Floral communities present complex and shifting resource landscapes for flower-foraging animals. Strong similarities among the floral displays of different plant species, paired with high variability in reward distributions across time and space, can weaken correlations between floral signals and reward status. As a result, it should be difficult for foragers to discriminate between rewarding and rewardless flowers. Building on signal detection theory in behavioural ecology, we use hypothetical probability density functions to examine graphically how plant signals pose challenges to forager decision-making. We argue that foraging costs associated with incorrect acceptance of rewardless flowers and incorrect rejection of rewarding ones interact with community-level reward availability to determine the extent to which rewardless and rewarding species should overlap in flowering time. We discuss the evolutionary consequences of these phenomena from both the forager and the plant perspectives. This article is part of the theme issue 'Signal detection theory in recognition systems: from evolving models to experimental tests'.