Task Analysis

What is cognitive task analysis? What is data driven knowledge elicitation and how does it relate to cognitive task analysis? Are there any tools available to help automated cognitive task analysis?
Are there any models of skill retention?
How good are models of human behavior as substitutes for missing team members?

What is cognitive task analysis? What is data driven knowledge elicitation and how does it relate to cognitive task analysis? Are there any tools available to help automate cognitive task analysis?

Cognitive task analysis identifies aspects of system design that place heavy demands on the user’s cognitive resources including memory, attention, and decision-making (Barnard & May, 2002). It is used to determine the thought processes that users follow to perform tasks at various levels, from novice to expert (Hanser, 1995). Cognitive task analysis looks at the system from the viewpoint of the user performing a specific task. The information gathered allows the designer to focus upon the system features that the user will find hardest to learn and be likely to make the most errors. Included in the analysis is examination into past critical incidents that may have occurred to shape the users feelings or expectations about the task (Schrraagen, Chipman, & Shalin, 2000). By identifying and highlighting where potential challenges could occur, designers can create a system that leaves more time for the user to perform the given task rather then struggle with using the interface (Barnard & May, 2002).

There are many different methods for conducting cognitive task analysis. At a minimum, cognitive task analysis should include the steps of mapping out the task, identifying the critical decision points, clustering, linking, and prioritizing them, and characterizing the strategies used (Klein, 1993). Cognitive task analysis relies on the technique of Data Driven Knowledge Elicitation (DDKE) to extract information about cognitive events, structures, and models. In depth interviews are used to probe into the cognitive processes of users who are performing the task. However, interviews are often very subjective, so the data collected may not present a complete and accurate representation of the cognitive processes involved. To account for this bias, controlled observation methods are also recommended. Controlled observation uses verbal protocol analysis of experts’ responses to the task when instructed to think out loud. An advantage of controlled observation is that key features of the task can be manipulated and data can be automatically collected by the system (Klein, 1993).

Recent advances in task analysis have introduced an automated way to conduct cognitive task analysis. The Decompose, Network, and Assess (DNA) program, for example, decomposes and classifies expert knowledge into organized knowledge structures in a user-friendly application. It is theory based and combines methods such as probed interview techniques and conceptual graph analysis to achieve an accurate representation of the task and the cognitive abilities of the user. DNA is the first automated cognitive task analysis application that has been developed for assessment and insight into human performance for training purposes (Regian, n.d.). DNA is especially useful in determining elements within instructional systems, and specifically in intelligent tutoring systems (Shute, Torreano, & Willis, 2000). (Picture courtesy of Cognition and Instructional Technologies Laboratory).

Are there any models of skill retention?

The natural phenomenon of forgetting is a critical training problem. If tasks are not regularly practiced, there is a high probability that they will be forgotten. In response to this problem, the Army Research Institute for Behavioral and Social Sciences (ARI) has developed a model for predicting how rapidly unpracticed tasks will be forgotten over a one year interval. This model is referred to as the ARI skill retention model (Macpherson, Patterson, & Mirrabella, 1989). The classification process requires that each task receive a rating score, which is based upon whether or not the task contains characteristics known to influence retention (i.e. memory aid, sequence, mental requirement etc). Tasks are rated on each of 10 task characteristics, and then assigned a numerical score. The sum of these scores yields the retention score, and the lower the score the more likely the task will be forgotten. The ARI Skill Retention Model can be invaluable to trainers deciding which individual tasks require the most training and when training should be scheduled (Rose, Radtke, & Shettel, 1985).

See also Measuring Skill Retention

How good are models of human behavior as substitutes for missing team members?

There have been significant improvements in the portrayal of human beings in real-time simulations. Virtual humans have been created which simulate the appearance, movement, actions, attention, and gestures of humans. Advances in engineering have allowed the construction of autonomous virtual humans, appearing to move and act independently, in addition to the creation of social virtual humans, who have the ability to interact with each other in a simulated environment. Advances in the simulation of motor skills of virtual humans include posture changes, reaching, grasping, looking, and exhibiting facial expressions. In addition, cognitive and perceptual skills are slowly being added to their capabilities, including language and attention. Virtual humans can also take on the roles and tasks of “missing team members”, meaning they can understand basic task components and function as an actual team member. Playing the role of missing team members allows other individuals to practice the task even when all team members are not present (Rickel & Lewis, 1999). The virtual human Steve, for example, can play the role of either coach (guiding the user) or missing team member. Although cutting edge software tools and applications have been developed to assist the designer in incorporating such technology into simulations, the use of virtual humans is not a perfect substitute for missing team members. More research is needed into the physiological and cognitive models of behavior to create a more realistic representation of human movements, actions, and decisions (Badler, Bindiganavale, Bourne, Allbeck, Shi, & Palmer,1997). (Image courtesy of Lycos, Inc)