Models of visual attention have, with few exceptions, proposed that attention is deployed to unitary regions of visual space. In this classical approach, attention is conceived as a spotlight or a zoom-lens. However, several recent studies have suggested that attention is considerably more flexible that previously believed such that, in many cases, attention may be focused on multiple non-contiguous areas of the visual field without selecting the region in between. In this paper, we examine the main empirical studies that have stood out as landmarks in this opposition between unitary and divided conception of attention. We analyze some of the theoretical consequences of such a more flexible approach of attention. Finally, we suggest sonic important questions which could provide interesting empirical ways for exploring the division of attention.