Visual selective attention (VSA) is the cognitive function that regulates ongoing processing of retinal input in order for selected representations to gain privileged access to perceptual awareness and guide behavior, facilitating analysis of currently relevant information while suppressing the less relevant input. Recent findings indicate that the deployment of VSA is shaped according to past outcomes. Targets whose selection has led to rewarding outcomes become relatively easier to select in the future, and distracters that have been ignored with higher gains are more easily discarded. Although outcomes (monetary rewards) were completely predetermined in our prior studies, participants were told that higher rewards would follow more efficient responses. In a new experiment we have eliminated the illusory link between performance and outcomes by informing subjects that rewards were randomly assigned. This trivial yet crucial manipulation led to strikingly different results. Items that were associated more frequently with higher gains became more difficult to ignore, regardless of the role (target or distracter) they played when differential rewards were delivered. Therefore, VSA is shaped by two distinct reward-related learning mechanisms: one requiring active monitoring of performance and outcome, and a second one detecting the sheer association between objects in the environment (whether attended or ignored) and the more-or-less rewarding events that accompany them.