Cybersickness (CS) is a pressing issue in virtual reality (VR) systems. While various mitigation methods (MMs) have been developed to counteract CS, their effects on human behavior remain largely unexplored, raising concerns about their potential applications. Using Jens Rasmussen's skill--rule--knowledge (SRK) model as a framework, our study investigated the effects of two widely adopted MMs---dynamic field of view and dynamic blurring---in VR. We compared these methods to a baseline condition where no MM was applied. We designed three VR tasks that align with the behavioral levels of the SRK model. In a within-subject study (N = 22), participants completed each task using these MMs. We measured task performance, CS symptoms, and locomotion control. Additionally, qualitative feedback was collected. Our results revealed that neither MM significantly alleviated CS across different VR scenarios. Furthermore, while some participants found MMs helpful, a larger portion reported visual hindrances, and a significant performance drop was measured in the skill-based task. More critically, participants indicated behavioral adaptations in response to the MMs, including changes in locomotion strategies and viewing behavior. Potential causes and implications were discussed. In conclusion, MMs offer promise, but their application necessitates a nuanced understanding of their impacts. We recommend a context-sensitive approach when designing and integrating MMs, prioritizing both maximizing CS mitigation and minimizing interference with the natural behaviors of users.