NEURAL CORRELATES OF OBJECT-BASED WARPING

Abstract

Object-based warping is a visual illusion in which space within an object appears expanded when points are located within the object and compressed when points are located on or just beyond the boundaries of an object. Other studies have used functional imaging to investigate the neural mechanisms of visual illusions, but this has yet to be done for object-based warping. In the current study, we tested feasibility of measuring positions of small stimuli (N = 5), a prerequisite for studying object-based warping using functional imaging; then, we delineated visual regions in subjects (N = 3) using probabilistic atlases and conducted multivoxel pattern analysis (MVPA) to determine which of these visual regions may hold representations of object-based warping based on classifier errors in position estimation. We trained classifiers on activity when a dot stimulus is isolated, when the dot is on a perceived figure region, and when the dot is on a perceived ground region. The classifier estimated locations above chance for nearly all visual regions in all subjects and tests, with greater accuracy for early than late visual regions. However, accuracy drops dramatically on runs with figure-ground displays, and early visual regions did not consistently exhibit systematic errors in estimates that would demonstrate expansion and compression effects involved in object-based warping. Our results do not support our hypothesis that early visual regions hold representations of position that are subject to object-based warping. We have demonstrated that this approach to studying object-based warping is ineffective, perhaps due to a limited amount of data and the use of figure-ground stimuli that are too high contrast and complex. Future work is needed to more precisely investigate the neural mechanisms of object-based warping.