Because most plants rely on animals for pollination, insights from animal sensory ecology and behavior are essential for understanding the evolution of flowers. In this review, we compare and contrast three main types of pollinator responses to floral signals--receiver bias, 'adaptive' innate preferences, and associative learning--and discuss how they can shape selection on floral signals. We show that pollinator-mediated selection on floral signals can be strong and that the molecular bases of floral signal variation are often surprisingly simple. These new empirical and conceptual insights into pollinator-mediated evolution provide a framework for understanding patterns of both convergent (pollination syndromes) and advergent (floral mimicry) floral signal evolution.