Sequential resolution of fragmented visual percepts: experimental investigation of a subject's perceptual experience after a right medial temporal stroke

This report concerns the fragmented visual percepts in a woman, TR, following a right entorhinal-perirhinal infarct. In a previous report, Weddell [Weddell, R. A. (2005). A visual disorder producing highly selective deletion of recurring letters. Cortex, 41, 471-485] linked TR's highly selective tendency to delete recurrent letters with her fragmented percepts. The conflation of same-identity form elements was attributed to anterior extrastriate damage, which reduced the amount of information sustainable in fully resolved visual percepts, and the present experimental investigation of her subjective account of segment formation and resolution completes the story. She said that complex objects and long words first appeared as blurred regions, which sometimes included form elements. It is argued that figure-centred attentional mechanisms subdivided this blurred region into up to 3-4 parts. String length, lexical status (word vs. nonword), and background colour and/or luminance determined fragment length. Two rules described the fragment resolution sequence: largest segments usually resolved first, left-to-right resolution accounting for a few sequences. This resolution sequence occurred when stimuli were exposed too briefly for saccadic exploration, implicating endogenous attentional shifts. Experiment 4 confirmed TR's assertion that spatial, orthographic, and phonological information were stored during the fragment resolution process. Moreover, TR exerted considerable voluntarily control over the fragment resolution sequence and some influence over fragment length. Finally, these findings were interpreted in terms of an extended version of a neural network model of vision largely derived from nonhuman primate studies.