Mary Catington¹, Michael Pratte¹; ¹Mississippi State University

In any particular moment only a subset of incoming visual information can be perceived due to limitations in attentional processing. Recent research has shown that low-level visual information about two objects, such as their color and luminance, can be perceived in parallel. However, high-level information such as word meaning is limited by a serial processing bottleneck, such that only one of two simultaneously presented words can be recognized. Although this limitation could be due to constraints on high-level aspects of word recognition, here we explore the possibility that it is driven by attentional limitations on the mid-level visual processes that make word identification possible. One such mid-level visual process is figure-ground segmentation, by which common local elements are grouped together to form a coherent surface. This is one of several mid-level processes known to rely on attention, and is crucial for boundary ownership and object identification. In Experiment 1 participants attempted to identify coherent textures in noise for one item, or for two items presented simultaneously. The results reveal a cost when attempting to process two items simultaneously such that, unlike color or luminance, texture segmentation could not be performed across multiple objects in parallel. However, unlike word identification, segmentation was not purely serial such that some information about two objects could be processed at the same time. Therefore, the inability to simultaneously read two words observed in previous studies may partially stem from attentional constraints in the mid-level visual processes that are necessary prerequisites for word recognition. There are several such mid-level processes, and future work will examine how limitations across them may give rise to the severe serial bottleneck in word recognition.

Acknowledgements: This work was supported by National Institutes of Health NIMH Grant #R15MH113075-02A1.