follows), keep prompts non-leading, mark reconstructed rationale as provisional, and sample several episodes — including ordinary ones — before treating any single case as the pattern.

How it implements the components¶

Cognitive Interview fills the probing components that a deep single-episode retrospective can reach:

cue_elicitation — context reinstatement recovers the specific signals the expert attended to in that episode, including ones they could not volunteer in the abstract.
decision_rationale_probe — counterfactual questioning recovers why the call was made, what alternative was rejected, and how confident the expert actually was.
exception_probe — the interview deliberately hunts the moment the ordinary rule stopped fitting and what signalled it.

It works from memory of one case, not the live scene (expert_practice_observation — Shadowing Session) and not the concurrent stream (sensory_or_embodied_cue_capture — Think-Aloud Protocol). It also does not build the broad, whole-task decomposition or test that the recovered knowledge reproduces (replication_probe — Cognitive Task Analysis).

Instantiates: Tacit Knowledge Elicitation — Cognitive Interview is the deep-probe method for recovering the reasoning and rule-breaks buried in a single expert judgment.
Sibling mechanisms: Cognitive Task Analysis · Critical Incident Technique · Expert Debrief · Decision Trace Review · Think-Aloud Protocol · Shadowing Session · Apprenticeship Observation · Simulation Replay · Case Library Capture · Judgment Aid · Task Analysis

Notes¶

Cognitive Interview and Critical Incident Technique both work from a single remembered case, but they differ in how the case is chosen and used. Critical Incident Technique selects cases because they are rare, high-information incidents and mines many of them for patterns; Cognitive Interview will happily go deep on an ordinary episode and cares about the full reconstructed texture of that one event, not about assembling an incident set.

References¶

[1] Reinstating the original context to improve recall, and following the interviewee's memory rather than leading it, are the core moves of the cognitive interview developed for eyewitness testimony (Fisher & Geiselman). The same literature is the source of the caution above: retrospective recall is reconstructive, and leading prompts can implant detail — which is why the method reinstates first and probes second. ↩