It is generally assumed that visual selection is both (1) stimulus-driven and (2) goal-driven. Very salient stimuli tend to automatically attract attention (1), but selection also depends on the relevance of the stimulus to the intentions of the observer (2). The relative contribution of these factors has been debated, with proponents of either extreme. Recent evidence suggests that saliency and relevance information do not continuously interact, but exert their effects within different time windows. Saliency information is not continuously available after stimulus onset but rapidly dissipates with time. Goal-driven information only affects performance after saliency information has gone. Interestingly, the relative roles of stimulus- and goal-driven mechanisms can be reversed: When the presentation of relevant and irrelevant information is separated in time. In the preview paradigm, participants are given a visual search task in which one set of distractors is presented first. The target, accompanied by a second set of distractors, is added later. Observers completely ignore the irrelevant first set and only search the second set. In this more dynamic situation, the goal-driven mechanism is thought to exert its effect before the stimulus-driven information. Using the preview paradigm, the project aims to further explore the dynamics of stimulus and goal-driven selection processes. By presenting stimuli at different moments in time, the paradigm is well suited for investigating the time courses of saliency and relevance information, and possible interactions. We seek to answer questions like how saliency is affected when goal-driven mechanisms are already fully operational, how saliency affects performance when it is defined across time rather than just space, and what happens to goal-driven processes when information is segmented in time, but groups across other dimensions (e.g. colour). Furthermore, neuropsychological studies will assess the functional problems in selection found after parietal brain lesions, broadening our understanding of the dynamics of underlying processes and throwing new light on the diagnosis and possibly also long-term rehabilitation. In the end, the findings are likely to provide important temporal boundary conditions for theories of attention like Bundesen s (N)TVA.