This document discusses cognitive architectures for physical agents. It provides an overview and background on cognitive architectures, examples of some existing architectures like Soar, ACT-R, PRODIGY and ICARUS, the key capabilities they aim to enable like perception, decision making, problem solving and learning. It also discusses important properties of cognitive architectures related to knowledge representation and utilization. The document outlines criteria for evaluating architectures and some open issues remaining around capabilities like episodic memory, natural language, emotions and enhanced learning abilities.
2. Overview
• Background & Motivation
• Example Cognitive Architectures
• Capabilities
• Properties
• Evaluation Criteria
• Open Issues
3. Background & Motivation
• What is a cognitive architecture?
• Why create a cognitive architecture?
• Emerging architectural classes
– Psychological phenomena
– Representation, organization, utilization, and
acquisition of knowledge
• Entering the commercial sector
• The need for integrated systems
4. Example Cognitive Architectures
• Soar (Laird, Newell, & Rosenbloom, 1987; Newell, 1990)
• ACT-R (Anderson & Lebiere, 1998; Anderson et al., 2005)
• PRODIGY (Carbonell, Knoblock, & Minton, 1990)
• ICARUS (Langley, Cummings, & Shapiro, 2004)
• The 3T Architecture (Bonasso et al., 1997)
5. Capabilities
• What is a well-defined architecture?
– Recognition & Categorization
– Decision Making & Choice
– Perception & Situation Assessment
– Prediction & Monitoring
– Problem Solving & Planning
– Reasoning & Belief Maintenance
– Execution & Action
– Interaction & Communication
– Remembering, Reflection, & Learning