State Chart Visualization of the Control Flow within an ACT-R/PM User Model Leon Urbas Technische Universität Berlin Center of Human-Machine Systems Jebensstr.1, Sekr J2-2, D-10623 Berlin ++49 (30) 314-72007 urbas@zmms.tu-berlin.de Ljudmilla Nekrasova Technische Universität Berlin Center of Human-Machine Systems lne@zmms.tu-berlin.de Sandro Leuchter Fraunhofer IITB Fraunhoferstr. 1, D-76131 Karlsruhe +49 (721) 60 91-424 sandro.leuchter@iitb.fraunhofer. de ABSTRACT We present a novel visualization for ACT-R/PM models of cognitive processes to support the model development. Because the underlying production system paradigm does not specify an explicit flow of control, it is rather difficult to grasp the structure of this kind of user models. Therefore, we developed an algorithm that analyzes the interdependencies of ACT-R/PM productions by resembling the main parts of the matching process of the production cycle. The algorithm produces a graph with nodes as specifications of the state of the declarative memory and edges as productions which are applicable in these states. States are generalized to reduce the complexity of the control flow. The graph is transformed into a state-chart like visual representation. Goal oriented behavior with sub-goaling is considered with sub- graphs. The algorithm is implemented as a plug-in for the integrated development environment eclipse. Categories and Subject Descriptors D.2.2, [Design Tools and Techniques] General Terms Algorithms, Documentation, Economics, Human Factors, Languages. Keywords Cognitive User Modeling. Visualization of Flow Control, State Chart 1. INTRODUCTION ACT-R [2-3,5-6] is a cognitive architecture and a programming environment for user models. These models describe the users’ cognitive structures and processes at a fairly atomic level. An ACT-R user model consists of a set of production rules and the specification of initial declarative memory elements organized in a semantic network. In each cycle of the production system, the condition part of all production rules is tested against the current state of the active declarative memory elements. Then one of the matching rules is selected by a conflict resolution algorithm which incorporates a network of sub-symbolic measures. Finally, the action part of the selected rule is executed to modify content and activity of declarative memory elements. ACT-R not only implements a theory of human associative memory but provides the modeler with detailed mechanisms for perception and motor action. The human mind is abstracted as a modular system. Central executive is realized as a production system core that interacts with perception, motor action, memory and other subsystems via buffers. These buffers implement laws and restrictions of data retrieval and access between the central production system and the modules. The authors of the latest version of ACT-R [3] are convinced that a mapping of some elements of the architecture to certain cortical regions like dorsolateral and ventrolateral prefrontal cortex (DLPFC, VLPFC) or basal ganglia is possible (Figure 1). The specification of the cognitive architecture imposes severe constraints on how to model user behavior. From an engineering point of view, these constraints are even supporting, because of the guidance they might give to the modeler. Therefore we think that it is more effective and efficient to use cognitive architectures than general purpose/AI languages like Prolog or cognitive toolboxes like COGENT [7]. Matching (Striatum) Selection (Pallidum) Execution (Thalamus) Productions (Basal Ganglia) Retrieval Buffer (VLPFC) Goal Buffer (DLPFC) Manual Buffer (Motor) Visual Buffer (Parietal) Declarative Module (Temporal/Hippocampus) Intentional module (not identified) Visual Module (Occipital/Parietal) Manual Module (Motor/Cerebellum) Task Enviroment Figure 1. Modular software architecture of the cognitive architecture ACT-R (adapted from [3]). Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Conference’04, Month 1–2, 2004, City, State, Country. Copyright 2004 ACM 1-58113-000-0/00/0004…$5.00. Urbas, L., Nekrasova, L. & Leuchter, S. (2005). State Chart Visualization of the Control Flow within an ACT-R/PM User Model. In A. Dix & A. Dittmar (Hrsg.) 4th International Workshop on TAsk MOdels and DIAgrams for user interface desgin TAMODIA 2005, Danzig (S. 43-48). New York: ACM. http://www.safety-critical.de/doc/tamodia2005.pdf