To cope with the heterogeneous nature of predation and the trade-off between predator avoidance and foraging, prey animals have evolved several cognitive rules. One of these is the risk allocation hypothesis, which predicts that in environments with long periods of sustained high risk, individuals should decrease their antipredator effort to satisfy their metabolic requirements. The neophobia hypothesis, in turn, predicts increased avoidance of novel cues in high-risk habitats. Despite the recent interest in predator-induced neophobia across different sensory channels, tests of such generalized neophobia are restricted to a single fish taxon, the Cichlidae. Hence, we re-tested the generalized neophobia hypothesis in fathead minnows Pimephales promelas, a small schooling North American cyprinid fish. From hatching onwards, minnows were exposed to conspecific alarm cues, which indicate predation risk, or distilled water in a split-clutch design. After one month, shoaling behavior was examined prior and subsequent to a mechanical predator disturbance. Fish previously exposed to elevated background risk formed compact shoals for a shorter time interval following the stimulus compared to controls. These results contrast previous studies of generalized neophobia, but match the risk allocation hypothesis. Consequently, risk allocation and generalized neophobia are not ubiquitous cognitive rules but instead evolved adaptations of different taxa to their respective environments.