You are in: Home » History of the Unit
6. VISUAL PERCEPTION
General Notes. This material has been scanned from the original typescript. While we have done our best to remove errors, some may well remain. You can access other parts of this particular Progress Report either from the menu at the bottom of this entry, by using your browser's back function, by navigating back to the Unit History Timeline, or by accessing the relevant section of the electronic archive. Reference for this report are indexed by number and these can be found in a dedicated section also accessible from the menu at the bottom of this entry.
6.1 PHOTOSENSITIVE EPILEPSY (Project 2) (Wilkins)
(a) About 5% of patients with epilepsy are photosensitive and liable to visually-induced seizures. Epileptiform electrographic abnormalities are induced by visual stimulation with certain very specific spatial and temporal characteristics. Wilkins has pointed out that these stimulus characteristics and the topography of the EEG discharges they induce suggest (52) that the epileptic discharge is triggered in the visual cortex of the brain and may be sustained within it (52), and (55) that the cerebral hemispheres may have different convulsive thresholds (53; 384) even in patients with primary generalised epilepsy. The pharmacological evidence (reviewed by Wilkins in collaboration with Or. B.S. Meldrum, Institute of Psychiatry, University of London) indicates that in photosensitive epilepsy there exists a minimal failure of cortical inhibitory processes (arising from GABAerglc or dopaminergic mechanisms), which has no consequences under normal conditions of excitation. When normal physiological excitation exceeds a critical mass, however, the inhibitory processes are unable to prevent the discharge from spreading (233). If the normal excitation is organised in a temporal pattern the epileptic process is facilitated (54).
(b) Television induces seizures partly because of the flicker generated by the flying spot as it scans down the screen, but mainly because of the pattern of interlaced lines (55, 180). This suggests several alternatives for the prevention of seizures by optical methods (385) and has implications for the design of visual display units (380).
(c) In some patients with photosensitive epilepsy, epileptiform EEG abnormalities occasionally appear after a slow closure of the eyes. The EEG changes associated with the epileptiform abnormalities appear before the eye closure suggesting that the eye closure is not simply a mechanism for the self-induction of seizures, as has previously been supposed (342).
(d) The successive lines of printed text comprise a pattern of stripes with characteristics within the epileptogenic range. A simple mask that covers the lines of text above and below those being read attenuates the epileptogenic properties of the pattern and enables patients with photosensitive epilepsy to read without the risk of seizures. The mask is now being sold as the "Cambridge Easy Reader" (385).
(e) In patients who are not obviously photosensitive, movements of the eyes can precipitate paroxysmal EEG abnormalities and this can disrupt reading. The "Cambridge Easy Reader" facilitates the movements of the eyes across text and presumably for this reason can improve reading in patients affected in this way (385).
(f) A study in collaboration with Dr. Zifkin and Dr. Andermann at the Montreal Neurological Institute has shown that in some patients seizures may be induced by mechanisms that are exclusively cognitive (390).
6.2 EYE-STRAIN AND HEADACHE (Project 45) (Wilkins)
(a) Wilkins has shown that certain patterns of stripes are judged unpleasant to look at and give rise to illusions and complaints of tired eyes and headaches. The spatial and temporal parameters of these stimuli are so similar to those of patterns that induce seizures in patients with photosensitive epilepsy as to suggest common mechanisms of induction of the various unpleasant effects. According to a theory of visual discomfort recently proposed, the illusions are generated in the visual cortex. Certain forms of visual stimulation cause large numbers of cells to fire in a temporally-organised pattern, and if this excitation exceeds a critical mass the discharge may spread. How far it spreads depends on the inhibitory mechanisms. Normally the spread is only slight, from cells that are excited by the stimulus to others responsible for the illusions. Sometimes the discharge spreads further and triggers a seizure. The patterns induce illusions to which people who suffer frequent headaches of the migraine kind are particularly susceptible. If their headaches are confined to one side, of the head the illusions tend to be asymmetric, more on one side of the pattern than the other. If, as the theory proposes, susceptibility to the illusions is a reflection of cortical inhibitory processes, these processes may have something to do with headaches (388). The above work has relied on the facility for testing large groups of subjects and continuation of the work has not been possible since the financial restrictions were imposed.
(b) The theory, has implications not only as regards mechanisms of headaches, but also for the design and display of reading material. Some printed texts form a pattern of stripes with parameters sufficiently close to those of epileptogenic patterns as to induce electrographic abnormalities in photosensitive patients and, in others, illusions, "eye strain" and headaches. These findings have implications for the design of textual material and have suggested that the clarity of text might be improved without increasing its cost by means of the previously described simple reading aid that darkens and blurs the lines of text above and below those being read. This device is similar to those invented in the last century to control "glare". The aid reduces the likelihood not only of eye-strain and headaches (386), but also of reading epilepsy (385).
(c) Computer displays have been associated with complaints of visual discomfort, and the theory outlined above might help explain why. Television can induce seizures in patients with light-sensitive epilepsy, particularly when the screen is viewed from close quarters. Computer displays usually combine the flicker of television with the stripes of text. According to the theory, the discomfort would be reduced if the lines of text were spaced further apart, and the screen illuminated at a higher frequency (381).
(d) The illusions are difficult to quantify and so a search is underway for other physiological indices with which discomfort and its precipitants might be associated. As a result of this search there are now findings that might eventually interrelate visual discomfort and the control of eye movements. Collaborative work with Dr. John Findlay at the University of Durham has shown that the patterns that induce illusions decrease the stability with which the eyes can maintain their gaze. The reading aid reduces considerably the number of movements that the eye makes in looking from one letter to another. Evidently the flanking lines of text compromise the control of the eyes. Such control 1s affected by flicker as well as stripes. When the text is intermittently illuminated by fluorescent light or on a computer screen the eye movements are slightly enlarged and increased in number (383 U). (see Section 3.7.4)
6.3 CLINICAL ASPECTS OF VISUAL PSYCHOPHYSICS (New project) (Watt, Wilkins)
6.3.1 Amblyopia
Case studies by Watt in collaboration with Drs. R. Hess and J. Pointer of the Physiological Lab. Cambridge University of amblyopes (anisometropia and strabismic) suggest that the only difference between an amblyopic eye and a normal eye is that the former is rather more spatially uncertain than the latter. The results suggest that the amblyopic visual field is somewhat scrambled.
Amblyopia affects about 1 in 7 of the population and varies considerably in degree. It is proposed to devise clinical test charts both to test the hypothesis more widely and to help evaluate treatment.
The work is supported in part by the Wellcome Foundation.
6.3.2 The grey-level stage in human vision
The initial stages of human vision effect a transformation from a continuous grey-level image to a symbolic representation recording features such as lines and edges along with certain of their properties. An account of how this is achieved has been worked out in detail by Watt in collaboration with Professor Morgan of University College London. This account has the surprising bonus that a whole range of grouping phenomena can be predicted because the output is a structured representation. The grouping may well be of practical importance for tasks like reading, as it appears to have considerable implications for visual processing of stimuli even when they are only 2 arc degrees from the fovea (344).
6.3.3 2D - Shape Representation in Vision
The next step up from the grey-level problem is the stage at which the shape and disposition of edge loci is represented. Watt argues that typically, 2D shape discrimination reflects high precision within a structured framework (343). The "edge-map" is segmented at points of contour termination (343; 347 U), points of contour intersection (345), and sharp angles (346 U). The resultant segments are thus, smooth (twice differentiable) and continuous, and are represented by "bump" primitive shapes (343; 350 U).
6.3.4 Clinical tests of contrast sensitivity
Prints of gratings with low contrast have been designed by Wilkins for the rapid "forced choice" examination of contrast sensitivity in the clinical setting. The gratings are cheap to reproduce and provide a test that is quick to administer. The test has been used in collaborative studies with Dr Sergio Delia Sala and others at the University of Milan. The studies have shown that a high proportion of patients with diabetes have impaired contrast sensitivity, regardless of retinopathy (88). Clement Clarke International have expressed an interest in publishing both the low contrast gratings and also another version of the test, currently under development, that uses low contrast letters.
6.4. VISUAL ATTENTION, MASKING AND CONSCIOUS AWARENESS (Project 42) (Duncan, Gathercole, Marcel)
While many different cognitive systems are dealt with under the general heading of "attention" (94), this section is concerned with the problem of limited capacity processes in vision. There is a strict limit to the amount of information that can be taken up from the visual field at any one time.
6.4.1 Directing attention.
Work on the problem of directing attention to the most important stimuli has led Duncan to the following general model (90). To be reported, any visual stimulus most gain access to a "limited capacity system" (LCS) which deals efficiently with only one stimulus at a time. This applies to the report of all stimulus properties: shape, size, brightness, colour, etc. LCS is preceded by a stage of perceptual analysis (the preselective system) which operates in parallel across simultaneous stimuli, and which serves to direct chosen stimli to LCS. To this end, all stimuli in the preselective system are interviewed for properties defining them to be important, and only stimuli possessing such properties pass on to LCS. For example, if the task is to report red letters in a display of letters in different colours, the colour of each stimulus is interviewed in the preselective system, and only red stimuli pass on to LCS.
An interesting implication of Duncan's model is that, in a sense, stimulus properties are known at the preselective stage, but cannot be reported without access to LCS. For example; colour can guide access to LCS, but cannot be reported until such access has occurred. This has been confirmed for both simple stimulus features such as tilt, and complex properties such as shape (95). It has often been claimed that simple stimulus features can be perceived "without attention", but thls claim is apparently wrong.
More generally, it appears that one should not speak of a single stage at which some aspect of a stimulus is "identified" (95). Stimulus Information is put to many different uses in the nervous system (directing "attention", underlying action and report, producing perceptual grouping, etc). Showing that information is used for one purpose, at a particular stage, does not imply it can be used for others. (The debate over "unconscious recognition" is a special case.)
Work has continued on the problem of which stimulus properties can be used to guide LCS access. While it still appears that quite complex categorizations (eg digit vs letter) can be employed (91), some results previously ascribed to this process seem in fact to reflect physical differences between categories (92). A new project is dealing with variations in the efficiency of selection using a particular property such as colour (95). Selection of target stimuli for entry into LCS is imperfect. In accounting for this, it may be important to take into account factors such as the interaction between preselective processes and later processes of choosing a speeded response.
Theories of why we are limited in our ability to deal with several stimuli at once fall into three broad classes. Object-based theories propose that LCS deals with only one perceptual object at a time. Discrlmination-based theories propose a limit on the number of discriminations that can be made. Space-based theories propose a limit on the spatial area from which information can be taken up.
Duncan's object-based theory is at least partly correct (93). Given a brief display, subjects can report two aspects of one object as accurately as they can report either on its own. They cannot, however, report aspects of two different objects.
A further question is whether there really is only one cause for limited capacity phenomena in vision. For example, we might be limited both in the number of objects we can see, and in the number of similar discriminations (e.g. shape discriminations) we can make. It is, of course, vital to know whether the phenomena of visual attention in fact reflect limits of several distinct sorts, work is in progress on this question.
6.4.2 The fate of unattended visual material
One technique frequently employed to assess the fate of unattended visual information 1s to look at whether responses to a target located in a multi-element display are influenced by the identity of the nontargets. Findings that the semantic relationship between target and nontargets falls to have any effect on target performance are typically interpreted as reflecting no detection of the nontargets.
Results obtained by Gathercole in a selective attention task in which nontargets are repeated after a delay, however, suggest that nontargets are registered even when they originally had no detectable effect on target performance. Future experimentation will explore this issue further, with the aim of discovering how the visual attention system handles unwanted information.
6.4.3 The influence of vision on touch
Marcel has investigated and extended an ignored illusion which bears on sensory integration. Subjects judge whether an index finger held in front of the face is touched by one or two points. Subjects focus on and monitor a visual display some distance beyond, which produces a double image of the finger. This significantly induces one touch to be felt as two. If, in addition, a filter is worn over one eye, delaying the visual signal from that retina, one touch is sometimes felt as two touches at different points in time. Conversely, stereoscopic fusion of two index fingers may result in two touches, one to each finger, being felt as a single touch. It is suggested that phenomenal sensations are experienced via the body image, which is constructed, and dominated by vision (227 U).
Other sections in the 1981-1984 report
2. COGNITIVE PSYCHOLOGY
3. COGNITIVE ERGONOMICS/APPLIED COGNITIVE PSYCHOLOGY
4. HEARING
5. MOTOR SKILL AND ACTION
6. VISUAL PERCEPTION
7. PSYCH0PHYSIOL0GY SECTION