Recent research on anxiety has increasingly focused on physiological markers and behavioral responses, often overlooking the subjective experience of anxiety itself. This is problematic, as anxiety is fundamentally a subjective emotional experience. To deepen our understanding of anxiety, it is essential to study the specific conditions that are most strongly linked to how individuals personally experience anxiety. To address this gap, we conducted an experiment using virtual reality (VR) to simulate a foraging task under threat of predation. Participants navigated virtual environments while facing various threats, and, after each trial rated their subjective experience of anxiety as well as other emotions. This setup allowed us to manipulate environmental variables and track how participants' actions, the immediate context, and the outcomes of those actions influenced their subjective emotional state. Using factor analysis and multilevel modeling, we analyzed the relative contributions of these different factors—environmental conditions, participant behavior, and outcomes—to the subjective experience of anxiety. Our results offer new insights into the dynamics of anxiety, revealing how different elements of a threatening situation interact to shape how anxiety is felt and perceived. This work highlights the importance of studying predictors of anxiety under naturalistic conditions, where participants are free to move without many constraints. By doing so, we can develop more comprehensive models of anxiety that better reflect its complex, real-world manifestations, ultimately improving both clinical interventions and theoretical understanding.