Achieving efficient and natural human-robot interaction requires robots to dynamically adapt their behavior to align with each user’s specific interaction style. This work presents an adaptive framework based on a comfort-driven architecture that enables social robots to adjust their behavior in response to individual user interaction styles, while maintaining their internal goals. The system leverages a modular architecture integrating perception, motivation, and reasoning components that allow the robot to identify the user’s preferred interaction modality and refine its behavior by regulating less-used modalities to ensure more balanced and effective communication. A simulation study demonstrated that the adaptive system allows the robot to maintain higher comfort levels and respond more efficiently to users with consistent interaction patterns, while adapting gradually to more variable profiles. By adapting its behavior, the proposed system aims to foster a more engaging and enjoyable interaction for both parties.