This study explores the complexities of animal decision-making in T-maze environments by integrating computational modeling with machine learning techniques. Specifically, it focuses on the binary decision-making process in T-mazes, examining how animals choose between two paths. Our research employs a mathematical model designed to capture the decision-making behavior of fish, providing analytical insights into their intricate behavioral patterns. Additionally, advanced machine learning algorithms are utilized to analyze behavioral data from zebrafish and rats. The combination of these techniques yields high predictive accuracies, demonstrating the effectiveness of computational methods in decoding animal behavior in controlled experiments. This study enhances our understanding of animal cognitive processes and underscores the crucial role of computational modeling and machine learning in advancing behavioral science.