Abstract: In contemporary analytic epistemology, emotions are usually not considered to be a good justification for an agent’s belief, in the same way perception, memory, reasoning or testimony are, due to emotions facing issues when it comes to being reliable and sensitive to defeaters. In this talk, I argue that this problem can be overcome if one reframes what is expected of emotions under rational guidance. Building on Jones (2003), I will propose that we need to consider emotions as participating to knowledge acquisition under a conception of epistemic agents as reason trackers rather than reason responders. This will enable me to build a model where the reliability and sensitivity of emotions can be explained through delayed calibration rather than immediate control. In order to strengthen this proposal, I will highlight how perception is following the same pattern, while benefiting from a better epistemic reputation than emotions. Finally, I will discuss how the calibration of emotions is a dual process, where calibration is influenced by both internal and social monitoring, which explains why our emotions can offer good justification for belief.