Human visual perception is highly adaptive: It flexibly prioritizes information that is relevant to our current tasks and goals, while it attempts to ignore information that is not, or no longer, relevant. For this to work, the cognitive system must somehow represent what is currently relevant. This representation is often called the attentional template, as it determines what we direct our attention to. Although there is no doubt on its existence, remarkably little is known about its representational nature. For example, does it represent features, combinations of features, or more abstract properties? Is it really visual in nature, as so often assumed, or may other types of knowledge contribute or even take over? And how does this change as a function of learning? The aim of our joint research proposal is to answer the fundamental question of what kind of representation the attentional template is, in terms of function (how it affects our behaviour), physiology (how it is implemented in the brain), and time (how it is affected by selection history). To achieve this, all of the labs involved will focus on the very same research questions, but approach these from their own methodological expertise. A thorough understanding of the representational properties of attentional templates will be a major step on the way towards a neuro-cognitive theory of goal-directed perception and action, and will have important implications for many clinical and professional settings.