What would a person do when exposed to some new computer interface,
cockpit control panel, or navigation problem? If we can answer
these questions before actual systems were built, the potential to
reduce training time, and decrease operational mistakes and their
associated costs is enormous.
Through User Modeling, we seek to answer these questions directly, allowing the use of "what if" simulations to quickly and inexpensively try out new scenarios. This is, in fact, one of our primary R&D activities: Capturing the behavior of a particular individual or group of individuals in the form of a running computer program (for more on the methods involved, see the Cognitive Modeling page).
User Modeling can start with a dataset specifying human behavior, or it can start from a first-principles cognitive engineering specification of expected behavior. In either case, the target of User Modeling is often some task that we have no direct way to collect human performance data from. For example, it may be a road intersection that hasn't been built, or a cockpit instrument that doesn't exist yet. The goal in both of these cases is to predict human performance in a new domain, rather than just describe behavior on some existing task or interface.
Obviously, no prediction of expected human behavior is perfect, but we can provide an analysis of existing data sets and a prediction of behavior in some alternative situation constrained by the available human data and the decision processes people are known to use. If you have a project in mind, contact us to set up an appointment to discuss it.