Random image structure evolution (RISE)

Abstract

Images of different objects, such as those used in conventional studies of object perception, can be thought of as isolated points drawn from a high-dimensional object-image space. Behavioral and neural measures corresponding to these points do not assess the variation in response associated with the variation in image attributes about these points. We have developed a technique called RISE (Random Image Structure Evolution), by which one may systematically sample continuous paths in a high-dimensional object space. A basic RISE sequence consists of a progressive transformation of an object's image into a random field, along with the reverse sequence depicting the evolution of the original image from randomness. The processing steps are designed to ensure that low-level image attributes such as the frequency spectrum and luminance are held constant throughout a RISE sequence. Experiments based on the RISE paradigm can be used to address some key issues in object perception. In the simplest RISE experiment, we measure the onset and offset of object recognition as subjects view images evolving from, then dissolving into, randomness. Shifts in the positions of the onset and offset markers across different conditions or populations serve as quantifiable and objectively verifiable measures of various perceptual phenomena. These include perceptual development, learning, and visual deficits. In our experiments so far to validate the RISE paradigm, we have used marker shifts as new indices of object priming. Doing so obviates the need for precise reaction-time measurements or tachistoscopic presentations. Additionally, we are using the perceptual hysteresis evident in the relative positions of the onset and offset markers as a signature to establish links between behavioral reports and neural responses. Javid Sadr is a Howard Hughes Medical Institute Pre-Doctoral Fellow.