Cognitive Modeling of Agents: Integrating Emotions, Goals, Needs, and Decision-Making

Traditional crowd simulations in complex environments like train stations often simplify human behavior by focusing solely on physical movement and neglecting psychological depth. This paper introduces a cognitive agent framework that integrates dynamic emotional states (e.g., valence, frustration) and physiological needs (thirst, hunger etc.) to model decision-making more realistically. Agents operate via a dual-mode architecture: during surplus time, they strategically pursue secondary goals using a utility-based mechanism that balances need intensity, spatial costs, and environmental opportunities; when needs exceed critical thresholds, they reactively prioritize urgent demands (e.g., finding a restroom). The framework also incorporates personalized factors (age, mobility, luggage) and agents’ evolving knowledge of Points of Interest (POIs), enabling them to reason about unknown POIs and anticipate need fulfillment on trains. Implemented in a simulated train station environment, the model demonstrates how agents generate context-sensitive, heterogeneous behaviors such as interrupting travel plans for urgent needs or dynamically rerouting driven by internal state fluctuations. Results show that this approach captures a richness in decision-making absent in conventional rule-based simulations, offering improved realism for applications in crowd management and spatial design.