Jans, Peters, and De Weerd (2010) examined the studies demonstrating that spatial attention can be split across 2 noncontiguous target locations. They find all these studies to be flawed and conclude that spatial attention only selects a single location at any given time. They do, however, suggest that there could be exceptional circumstances that allow for split attention. As Jans et al. point out, there are credible alternative explanations for many of the split-attention results, and it is necessary to establish high methodological standards for this question. However, a close examination of the criteria that they propose for split attention shows that they are unnecessarily constraining, making it virtually impossible for any experiment to meet all the criteria. Jans et al. also apply their criteria inconsistently and do not hold their unified attention account to the same standard that they demand for split-attention accounts. When a more even-handed standard is adopted, the split attention account emerges as a viable explanation and perhaps the most parsimonious explanation for some of the experimental results. Jans et al. do not consider visual tasks, such as extracting the gist from a complex scene, that suggest that in some circumstances complex visual processing is possible while multiple locations are selected. We propose a new theoretical framework, based on a minimal filtering principle, to explain how the attentional system adjusts between situations in which multiple locations can be selected simultaneously and situations in which only a single location must be selected.