Cognitive Modeling for Personalized, Adaptive Signaling for Cyber Deception

Abstract

This chapter presents an overview of the use of cognitive models for personalized adaptive signaling for cyber deception. It provides a general introduction to cognitive modeling techniques and concepts, including general goals, capabilities and limitations, cognitive architectures, and instance-based learning. It establishes that cognitive models’ reliance on generative mechanisms has predictive capabilities beyond those of purely data-driven techniques such as machine learning that can be used to evaluate the effectiveness of cyber defense techniques without requiring full implementation and test. Cognitive models can account for the entire range of human performance, including levels of expertise and individual differences. Techniques such as knowledge-tracing and model-tracing can align a specific cognitive model against an individual behavior trace, enabling personalized interventions. Because cognitive models are analytically tractable, they can guide, inform, and optimize the design of cyber deception techniques. We illustrate these concepts using an insider attacking game meant to abstract the dynamics and decision-making characteristics of real-world cyber defense. Finally, we conclude by discussing future research directions in the development and application of cognitive models to cyber deception for defense.