Skip to content

Cognitive Task Analysis

Design Method · All design methods
Evidence · 1 claim (1 mixed) · 5 studies (3 review, 1 causal, 1 theoretical), q1–q3 · 0 of 5 report an effect size

Description

Ask an expert how they do something and they will describe the procedure. Watch them and you will see the procedure plus a great deal of judgement: which cue they attended to, which they dismissed, when they decided the situation had changed, what they were ready for that did not happen. Cognitive task analysis (CTA) is the set of methods for recovering that second layer.

It exists because of a specific and well-attested failure. Experts are unreliable reporters of their own expertise — the knowledge that has become automatic is the knowledge they no longer notice using, so it is systematically omitted from their account. A procedural task analysis faithfully records what the expert said, and the resulting instruction teaches the steps while leaving out the judgements that make the steps work. Learners then perform the procedure correctly in the classroom and fail in the situations it was meant to prepare them for.

CTA is the missing third method beside the two this wiki already documents. Task Analysis decomposes a task into steps; Learning Hierarchy Task Analysis orders skills by prerequisite; CTA goes after what the practitioner decides. The three are complementary and answer different questions, and the giveaway that CTA is the one you need is a step in the procedure that reads "assess the situation", "use judgement", or "as appropriate".

The best-known packaging is ACTA — applied cognitive task analysis — a practitioner's toolkit designed to be usable by people who are not cognitive psychologists, comprising a task diagram, a knowledge audit, a simulation interview, and a cognitive demands table that turns the findings into design decisions.

Design Implications

Context

Requirements

  • Access to genuine experts, and to more than one: a single expert's account is a case study, and CTA's characteristic findings are the differences between practitioners
  • Real incidents or realistic simulations to interview against. CTA interviews probe specific remembered events; asked in general, an expert reverts to the procedure
  • Interviewer skill. The method is a set of probes, and an interviewer who accepts the first answer collects the same content a procedural analysis would
  • Somewhere for the findings to go. A cognitive demands table that does not become assessment items, case selection or scaffolding is analysis nobody used

Constraints

  • Expensive. Several interviews, transcription and analysis, for one task. It is not the default method; it earns its cost where the judgement is the skill
  • Poor fit for genuinely procedural work. If the task really is a fixed sequence, CTA finds nothing a procedural analysis missed
  • Expert disagreement is data, not noise, and it is uncomfortable data: the method routinely reveals that two respected practitioners solve the same situation on different cues
  • Over-teaching the tacit. Making every expert cue explicit and handing it to a novice can produce exactly the redundancy Guidance that helps novices can become redundant as expertise grows. [~M] describes — the findings should shape what learners practise deciding, not become another list to memorise
  • Retrospective accounts are reconstructions, and confident ones. The simulation interview exists because memory of a decision is more reliable when the situation is in front of the expert again

Implementation Variability

  • ACTA — the four-instrument toolkit, designed for non-specialists and the usual starting point
  • Critical decision method — deep retrospective interviews around one non-routine incident, sweeping the timeline repeatedly at increasing depth
  • Concept mapping / structural methods — recovering how an expert organises the domain rather than how they act in it
  • Observation-led variants — think-aloud during real or simulated performance, where the work can be watched without being disturbed

Target Learners

  • Designers serving learners who must decide rather than execute — clinical, diagnostic, tactical, editorial and troubleshooting work
  • Novices moving into contexts with high variability, where the transfer failure CTA prevents is most costly
  • Less useful where learners need fluency in a stable procedure; there, Task Analysis and part-task practice serve better Guidance that helps novices can become redundant as expertise grows. [~M]

Target Learning Goals

  • Cue recognition — noticing what matters in a situation before acting on it
  • Decision rules and their boundaries, including when a practitioner abandons the standard approach
  • Situation assessment and anticipation: what an expert expects next, and what would surprise them
  • Case and scenario selection: which situations a curriculum must include for the judgement to be exercised

Instructions

  1. Bound the task. Name the performance and the conditions. CTA on "teaching" yields nothing; CTA on "deciding when to abandon a lesson plan" yields a great deal
  2. Build a task diagram with the expert — a coarse three-to-six step decomposition — and ask which steps carry the cognitive difficulty. This scopes the expensive work onto one or two steps
  3. Run a knowledge audit on those steps: probe for past-and-future thinking, big-picture awareness, noticing, improvising, and the errors a novice characteristically makes. Take a concrete example for every probe answered
  4. Run a simulation interview. Present a realistic scenario, stop it at decision points, and ask what they notice, what they expect, what they would do, and what a novice would get wrong here
  5. Interview several experts and record where they diverge; do not average them into one composite practitioner
  6. Build the cognitive demands table — for each difficulty: why it is hard, what errors novices make, and the cues and strategies experts use
  7. Convert it into design. The demands table maps onto Case Studies and scenario selection, Think Aloud modelling of the expert's reasoning, assessment items that require a judgement rather than a step, and the difficulty ordering for simple-to-complex sequencing
  8. Validate the findings with an expert who was not interviewed

Examples

  • Clinical reasoning curricula — CTA of diagnosis reveals the cue patterns that distinguish an experienced clinician's early hypothesis from a novice's exhaustive checklist, and turns them into case selection
  • Troubleshooting and maintenance training — the difference between the manual's procedure and how a technician actually narrows a fault
  • Firefighting, aviation and other high-tempo domains — the setting most CTA methods were developed in, where the decisions are consequential and the procedures alone demonstrably insufficient

Key Sources

  • Militello, & Hutton. Applied cognitive task analysis (ACTA).
  • Stefaniak. Ch. 7 in Design for Learning. EdTech Books.
  • Clark, R. E., Feldon, D. F., van Merriënboer, J. J. G., Yates, K., & Early, S. (2008). Cognitive task analysis. In J. M. Spector et al. (Eds.), Handbook of research on educational communications and technology (3rd ed., pp. 577–593). Lawrence Erlbaum Associates.
  • Schraagen, J. M., Chipman, S. F., & Shalin, V. L. (Eds.). (2000). Cognitive task analysis. Lawrence Erlbaum Associates. doi:10.4324/9781410605795
Page metadata
Field Value
type method
id cognitive-task-analysis
title Cognitive Task Analysis
description A family of interview and observation methods for eliciting the cues, judgements and decision rules an expert uses but cannot readily articulate — the tacit content that a procedural task analysis records as a single step.
status review
generated by claude/unspecified on 2026-09-02
trust unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.