Memory and Learning Mechanisms

View at publisher Summarise with VIRAT
Authors: Kuldeep Singh Kaswan, Jagjit Singh Dhatterwal, Anand Nayyar
Cited by: 0

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

The main focus of this chapter is on developing an approach for learning and memory mechanisms as considered in cognitive architectures that allows intelligent systems to be adaptable, allow for optimization and make rational decisions. Episodic Memory and Chunking Integration in Soar: A General Intelligence Cognitive Architecture Emphasizing the system’s ability to learn about its experiences through episodic memory aids in storing and recalling past experiences for translating the context into decisions made, and chunking encodes learned action sequences into production rules for procedural learning of skills in repeated problem-solving contexts. Combined, these mechanisms allow Soar to dynamically learn from experience, manage computational resources more effectively, and improve performance over time. This chapter explores the architecture that coordinates episodic memory retrieval with chunking, striking an optimal tradeoff between memory efficiency and plastic learning in the service of memory efficiency and flexible learning. It also analyses issues like retrieval optimization, chunk pruning, and the tradeoff between specificity and generality in auxiliary knowledge. From its contribution to artificial intelligence and robotics to decision support systems and adaptive learning environments, we see how cognitive architectures can inform our understanding of human-like reasoning and learning in practical applications. Subsequent developments could integrate deep learning and reinforcement learning methods to fine-tune memory management and decision optimization even more. Drawing from long-term experimental findings, this chapter illustrates that the memory and learning mechanisms of Soar can be leveraged to move toward human-like problem-solving to create scalable, more efficient, and more intelligent AI systems.

Fields of Study & Subjects

AI-based Problem Solving and Planning Psychiatry Mental Health Neuroscience Cognitive Computing and Networks

Publication details

Journal / source
CRC Press eBooks
Pages
88–111
Publication date
26 March 2026
Language
EN
Document type
Book chapter