Learning plays a crucial role in predator avoidance [1-3], but little is known about how the type of experience with predators molds future prey behavior. Specifically, is predator-avoidance learning and memory retention disrupted by cryptic coloration of predators, such as crab spiders [4, 5]? How does experience with different predators affect foraging decisions? We evaluated these questions by exposing foraging bumblebees to controlled predation risk from predators (robotic crab spiders) that were either cryptic or highly contrasting, as assessed by a quantitative model of bee color perception . Our results from 3D tracking software reveal a speed-accuracy tradeoff : Bees slow their inspection flights after learning that there is a risk from cryptic spiders. The adjustment of inspection effort results in accurate predator detection, leveling out predation risk at the expense of foraging time. Overnight-retention tests reveal no decline in performance, but bees that had experienced cryptic predators are more prone to "false alarms" (rejection of foraging opportunities on safe flowers) than those that had experienced conspicuous predators. Therefore, bees in the cryptic-spider treatment made a functional decision to trade off reduced foraging efficiency via increased inspection times and false-alarm rates against higher potential fitness loss from being injured or eaten.