AbstractsPsychology

Where a reader's eyes fixate within a word influences how accurately and quickly a word is recognized. Mapping performance against fixation within a word gives the “viewing position (VP) curve,” which has an inverted-U shape with better recognition at word beginning than end and best recognition at a fixation slightly left-of-center (optimal viewing position, OVP). There are data supporting each of the major, non-mutually exclusive accounts of these VP asymmetries: informativeness, cerebral laterality, and perceptual learning. However, none of them can explain why reverse VP asymmetries have not been observed. It is my thesis that these issues are in large part because stimulus and reader characteristics both are critical in visual word recognition. We addressed the issues by using a target discrimination task in a VP paradigm for a variety of more-or-less word-like strings. Behavioral results show (1) all strings yield an U-shaped VP curve; (2) all but strings with no letter-like features show a beginning-vs-end VP asymmetry, ruling out an account based wholly on either informativeness or left cerebral specialization for word processing; (3) only words exhibit a left-to-center OVP, suggesting cerebral laterality and/or informativeness may play a role; (4) perceptual learning also falls short of explaining the deleterious effect on performance of the number of characters in the target's VF (i.e., visual crowding). Using event-related potentials, our investigation on the neural mechanisms of the VP effects find (1) the continuity of fovea to parafoveal processing assumption, required for the cerebral laterality account, however, interacts with VF, and (2) an electrophysiological index of the OVP effect in the early sensory P1 component. Compared to prior research, we use a broader range of stimulus types, at more locations in visual space, in a target discrimination task, thereby affording a not only finer-grained analysis of the VP curve but also filling significant empirical gaps. For the first time, we dissociate the beginning-vs-end and OVP asymmetries, and further uncover an important role for an unexamined factor: visual crowding. Moreover, by filling these empirical gaps, we provide additional constraints on any empirically adequate account and conclude that no current alternative is wholly satisfactory.