The purpose of this research is to better understand the impact of cortically-induced blindness (CB) and the compensatory strategies subjects with this condition may develop on naturalistic behaviors, specifically, driving. Using a novel Virtual Reality (VR) program, the researchers will gather data on steering behavior in a variety of simulated naturalistic environments. Through the combined use of computer vision, deep learning, and gaze-contingent manipulations of the visual field, this work will test the central hypothesis that changes to visually guided steering behaviors in CB are a consequence of changes to the visual sampling and processing of task-related motion information (i.e., optic flow).
Residents of the United States or Canada
Presence of one-sided stroke or stroke-like damage to primary visual cortex or its immediate afferent white matter sustained within the specified age range of 21 - 75 years (verified by MRI and/or CT scans)
Reliable visual field defects in both eyes (homonymous defects) as measured by Humphrey or equivalent perimetry.
Willing, able, and competent to provide their own informed consent
Those who have never driven or earned a drivers' license
Past or present ocular disease interfering with visual acuity
Best corrected visual acuity (BCVA) worse than 20/40 in either eye
Sustained damage to the dorsal lateral geniculate nucleus
Parallel
Virtual Reality Driving Task
Other interventionParticipants will steer a virtual car with the goal of staying in the center of a single-lane roadway while traveling at a constant speed of 26.6 m/s (approximately 60 miles/hr). The roadway alternates between a series of straights and turns of different radii to both the left and the right. This allows for careful control of task difficulty, and for the repeated presentation of specific conditions across multiple "trials" (i.e. turns in the road) in a randomized order. In addition, the density of the visual texture elements in the virtual environment that provide optic flow (OF) signal is also varied. The low-density OF condition has no road texture or foliage, and only the solid road edges on a flat-black ground plane. The medium-density OF condition has sparse textural elements distributed on the ground plane, and the high-density OF condition has high density road texture and a canopy of road-side trees that provide texture extending far above the horizon.
Virtual Reality (VR) Lane Deviation / Offset
Virtual lane offset is a measure of driving accuracy, which involves computing the divergence (in virtual meters) from the center of the single-lane virtual roadway, relative to the position of the driver's head and averaged over the duration of each turn. The researchers will compare the variance of lane offset in virtual meters between Cortically Blind (CB) and control subjects.
Click on the lead sponsor to view all of their trials.
University of Rochester
Lead sponsor
Rochester Institute of Technology
Collaborator
