ACT-CV: Bridging the Gap between Cognitive Models and the Outer World

Cognitive modeling can provide important insights about the usability of software or hard-ware products at all stages of product design. It is not used very often, though, as the cre-ation of cognitive user models is typically done by highly trained experts. These specialists need not only develop the models, but have to spend a lot of additional time on software interface problems: The cognitive models need to be connected to the interactive system under evaluation. This poster presents ACT-CV, a software library that aims at solving the interface problems of user modeling. The Challenge: Modeling interactive Behavior Cognitive modeling has proven more and more useful for the design and evaluation of hu-man machine interfaces as it allows to predict usability metrics like the time to complete a task or the types of errors that may occur. Modeling relies heavily opon cognitive architec-tures like ACT-R (Anderson, 2007) which closely mimic the cognitive, perceptual, and motor capabilities of human users. Unfortunately, cognitive architectures tend to lock the modeler into highly specialized computer systems (Lisp in the case of ACT-R), creating significant hurdles for the application of a user model to the human machine interface in question (Byrne, 2013). Existing tools to overcome these hurdles either revert to mock-up prototypes of the interface in question (e.g. CogTool, John et al., 2004), require changes to the source code of the original software (e.g. AGIMap, Urbas et al., 2006), or search the computer screen for special pixel color combinations (Segman, Ritter et al., 2006)