Guarding Against Malicious Biased Threats (GAMBiT) Experiment 3

The GAMBiT (Guarding Against Malicious Biased Threats) Experiment 3 dataset captures attacker behavior in a high-fidelity cyber range designed to study the impact of cognitive biases on adversarial decision-making. The study involved 22 red team participants conducting self-paced cyberattacks over two days within a simulated enterprise network built using the SimSpace Cyber Force Platform. The environment featured 40 virtual devices and 25 embedded \u201ctriggers\u201d\u2014deceptive cues designed to elicit cognitive biases such as confirmation bias, sunk cost fallacy, base rate neglect, availability heuristic, and loss aversion. Participants\u2019 behaviors were monitored through network telemetry, keystrokes, terminal histories, and psychometric assessments, including reasoning and affect measures. The dataset provides detailed behavioral logs and supports hypothesis-driven analysis of attacker susceptibility to cognitive manipulation. It offers valuable insights for developing adaptive cyber defenses that exploit predictable biases in human adversaries, reinforcing the potential of cognitive modeling in cybersecurity strategy.