The effects of spatial or featural attention on the activity of neurons have been studied in many experiments that have used a variety of neurophysiological approaches. Other experiments have examined how expectations about reward are represented in neuronal activity in various brain regions. Although attention and reward are distinct concepts, I argue here that many neurophysiological experiments on attention and reward do not permit a clean dissociation between the two. This problem arises in part because reward contingencies are the only parameter manipulated in any of these experiments. I describe how attention and reward expectations have been confounded, giving rise to uncertainty about how signals related to attention and reward are distributed in the brain.