Adaptive Learning in Soar-Based Systems
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Soar-based adaptive learning is a powerful and intelligent system that can modify its decision-making strategies as needed based on its past experiences and received feedback. Loosely based on cognitive psychology findings, Soar is a rule-based cognitive architecture that incorporates invariant components of reinforcement learning, chunking, and decision-based adaptive learning, making it well suited for applications that need to adapt and improve continuously. The chapter describes how emotional adaptation learning, which adjusts agents’ actions based on emotional reinforcement, is implemented in Soar. The implementation of emotional state representations, like the mentioned happiness and frustration, allows Soar-based agents to facilitate actions that ultimately produce reward and avoid actions which may be detrimental to the agent. This will enable long-term learning since the system can adapt its decision making based on previous experiences. The architecture utilizes reinforcement learning techniques to identify the most beneficial strategies, yielding greater efficiency and robustness over time. This chapter uses case studies and simulations to show how the ability to adapt emotionally profoundly augments the ability for around the clock problem solving through AI systems. The proposed Soar-based model combines the cognitive and emotional learning mechanisms of intelligent agents, allowing them to respond dynamically to changing environments. Such conclusion lays the foundation capable of emotional intelligence for building more human-like AI systems than enhancing autonomous learning. The results highlight Soar’s applicability across a range of fields, such as robotics, decision support systems, and cognitive computing, where adaptability is essential for effective learning.
Fields of Study & Subjects
Psychiatry
Mental Health
Neuroscience
AI-based Problem Solving and Planning
Reinforcement Learning in Robotics
Publication details
- Journal / source
- CRC Press eBooks
- Pages
- 112–145
- Publication date
- 26 March 2026
- Language
- EN
- Document type
- Book chapter