Helene Intraub, Ph.D. - US grants

A useful approach to studying the processes that underlie visual cognition is to study the errors that people make. By examining patterns of errors and the conditions that predictably cause them, it is possible to infer the principles that guide visual perception and memory. Two lines of research will test the effects of viewers' general knowledge about the structure of naturalistic scenes on their ability to perceive and remember pictures. One series of experiments will examine a perceptual illusion that occurs during high-speed presentation of pictures (e.g., 9/sec), in which an object from one picture is frequently misperceived as having occurred in the immediately preceding or following display in the sequence. The relationship of the object to the display in which it actually appears, and to those that surround it, will be manipulated such that the object either will or will not conform to the observer's knowledge of real-world scenes. The research will evaluate the effect of such manipulations on frequency of misperception. The other series of experiments will test a memory distortion that occurs when viewers try to remember close-up photographs of objects. Observers tend to recollect seeing information that was not present in the photograph, but is likely to have existed just outside the camera's point of view. This suggests that general scene knowledge is important in understanding pictorial representation and that it alters memory in a predictable and measurable way. This basic research on picture comprehension and memory has implications for visual communications (e.g., video displays, control panel design), particulary under conditions in which errors and brief delays may be crucial to avoid.

DESCRIPTION (Applicant's Abstract): The proposed research tests and extends a model of visual scene processing and spatial representation. The model takes into account the fact that input to the visual/cognitive system is comprised of partial views of a continuous visual world. It is proposed that spatial views activate a mental schema that extrapolates beyond the depicted view, allowing it to be understood within a larger context, and thus to help support the experience of a continuous visual world. This process affects memory, however, in that observers tend to remember having seen scene structure that was never presented but had only been inferred through schema activation alone. In one such distortion, called boundary extension, the observer remembers having seen a greater expanse of a scene than had been presented. The model raises specific predictions about the likelihood of various types of spatial distortions under a wide range of conditions. In addition to broadening the understanding of spatial representation, the proposed research has implications for individuals with brain damage whose visual/spatial capacities are compromised. The research is divided into seven sections: a) objects vs. scenes; b) scene "decomposition;" c) imagination; d) memory for spatial relations within a picture; e) eye movements and the perceptual schema; f) tactile spatial representations; and g) changes in pictorial representations over time. The procedures involve presentation of visual or tactile depictions of scenes followed by tests of spatial recall and recognition. To test recall, subjects' drawings will be digitized and areas and distances within their drawing will be measured using a computer graphics system. Recognition tests will include distractors and rating scales that will be sensitive to subtle differences between the test item and the viewer's internal representation.

DESCRIPTION (provided by applicant): The proposed research tests a theory of visual/spatial representation that addresses the fact that at any moment in time, sensory input delivers only a partial view of the continuous world. In visual scanning, the eyes sample the environment as rapidly as 3-4 times per second, and in haptic exploration (without vision) hands successively examine relatively small regions over time. It is proposed that a single limited "view" of a scene activates representation of anticipated surrounding space. This anticipatory projection serves to facilitate integration of successive inputs, but also has important implications for memory. Observers remember having seen (or touched) parts of a scene that were anticipated, but never actually experienced (e.g. a phenomenon called "boundary extension".) In addition to broadening our understanding of spatial representation, the proposed research has implications for individuals with brain damage (e.g., hemi-neglect), and those with sensory deficits. The proposed research will address: a) the effects of planning an eye movement on the ensuing spatial representation of a scene; in particular addressing the issue of whether boundary extension will increase or decrease in the about-to-be-fixated region b) the similarities and differences in spatial representation following visual Vs. haptic (while blindfolded) exploration by sighted observers, and in some cases, those with sensory deficits, c) the accuracy of cross-modal transfer of spatial information, and c) how the relations among objects influence accuracy in memory of spatial layout. Procedures include visual and/or haptic (without vision) exploration of real scenes, or visual exploration of digitized photographs in conjunction with eye tracking techniques. Memory tests will involve recall and recognition procedures as well as new interactive memory tests involving the reconstruction of a view in real space or on a computer screen.