Modern philosophy of emotion has been largely dominated by what I call the Tracking Dogma, according to which emotions aim at tracking “core relational themes,” features of the environment that bear on our well-being (e.g. fear tracks dangers, anger tracks wrongs). The paper inquires into the empirical credentials of Strong and Weak versions of this dogma. I argue that there is currently insufficient scientific evidence in favor of the Tracking Dogma; and I show that there is a considerable weight of common knowledge against it. I conclude that most emotions are insensitive to the circumstances that might be thought to elicit them and often unfitting to the circumstances in which they arise. Taking Darwin’s lessons seriously, even predictable emotional responses to biologically basic objects (e.g. bears, heights), should not be understood as tracking abstract categories (e.g. danger). This renders most contemporary accounts of emotion implausible. We are left with two options: one may still continue to claim that emotions aim at tracking, even if they often fail; or one may abandon the Tracking Dogma in favor of a non-representational view.