High-Level ACT-R Modeling based on SGT Task Models


Today’s most important constraint on practical application of cognitive modeling as evaluation method is the high cost/low benefit ratio caused by a lack of support tools for modeling and by high requirements on sophisticated knowledge in cognitive psychology as well as artificial intelligence programming. The development of high-level languages to model human cognition based on low-level cognitive architecture is a current matter in the cognitive research community (see Ritter et al., 2006). To open cognitive modeling to a wider user group and make the developing task easier and more accessible, the Hierarchical Task Analysis Mapper approach (HTAmap) has been developed. The main idea behind HTAmap is modeling on a higher level of abstraction. The foundation of the HTAmap approach is determined by: (1) HTAmap modeling process: A formalized modeling process to minimize the transformation-gap between semi-formal, high-level sub-goal template task models (SGT; Ormerod & Shepherd, 2004) and formal, low-level ACT-R models (Anderson et al., 2004). (2) HTAmap modeling language: An XML-based representation of task- and device-related knowledge within an integrated ACT-R model as well as strategies for model reuse and adaptation with regard to dynamic task environments and user skills. (3) HTAmap editor: A Java-based software tool for systematic and semi-automated ACT-R modeling based on predefined model fragments. The HTAmap modeling approach delivers cognitive modeling functionality based on predefined and modifiable cognitive activity patterns (CAP). Options for integrating separately defined device models of complex dynamic task environments (ACT-R graphical interface, AGI) as well as interfaces for embedded perception and action-models (AGI strategies) are also possible.

Back to Table of Contents