This study explores group cohesion on the base of social exchange theory, focusing on the dynamics within groups as influenced by memory and learning mechanisms. Using agent-based modeling, we examine how these factors affect individual and group-level cohesion. Specifically, we model the average time agents spend in groups (duration of individual membership), the duration of group existence (aggregate cohesion), and overall agent satisfaction under three conditions: No Memory, Memory, and Memory+Learning, across two network structures: Fully Connected (FC) and Preferential Attachment (PA). Our model incorporates dynamic cultural adaptation and reciprocal exchanges to simulate realistic social dynamics. Dynamic cultural adaptation in our model refers to the evolving behaviors of agents as they adjust their individualism and preferences based on past interactions and group experiences. We find that memory of past interactions enhances group cohesion and stability, while learning introduces new dynamics that influence group persistence.