Humans can dynamically adjust the priority of stimuli to determine which are relevant to a given objective in a specific context. Humans are also able to detect stimuli crucial for survival, even when they are unrelated to the current objective. Commonly, computational models prioritize stimuli using static thresholds, evaluating relevance according to the demands of the task. However, to achieve human-like flexibility, it is necessary to adjust stimulus priorities according to both the task at hand and the appearance of new relevant stimuli. This work presents an approach to modeling dynamic stimulus prioritization using thresholds informed by neuroscientific evidence. It draws on three key brain networks involved in detecting significant stimuli that can trigger changes in sensitivity. To validate the proposal, a case study was used to simulate an agent capable of adjusting stimulus priorities and sending alerts when dangerous events are detected, illustrating how sensitivity thresholds are adapted according to the assigned task.