Personalized Model-Driven Interventions for Decisions From Experience

Area of Science:

• Cognitive Science
• Artificial Intelligence
• Cybersecurity

Background:

• Individual differences in knowledge influence behavior.
• Instance-based learning theory (IBLT) models decision-making from experience.
• Cognitive architectures integrate learning and decision processes.

Purpose of the Study:

• To demonstrate an instance-based learning (IBL) cognitive model for predicting individual behavior in dynamic environments.
• To account for population averages and individual variances in decision-making.
• To develop personalized signaling algorithms for cybersecurity defense.

Main Methods:

• Implementation of an IBL cognitive model within the Adaptive Control of Thought-Rational (ACT-R) architecture.
• Utilizing recurrence quantification analyses to examine sequential trial-to-trial behavior.
• Applying model-tracing and knowledge-tracing for real-time individual alignment.

Main Results:

• The IBL model, with identical parameters, generated diverse human behaviors via stochastic memory retrieval.
• Personalized modeling successfully aligned with individual decision-making in a cybersecurity task.
• Cognitive model introspection revealed salient features influencing individual choices.

Conclusions:

• The combined techniques offer a blueprint for personalized cognitive modeling.
• This adaptive, personalized approach has implications for cybersecurity defense and intelligent systems.
• Tailoring intelligent artifacts to individual differences is crucial for domains like human-machine teaming.