Clark Quinn's Learnings about Learning
(The Official Quinnovation blog)

19 September 2011

Cognitive Task Analysis

Clark @ 6:31 AM

While I argue strongly for stepping away more frequently from formally structured learning, not least because we overuse it, there are times when it is crucial.  As naysayers of informal learning like to point out, you wouldn’t want your pilot or heart surgeon to have picked up the task by reading a book. When performance is critical, you really want to understand what the important elements are, whether to train them or provide support.

A technique for doing that is Cognitive Task Analysis (CTA).  This is not a shortcut, it’s deep in terms of the knowledge elicitation techniques, the analytical task, and the representation of results.  Based in decades of cognitive research, integrating work on mental models, expertise, and more, it provides a mechanism to try to unearth the tacit understanding experts hold. Because experts compile away their knowledge to the point that they no longer have access to it, it is hard to get at this knowledge, and it takes a rigorous process.

While useful for system design, CTA is also valuable for designing performance support, and training.  The deep elicitation process can derive what the task really is, and what should be in the learner’s head and what support can and should be available.  When I talk about the performance ecosystem, particularly for complex tasks, you want just this sort of support to determine what should be distributed across formal learning and performance support.

One of the problems with CTA is that there have been a number of different approaches, and they tend to be buried in academic papers or proprietary processes. The good news is that there’s now a book about CTA, Working Minds, by Beth Crandall, Gary Klein, & Robert Hoffman, academics and practitioners.  It boils down the divergence into a fairly reasonable set of steps, with techniques that can be used at each stage.  The bad news is, of course, that it still is a daunting read, with considerable depth.

If you’ve got performances that absolutely have to be right, you’ll want to do the analysis ala CTA, and use it to decide what really needs to be in training, checklists, etc.  This goes deeper than HPT even, tho’ I think it’s as weak when it comes to the benefits of social learning, but I reckon it’s for expert *performance*, not innovation. That’s another layer.  Still, a valuable tool in the quiver of supporting performance.

Please at least understand what CTA is, and know when you need it. You may not need to be an expert in it, but you should at least be aware.


  1. Your last two posts have been really validating for us, Clark. We’ve just finished a 1 year effort to rewrite our process guides for selection and estimation, initiation, pre-design, design, development, testing, implementation, and lifecycle activities for digital solutions. Two huge themes in this rewrite were meaningful activities and a re-rack on expectations. This change effort targeted short-sighted views of technology that usually resulted in the construction of “content bombs”.

    Here’s an excerpt from the pre-design analysis section that touches on cognitive task analysis:

    “Among the strengths of the digital environment is the opportunity for ample practice of “thinking tasks” or cognitive tasks and the acquisition of mental models. Thinking tasks and cognitive tasks are referred to as covert tasks in the SABA Peak Performance System. The success of practice opportunities presented in the digital environment is largely dependent on the discovery of these cognitive tasks and the explication of mental models. Covert task definition is critical to the definition of accurate instructional objectives, meaningful activities, and successful acquisition of real world skills using Self-Paced eLearning.”

    We aren’t very far down the road validating the practices or defining the context of application for CTA. That work will begin this year. We will be using Crandall, Klein, and Hoffman’s book on the subject in a few test cases. We’d like to identify how much / how deep is enough and in what cases this type of analysis will be most beneficial.

    We’re a Harless Methodology shop and have been for well over a decade. This works pretty well for identifying the big parts but I believe it can miss some of the nuances that other methods can reveal. The challenge is in discovering which cases will make it worth the specialized effort.

    Comment by Steve — 20 September 2011 @ 6:52 AM

  2. Hi, Clark.

    Thanks for writing this post. CTA is useful whenever job tasks involve a lot of otherwise invisible cognitive stuff involving situation recognition, decision making, and problem solving. Representations of expertise from CTA have found their way into better user interface design, online information systems, and training that helped others solve workplace problems more like the organization’s best.

    CTA also puts controls in place that guard against the inaccurate and incomplete self-reporting that otherwise occurs when experts tell you want it is they think they do on the job. Ever try to get a recipe from a good cook? They often can’t tell you what they do because expertise involves unconscious competence. Regardless of the particular CTA strategy, every CTA strategy puts controls in place to address this issue.

    As you note, CTA is expensive. Ultimately, the decision to use CTA or not may involve whether the cost of collecting, representing, and applying CTA data is outweighed by the benefits of better on-the-job problem solving.

    Comment by Steve Villachica — 21 September 2011 @ 6:49 PM

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress