Cognitive agent-based modeling to optimize complex project management under uncertainty

This paper presents an agent-based simulation model with cognitive elements, developed to explore strategies for managing complex projects under uncertainty. The model was implemented using NetLogo, which provided a matrix-structured environment where agents—identified as assigners and workers—interact based on factors such as expertise level, intensity of collaboration, frequency of task rotation, and workload balance. The system was simulated over 100 time steps, allowing the identification of behavioral patterns related to organizational learning and performance under varied experimental conditions. Methodologically, the model integrates mechanisms for distributing tasks, rules for agent assignment, adjustable collaboration ranges, and decision-making based on utility functions. Cognitive aspects are addressed through decentralized decision processes and adaptive behaviors that evolve during the simulation. A total of eight configurations were analyzed, varying turnover rates, collaboration distances, and the initial competencies of the agents to assess how these variables influence the accumulation of experience and project effectiveness. The results indicate that when agents build up experience over time, project outcomes improve, following a growth curve with diminishing marginal returns. Conversely, high turnover interferes with knowledge retention and weakens learning processes. Additionally, too much collaboration can overwhelm highly connected agents, thereby decreasing the system’s overall efficiency. Statistical validation was carried out through ANOVA and Spearman’s correlation tests, confirming the consistency and relevance of the observed effects. Overall, the study underscores the importance of organizational continuity and the thoughtful design of collaboration parameters to achieve efficient performance in complex project settings. The proposed model serves as a useful framework for examining how task distribution, talent retention, and interaction patterns can be optimized, and it opens the door to future developments involving more advanced cognitive modeling techniques.