A Cognitive Approach to Cyber Defense and Building and Testing a Prototype Can Make the Internet a Safer Place

Cyberattacks on digital infrastructure are increasing rapidly, making it essential to understand how financial incentives, technical limitations, and environmental information shape the decisions of attackers and defenders. This thesis examined these aspects through a series of controlled behavioral experiments combined with computational cognitive modeling. Across three financial incentive–based studies, it was consistently observed that monetary rewards influenced players’ decision strategies in non-intuitive ways. Instead of encouraging active engagement, financial rewards led human participants to reduce both attack and defense actions, suggesting that incentives may create a risk-averse mindset in cyber game environments. Conversely, financial penalties for analysts—particularly penalties for misses and false alarms—had a substantial impact on strategy selection, pushing analysts to become more vigilant but also causing greater deviation from the ideal Nash equilibrium. When participants faced real human opponents rather than algorithmic Nash players, these motivational effects intensified, demonstrating that human-human interaction is a major driver of strategic variability. The Instance-Based Learning (IBL) cognitive models developed in this research successfully reproduced these human patterns, underscoring the importance of cognitive factors—such as recency of feedback, frequency of outcomes, and memory activation—in shaping cyber decision-making under uncertainty.