“…Saliency, as a composite measure of low-level visual information, is positively correlated with the presence of objects (Elazary & Itti, 2008), and found to guide attention in tasks where targets are underspecified (e.g., memorization, Underwood & Foulsham, 2006), and also utilized during situated language production (Gleitman et al, 2007). In our view, the most appropriate way to define visual saliency is in terms of visual clutter, calculated by integrating low-level visual information, e.g., color, with edge information (Rosenholtz, Mansfield, & Jin, 2005). The inclusion of edges sets apart clutter from standard visual saliency (Itti & Koch, 2000), which is based solely on low-level features, without taking object-based visual information such as edges into account.…”