Abstract
When looking out at a scene, we can flexibly direct our attention to individual objects (e.g., a specific tree), or to the whole scene (e.g., a forest). Here we examine how the learning of statistical regularities prioritizes individual objects in the array (local attention) or the entire array (global attention). In Experiment 1, we examined whether local regularities draw attention to a local scale. Observers viewed arrays of nine colored objects arranged in a 3x3 matrix. Each matrix was either in the shape of a square or a diamond. Each individual object was either a square or a diamond. Unbeknownst to the observers, the matrix either contained three triplets of colored objects (i.e., local regularities) in the structured condition, or contained colored objects in a random arrangement in the random condition. The task was to indicate, as fast as possible, either the shape of the individual object, or the global shape. We found that observers were reliably faster at identifying individual objects but slower in identifying the global shape, when the array was structured vs. random. This suggests that local regularities facilitate local attention and impede global processing. In Experiment 2, we examined whether global regularities cue global attention. Everything was the same as in Expt1 except that there were no triplets. Instead, the four corners of the matrix contained a color quadruple (i.e., global regularities) in the structured condition. We found that observers were reliably faster at identifying the global shape but slower in identifying individual objects, in the structured vs. random condition. This suggests that global regularities facilitate global attention and impede local processing. These findings demonstrate that spatial regularities can determine whether attention is directed to individual elements or to the entire scene, providing evidence for the influence of statistical learning on the spatial scale of attention.
Meeting abstract presented at VSS 2014