James E. Cutting - US grants

The aim of this project is to identify the means by which human observers organize moving structures and perceive them as dynamic wholes. The particular moving structures used will be computer-generated displays of dots that, in their movement, mimic human walkers, rolling wheels, and other everyday events. Previous research has shown that these displays yield robust percepts, despite their relative impoverishment of seen elements. And the dearth of elements involved allows simple and concise mathematical description. This project will proceed on three fronts. The first will study perceptual organization with a particular focus on how much time is needed to organize a dynamic stimulus. Brief displays of systems of dots will be displayed for varying amounts of time in order to assess the durations necessary for the identification of familiar movements of near-familiar objects. The second will study perceptual organization with a focus on the effectiveness of various kinds of camouflage in inhibiting identification of common objects in motion. Static and dynamic camouflages of several types will be employed, with a particular emphasis on the movement parameters of a dynamic camouflage that are most effective in interfering with identification. The third will study perception organization with a focus on the information available about movement in iconic memory. Previous research has shown that such information is available, and the nature of the coded form of that information will be assessed. The idea is that if information about many parameters of movement is available as early as the icon, then the perceiver is clearly attuned to that information and appears not to need to elaborate it in a computational fashion. The disciplines involved in this study are, primarily, psychology and visual perception, and, secondarily, artificial intelligence and computer simulation. The health relatedness of this project is to be measured in terms of understanding the normal function of the human visual/perceptual system, both in situations of natural complexity and in reduced complexity such as when viewing elements on radar screens, and when watching for pedestrians crossing the streets at night.

People typically find their way through woods and buildings without harm or injury. That is, while traveling at reasonable speed, we can determine our heading and avoid obstacles without doing damage to them or to ourselves. Constraints of physics and human reaction time dictate that we must be able to determine our heading to within one degree of visual angle, or about half the width of one's thumb head at arm's length. Otherwise, we risk injury in our daily movements. Perceptual scientists generally agree that visual information provides the usual means for locomotor guidance and that this information lies in the relative motions of stationary objects around a moving observer. There is disagreement, however, on how best to characterize these relative motions to determine exactly what information is used. This research will test a theory positing a recursive two-step process. According to the theory, people first search out objects to look at and follow them (using pursuit eye movements) while they are moving forward and second monitor the relative motions of objects nearer and farther than the one they are looking at, but in its vicinity. In general, nearer objects move faster than, and in the opposite direction from, farther objects and near objects move in a direction opposite to one's heading. Thus, to be able to determine one's heading, one simply gazes at an object anywhere in the field of view and observes the direction of the fastest movements around the point of gaze. If these move right, the heading is usually to the left; if these move left, the heading is usually to the right. This research will test the theory in a number of ways, using computer- generated displays, which will simulate various types of environments and various types of observer movements through them. In particular, the research will explore conditions in which people make errors in determining their heading and the reasons for those errors. An understanding of these errors will have implications for air-traffic safety, highway safety, and the safety of visually-impaired pedestrians.

Human beings generally find their way through environments with ease and without injury. During a normal walk, a pedestrian must judge direction of movement with an accuracy of about 4 degrees of visual angle (about the width of two thumbs held at arms' length). Otherwise he or she risks injury; a driver or pilot risks worse, and must attain considerably greater accuracy. Previously it has been shown that people can find their way based on information in the global pattern of movement at the back of the eye, called the retinal flow. More particularly, as one moves through the environment, an individual's natural mode of looking entails a cycle of pursuit fixations and saccades. That is, an observer looks at a given object (say to the left of his or her path) while moving forward on a linear or curvilinear path. The pursuit fixation entails a smooth counterclockwise eye rotation during locomotion. He or she then saccades rightward to a new object, entailing an abrupt clockwise eye movement. This cycle generally repeats, either again on the left side or reversing to the right. Several sources of information are available in the retinal flow during pursuit fixation. The two most important are differential motion parallax (DMP), the relative motion around a fixated object where near objects move faster than, and in the opposite direction from, far objects, and inward motion (IM), the motion of objects toward the fovea. These studies will continue this basic line of research, investigating further the utility of DMP and IM, and exploring new sources of information, including several entailed in a constant gaze-movement angle which may be useful in detecting collisions. In particular, the experiments will determine how moving observers negotiate moving objects in a rigid environment. The research should result in a deeper theoretical understanding of the limits and nature of human beings' wayfinding ability, with a long-term aim of promoting traffic safety. Stimuli will be presented to individual viewers on a computer screen and will be manipulated in ways that obey the mechanics of the real world and the biomechanics of human locomotion.