Gerald Westheimer

Interactions between attention, context and learning in primary visual cortex

Attention in early visual processing engages the higher order, context dependent properties of neurons. Even at the earliest stages of visual cortical processing neurons play a role in intermediate level vision - contour integration and surface segmentation. The contextual influences mediating this process may be derived from long range connections within primary visual cortex (V1). These influences are subject to perceptual learning, and are strongly modulated by visuospatial attention, which is itself a learning dependent process. The attentional influences may involve interactions between feedback and horizontal connections in V1. V1 is therefore a dynamic and active processor, subject to top-down influences.

A quantitative measure for short-term cortical plasticity in human vision

The human visual system is normally very good at determining the relative positions of objects in space, but under certain conditions contextual influences can cause significant errors in this process. We studied spatial localization around an artificial scotoma, a small mask that occludes part of the visual field while a dynamic pattern is shown over a surrounding region, and found that the ability to determine the position of short line segments was strongly biased toward the interior of the scotoma. We attribute this “shift” or misassignment of position to receptive field (RF) expansions within the artificial scotoma as seen in recent physiological studies. Furthermore, our findings show that this shift begins within 1 sec of stimulus presentation, suggesting that RFs are constantly altered by their local context and that these dynamics are a part of normal vision.

Attention and Perceptual Learning Modulate Contextual Influences on Visual Perception

Brightness discrimination thresholds and facilitation by lateral interaction were measured in five human observers and two monkeys. The subjects judged the brightness of one of four peripherally seen lines against a reference. This experiment was performed both when the observer was cued to the position of the test line (focused attention) and when there was no cue (distributed attention). Discrimination was better with focused than with distributed attention. When the test line had a collinear flank, its brightness was enhanced; this enhancement was four times more prominent with distributed than with focused attention. After training, thresholds improved and collinear facilitation decreased under distributed but not under focused attention. The findings show that there are fewer benefits from contextual interaction once attention is directed toward a visual location, and that the attentional effects are subject to training.

Diffraction theory and visual hyperacuity.

&NA; The implications of the diffraction theory of light are examined as they relate to visual acuity and hyperacuity. Two incoherent point sources of light give rise to a double‐peaked light distribution whose trough is 26% below the adjoining peaks when the sources are separated by the Rayleigh limit of resolution, but the dip has vanished when the separation is 80% of the Rayleigh limit. Diffraction theory, however, places no restriction on the precision with which any single source can be localized. No paradox is therefore involved in having a point‐spread function with half‐width one minute of arc and, at the same time, a localization threshold of a few seconds of arc, although the need for a sophisticated processing mechanism to achieve such low hyperacuity thresholds is emphasized.

Discrimination of short time intervals by the human observer.

Abstract For time intervals in the 150–1500 ms range, the difference-discrimination thresholds are about 5%. The value of this Weber fraction varies somewhat depending whether the stimulus modality is vision, hearing or touch. Thresholds are higher when a time interval signaled in one modality has to be compared with one in another, and also when two different modalities are used to delineate a single time interval, as well as when onset and offset are in the same modality but signaled to opposite cortical hemispheres. There is a prominent practice effect. This effect was used to show that there is complete transfer of training between the two visual hemispheres. These findings imply that the time-discrimination mechanism is not located at an early stage of visual processing. If there is a single central time-discrimination apparatus, the observed intermodal differences must relate to the relative ease of access to it via different modalities. The mechanism involved needs elucidating. Counting of spikes or internal time modules would seem to be too simplistic a concept; there is still a need for a process in which the duration of a just concluded presentation and an internally stored interval duration can be compared.

Gestalt theory reconfigured: Max Wertheimer's anticipation of recent developments in visual neuroscience.

In the 1920s Max Wertheimer enunciated a credo of Gestalt theory: the properties of any of the parts are governed by the structural laws of the whole. Intense efforts at the time to discover these laws had only very limited success. Psychology was in the grips of the Fechnerian tradition to seek exact relationships between the material and the mental and, because the Gestalt movement could not deliver these, it never attained a major standing among students of perception. However, as neurophysiological research into cortical processing of visual stimuli progresses the need for organizing principles is increasingly making itself felt. Concepts like contour salience and figure segregation, once the province of Gestalt psychology, are now taking on renewed significance as investigators combine neural modeling and psychophysical approaches with electrophysiological ones to characterize neural mechanisms of cognition. But it would be perilous not to take heed of some of the lessons that the history of the Gestalt movement teaches.

The distribution of preferred orientations in the peripheral visual field

Orientation discrimination was measured for line targets in the fovea and in locations along four meridians (vertical, horizontal, 45 degrees and 135 degrees ) in the peripheral visual field. In all locations, the performance for vertical and horizontal contours was on average almost twice as good as for oblique ones. In addition, especially in peripheral locations along oblique meridians, thresholds tend to be better for radially-oriented contours compared to tangential ones. These effects are robust to length of the test lines and to small changes in position and cannot therefore be due to cortical singularities (pinwheels). The results suggest that the advantage of horizontal and vertical contours, as well as possibly of radial ones, is a manifestation of a characteristic structural property of the organization of the visual pathway.

Target crowding in foveal and peripheral stereoacuity.

Disparity thresholds were obtained for a single point target when surrounded by a hexagonal array of comparison point targets. The experiment was carried out in two observers at the fovea and at retinal eccentricities of 3, 6, and 9 degrees. In each case the array diameter for best stereoacuity was determined. Care was taken that the data represented the optimum performance of the observer for the level of training in peripheral stereo tasks and also for the distance from the horopter. The results show a steady rise with eccentricity both for the stereothreshold and for the minimum target separations needed for uncrowded stereo performance. Both increase by a factor of about 10 between the fovea and the 9 degrees periphery, where a clear zone of at least 2 degrees diameter between test and comparison targets is needed for best stereoacuity.

Visual acuity and hyperacuity: resolution, localization, form.

The roles of optical resolution, of spatial localizing ability, and of form perception are assessed as they relate to visual acuity and hyperacuity and to the measurement of a patients performance in these tasks.

Center-surround antagonism in spatial vision: Retinal or cortical locus?

Mach and Hering had early advanced a model of spatial visual processing featuring an antagonistic interaction between adjoining areas in the visual field. Spatial opponency was one of the first findings when single-unit studies of the retina were begun. Not long afterwards psychophysical experiments revealed a center-surround organization closely matching that found in the mammalian retina. It hinged on the demonstration of reduction of sensitivity in a small patch of the visual field when its surround was changed from dark to bright. Because such patterns inevitably produce borders, well-known phenomena of border interaction could be seen as providing alternative explanations, whose substrate would most likely be in the visual cortex. These competing viewpoints are discussed especially as they pertain to the recent demonstration of spatial differences in the center/surround organization between the normal and affected eyes of amblyopes. To the extent that most findings favor a retinal site for the psychophysically measured antagonism, and that evidence is accumulating for a direct effect on the mammalian retina of stimulus manipulation during visual development, the difference in spatial parameters of center/surround antagonism in amblyopia suggests that the dysfunction in amblyopia begins already in the retina.