Oxyopias Archive 1998

Oxyopias Schedule

December 4, 4:00 PM, 489 Minor Hall

Suzanne McKee
Smith-Kettlewell Eye Research Institute, San Francisco, CA
Host: Cliff Schor

Visual Acuity and Binocularity in Amblyopic Patients

Claudio Galletti
University of Bolonga, Bolonga Italy
Host Elwin Marg

Area V6a: a Bridge from Vision to Movement

Eric Hornstein
University of California, School of Optometry-Vision Science
Host Ted Cohn

Sex differences in spatio-temporal properties of housefly (Musca) photoreceptors

Yair Weiss
EECS, University of California at Berkeley
Host: Jitendra Malik

Psychophysics and Modeling of Human Motion Perception

Gerald Westheimer
Division of Neurobiology, University of California at Berkeley
Host: Stan Klein

Position and Line Orientation as Distinguishable Primitives in SpatialVision

Bill Prinzmetal
Psychology Department, University of California at Berkeley
Co-sponsored with the Institute for Cognitive Studies
Host: Steve Palmer

Visual Illusions, the Mystery Spot, and the Anti-Gravity Chair

Abstract

Geometric illusions, such as the Ponzo, Zollner, and Wündt-Hering illusions, demonstrate that simple line drawings can effectively distort the perceived length, orientation, or curvature of line segments. Despite over a century of analysis, there are still questions about the perceptual mechanisms that underlie these illusions.

I will present a new theory that accounts for many of the classic visual illusions. The theory, called the Orientation Framing Theory (OFT), developed with Art Shimamura, claims that many illusions can be attributed to the effect of local visual cues to our sense of orientation perception (i.e., the perception of vertical and horizontal).

The inspiration for this theory comes from the Mystery Spot in Santa Cruz, and other anti-gravity houses. In these locations, one' sense of vertical and horizontal is distorted by the visual environment. I will demonstrate how the same process can account for the tilt-induction effect, the Zöllner illusion, the Poggendorff illusion, the cafe wall illusion, the Ponzo illusion, and many other phenomena.

I will present two lines of evidence for the theory. First, we took the Ponzo illusion as a test case and conducted a number of experiments comparing the OFT account of this illusion with theories based on low spatial frequency filtering (Ginsburg), linear perspective (Gregory) and other factors.
The second line of evidence used the anti-gravity chair. It has been found that the effect of visual context on orientation perception is increased by having observers sit in a tilted chair. (Because sitting in such a chair increases the effect of visual as opposed to gravity-based cues to orientation, I call it the anti-gravity chair). The anti-gravity chair increased the magnitude of those illusion explained by the OFT, but did not affect a control illusion.

David Merwine, PhD
Research Associate
Smith-Kettlewell Eye Research Institute, San Francisco, CA
Host: Izumi Ohzawa

Complementary pathways supporting retinal directional selectivity

Abstract

Computation of directional selectivity in the rabbit retina. Two separate asymmetries combine to produce the robustly directional output of ON-OFF directionally selective ganglion cells.

Bruce Bridgeman
Psychology Department, University of California, Santa Cruz
Host Lawrence Stark

Perception and Oculomotor Behavior in a Patient Without Efference Copy

Abstract

A German neurological patient, RW, was the subject of a paper in Nature (Haarlemmer et al.) earlier this year. He has bilateral damage to nonstriate visual cortex, diagnosed with MRI scans, and suffers from intrusive Fihlene illusions (the world seems to sweep in the opposite direction when he tracks a moving object). In our Munich lab RW did not see an afterimage jump when he made a saccade with closed eyes, evidence that oculomotor efference copy does not affect his perception. But he does not experience jumps of the visual world during normal saccades, evidence consistent with a new 'reference object' theory of space constancy (Deubel et al., 1998). The theory holds that efference copy is not used in re-establishing visual calibration after a saccade if the saccade goal or another reference object is found within a spatiotemporal window at the start of the new fixation. RW shows very high thresholds for saccadic suppression of image displacement with a forced-choice measure. But he can point to targets in the dark with or without fixating them, and can point straight ahead even with eccentric fixation, indicating that his efference copy may be available at a motor level but not in perception.

Eli Peli
Schepens Eye Research Inst
Harvard Medical SchoolBoston MA
Host: Stan Klein

Prism treatment for hemianopia

Lawrence Stark
School of Optometry, Vision Science, University of California, Berkeley
Host: Cliff Schor

Locating the Scanpath Cognitive Model

David Forsyth
Computer Science, University of California at Berkeley
Co sponsored by Institute for Cognitive Studies
Host: Steve Palmer

Finding people and animals in large collections of images

Abstract

Segmentation and recognition tend to be seen as separate activities in theories of human and of machine vision. Building programs that can handle interesting applications of object recognition - for example, recovering pictures that contain particular objects from poorly structured collections - requires object representations that can represent objects at a reasonable level of abstraction and can help segment objects from the image. Ideally, we should be able to learn these representations from images. I have built a number of programs that, illustrate the issues to deal with. One can tell, quite accurately, whether an image contains a naked or scantily-clad person; another can tell whether an image contains a horse or not. These programs have been tested on large collections of pictures of diverse content. The representations can be learned from image data. Each program uses a bottom up process only. I will describe our current work on using MCMC methods as a source of top-down information for grouping human limbs and image regions that look like clothing, and speculate on implications for theories of human vision.

Ruth Rosenholtz
Xerox PARC; Co-sponsored with the Institute for Cognitive Studies
Host: Jitendra Malik

Visual Search Among Heterogeneous Distractors

Abstract

A model for visual search based upon signal detection theory (SDT) has been shown to predict a number of results. However, aside from conjunction and disjunction search tasks, this model has largely been tested on search for a target among homogeneous distractors. We will present experimental evidence and simulation results for an orientation search task, which suggest that the basic signal detection theory model cannot fully explain search for a target among heterogeneous distractors. Further simulation results suggest a model in which: saliency of a display element is measured as in a parametric test for outliers; greater saliency relative to that of the other elements in the display implies improved search efficiency; and performance with equally-salient elements is modeled by SDT. Finally, we will show that this model qualitatively explains a number of classic motion search asymmetries.

April 15 Wednesday, 12:00 noon, 489 Minor Hall

Akiyuki Anzai
UC Berkeley, Vision Science Group
Host Ralph Freeman

Nonlinear analysis of binocular neurons in the primary visual cortex

Abstract

Nonlinear properties of neurons provide important clues as to what kind of computations they perform. In this talk, I will describe spatio-temporal nonlinear interactions exhibited by binocular neurons in the cat's striate cortex. First, responses of neurons due to nonlinear interactions between a pair of bars, one in each eye, are examined. The binocular interactions are found to be multiplicative, and are consistent with the computation of the interocular cross-correlation of images in a local region of space. Since an interocular cross-correlation is one of the algorithms for helping solve the stereo correspondence problem, these neurons may perform initial computations necessary to solve the problem. Second, responses due to nonlinear interactions between a pair of bars separated in space (either monocularly or dichoptically) and time are examined. Complex cells exhibit responses associated with monocular as well as dichoptic motion energy of the stimulus. This could explain perceptual phenomena such as dichoptic apparent motion and an apparent depth change due to an interocular stimulus delay (Pulfrich effect). These results suggest that neurons in the striate cortex perform nonlinear computations that are fundamental for the analysis of motion and depth.

Jochin Braun
Computational and Neural Systems, Division of Biology, California Institute of Technology, Pasadena, CA
Host: Jack Gallant

Visual attention and cortical circuits

Abstract

A good way to study visual attention is to confront observers with concurrent visual tasks. Such experiments quantify the attentional cost of different visual tasks, and show that many demanding tasks are performed well with little or no attention. This vision without attention seems to reflect relatively early levels of processing. Further experiments of this type reveal how attention changes the tuning properties and contrast response function of visual filters. These observations are summarized by a computational model of spatial vision, and of attentional changes in spatial vision, which maps quite well onto the connectivity of a hypercolumn in striate cortex.

Michael Tarr
Host: Steve Palmer

Is face recognition 'special'? or Understanding normal object recognition by testing its limits. Evidence from brain injury, fMRI, and behavior

Abstract

Three sources of evidence have been cited for face-specific processing in human object recognition. First, there are neuropsychological case studies where brain-injured subjects appear to be disproportionately impaired at face recognition as compared to the recognition of common non-face objects (prosopagnosia). Second, functional brain imaging studies (fMRI and PET) appear to show a dedicated neural substrate -- a portion of inferior temporal lobe (IT; part of the fusiform and inferior temporal gyri) -- in humans that is more active when viewing faces as compared to when viewing common non-face objects. Third, psychophysical studies have revealed a range of putatively face-specific behavioral effects -- in particular, extreme sensitivity to the configural aspects of stimuli across a variety of recognition tasks.

It is our contention, however, that face recognition should be considered a particular case of within-category/item-specific recognition, in which members of a visually homogeneous class must be distinguished by experts. Using this framework we tested claims regarding face-exclusive processing for all three sources of evidence. First, across multiple experiments we found that prosopagnosic subjects are disproportionally sensitive to increasingly specific levels of categorization with both familiar and novel non-face objects (rather than faces _per se_). Second, in an fMRI study we found that the same region of IT associated with faces is engaged by the subordinate-level recognition of familiar non-face objects. Third, in behavioral studies we found that experts, but not novices, with a novel class of non-face objects (Greebles) showed configural sensitivity similar to that obtained with faces. Finally, we combined the latter two techniques and found individual subjects' face areas were active during Greeble processing, but only following expertise training. Together, these results suggest that the level of visual categorization and the degree of expertise, not just stimulus class membership, may be important mediating factors in dissociations generally found between face and object recognition in the human ventral pathway. At a more general level, these results suggest that models of visual recognition cannot simply assume a restricted domain of explanation -- the human visual recognition system is highly flexible and theories must be able to account for a wide range of recognition behaviors. At a minimum, our results provide additional controls for proposed of face-specific mechanisms and neural substrates.

Ken Nakayama
Harvard University
Irvin Rock Memorial Lecture; co-sponsored with the Institute for Cognitive Studies
Host: Steve Plamer

Erik Viirre
Research Scientist, Human Interface Technology Laboratory, University of Washington
Host: Cliff Schor

Vision Technologies and Vision Research at the Human Interface Technology Laboratory

Pietro Perona
California Institute of Technology, Pasadena
Host: Jitendra Malik

Shape, Reflectance and Pop-out

Julian S. Joseph
Department of Psychology, University of Nevada
Host Stan Klein

The Architecture of Attentive and Preattentive Vision

March 10, noon, room 3201 Tolman Hall

Izumi Ohzawa
UC Berkeley

Roles of complex cells in the primary visual cortex

Dana Ballard
Host: Jack Gallant

Natural computation

Mike Mustari
University of Texas
Host: Cliff Schor

The role of binocular visual experience in oculomotor system development

Abstract

Humans or simians experiencing Disruption of binocular pattern vision early in life, subsequently develop significant deficits in their oculomotor gaze holding systems. Prominent among these deficits is apermanent horizontal nystagmus, resembling latent nystagmus (LN),revealed during monocular viewing. Subjects also have an asymmetric optokinetic nystagmus (OKN), with little or no nasal-temporal response,under monocular viewing conditions. Both the horizontal OKN deficitsand the presence of horizontal LN suggest involvement of the pretectal nucleus of the optic tract (NOT) in these deficits. This is because the NOT is essential for generating slow phase eye movements in the responseto horizontal full-field visual motion. The goal of our studies was to determine whether the NOT plays a critical role in LN. Therefore, we recorded NOT unit activity during LN, smooth pursuit and during visual receptive field testing in two trained animals (> 2 yr old), which lacked binocular pattern visual experience for the first 25 or 55 days of life. Recordings from the NOT of animals reared with normal binocular visual experience served as controls (Mustari and Fuchs,1990). Eye movements were measured with an electromagnetic method employing scleral search coils. Animals were trained to track a moving target spot with either smooth pursuit or saccadic eye movements, or to fixate a stationary target spot during visual receptive field testing.In addition, we evaluated the effects of bilateral pharmacological blockade of the NOT on LN.

We found that LN was initiated at short latency, averaging (average 75ms), when a shutter was closed in front of one eye. Slow phase eye velocity reached at least 40 deg/s within a few seconds after the onset of monocular viewing. The short LN onset times we observed are consistent with direct activation of appropriate optokinetic pathways possibly involving the NOT. The direction of LN slow phases depends upon which eye is viewing. NOT units of normally reared and BD monkeys are directionally tuned. Most NOT units responded preferentially to movement of a large-field visual stimulus toward the side of recording.However, we did find an increased percentage of units in BD monkeys that responded preferentially during contraversive visual motion.Interestingly, such contraversive units were strongly modulated duringLN per se. The visual dependency of this LN related response was shownby briefly turning the visual stimulus off during LN; during this time unit response ceased or fell to the resting level. NOT units in BD moneys encode visual stimulus velocity in a manner similar to that observed in normally reared monkeys. In normally reared monkeys, NOT units are binocularly driven, responding well to visual stimuli presented to either eye. In contrast, NOT units in BD animals were either driven exclusively by the contralateral eye or were contralateral eye dominated. This contralateral eye dominance is potentially important for eliciting LN. For example, during monocular viewing through the left eye, the right NOT will be activated preferentially, thus producing rightward (slow phase) nystagmus. Consistent with this interpretation, we found that LN is abolished following reversible inactivation of the NOT with the potent GABA agonist, muscimol. We useda single injections (200-400 nanoliter) of 2% muscimol delivered intothe NOT in these experiments. Horizontal optokinetic eye movements were also virtually abolished following such bilateral injections. Insummary, the NOT plays an important role in the etiology of LN. We performed further experiments where vision was alternately restricted to one or the other eye for the first six months of life. For this testing fitted contact lens were alternated on a daily basis, commencing within 24 hours of birth. We expected that such monocular visual experience would produce profound LN and a monocular NOT. Interestingly, we found neither result in the first two subjects tested. This special form of restricted visual experience will be discussed in terms of likely developmental mechanisms important for oculomotor and visual system development.

David Williams
Center for Visual Science, University of Rochester
Host: Stan Klein

Adaptive Optics for the Human Eye

Ben Backus
UC Berkeley
Host Marty Banks

Use of vertical disparity and extra-retinal eye position cues during stereopsis

Abstract

Human observers are able to determine the orientation of a surface stereoscopically, despite the fact that horizontal disparity does not contain sufficient information. How do they do this? Two additional sources of information are vertical disparity and felt eye position (e.g., from efference copy). We (in the Banks lab) manipulated these sources of information independently, and found that the visual system uses both sources of information. Furthermore, the visual system dynamically re-weights these two sources of information depending on their relative reliability under current viewing conditions.

Simon Rushton
Research Psychologist
Hewlett-Packard Labs, Bristol, Bristol UK
Host: Marty Banks

The neglect of perceived location - locomotion on foot does not rely on optic or retinal flow

Abstract

Movement through an environment produces an optic flow field. Thedirection of locomotor heading is specified by invariants (patterns) within the flow-field. Since Gibson, research on locomotor heading has been concerned with identifying flow field invariants and mechanisms for picking them up.

Provoked by the veering trajectory taken by a patient with unilateral visual neglect, we explored the role of perceived location in the guidance of locomotion in normal observers. We used displacing prism glasses to manipulate perceived location and recorded walking trajectories.

The results suggest, in contradiction to a body of previous research, that perceived location, rather than optic or retinal flow, guides locomotion on foot.