Research Lecture Advance Programme 2002/03

Vision Science and its Clinical Applications

All rooms with the CM prefix are in the Tait Building, City University, Northampton Square.

For enquiries please contact:

Abstracts

Morphology and Physiology of Primate M and P Cells

New and Old World anthropoids differ in their cone photopigments. Evolution of this difference is uncertain. Post-receptoral mechanisms must have co-evolved with the receptors to provide colour vision, and so it is important to compare post-receptoral processing in these two primate groups, both from anatomical and physiological perspectives.

The morphology of ganglion cells and other neurons has been studied in the retinae of Cebus, a diurnal platyrrhine, and of the nocturnal Aotus, which has a single cone pigment. Electrophysiological recordings were made in retinal ganglion cells of Cebus and Aotus. Animals phenotypes were molecularly confirmed

Diurnal platyrrhines, both di- and trichromats, have cell classes very similar to those found in catharrhines: M (parasol), P (midget), and small-field bistratified ganglion cells, diffuse and midget bipolar cells, and H1 and H2 horizontal cells. In fovea, P cell dendritic trees contact single midget bipolars, which contact single cones. The Aotus retina has far fewer cones than diurnal species, but M and P ganglion cells are similar to those in diurnal primates although of larger size. Small bistratified ganglion cells and H2 horizontal cells are rare or absent in the Aotus. Recordings from Cebus showed very similar properties to the catharrhine macaque, except that P cells of dichromatic animals lacked cone opponency. Recordings from Aotus showed cell response dominated by rod signals even at higher luminance levels.

These results give support to the Mollon-Jordan hypothesis of two distinct pathways for color vision.

1/f channel re-weighting predicts many aspects of lightness perception

The perceived lightness of a surface remains consistent under huge variations in illumination. To achieve such remarkable lightness constancy the visual system must take one effect the amount of light landing on the retina and disentangle the contribution of two causes: the reflectance of surfaces and how intensely they are illuminated. To solve such an under-constrained problem requires that one make assumptions. Such assumptions could be based on high-level knowledge of objects, but here we describe a low-level scheme which is predicated on the statistics of natural scenes. Specifically, we pose lightness perception as a reconstruction problem where the visual system infers the image most likely to have caused the particular response of a bank of spatial-frequency tuned filters. It does this based on two assumptions. First that images exhibit scale-invariance: the responses of filters at different scales are related by a simple linear relationship on log-log axes, (i.e. 1/f statistics). Second, that illumination tends to vary slowly over scenes (compared to image features, such as edges), so that its effect may be minimised simply by switching out a subset of the coarse-scale filters (local gain control). We show that imposition of these two properties on images can (a) adequately reconstruct natural scenes, (b) reduce the disruptive effect of gross luminance variation on image features and (c) account for the presence and magnitude of a wide range of lightness illusions.

Non-invasive measurement of short wavelength ganglion cell loss

There is much speculation about selective loss of short wavelength driven ganglion cells with age and in diseases such as glaucoma. Different theories explain the blue-yellow defects observed in a variety of eye conditions but is there really a selective loss? We have developed techniques to measure the density of SWS driven ganglion cells relative to other cell types in vivo and will present initial findings.

High Resolution Imaging of the Human Retina using a Fourier Deconvolution Technique

The use of a Fourier deconvolution technique, based on multiple wavefront aberration data and degraded images to provide an ensemble maximum likelihood estimate of the original object, is a novel technique for imaging in the eye. The technique will be introduced, and illustrated, from its astronomical origins. A high resolution retinal imaging camera, that uses a Shack-Hartmann wavefront sensor and a Fourier deconvolution imaging technique will be presented. High resolution retinal images of the human cone mosaic, resolved using the Fourier technique, for a retinal patch approximately 10 minutes of arc in diameter from two different retinal locations will be shown and discussed in detail. The Fourier resolved images are compared with anatomical images of the cone mosaic from the same retinal locations.

Electrophysiology of the S-cone pathway in early glaucoma

The blue sensation in human colour vision is mediated primarily by the blue cone or S-cone photo-receptor. S-cones constitute less than 0.5% of the photoreceptors, their spectral sensitivity overlaps with the remainder, and is affected by individual lenticular absorption characteristics. A stimulation technique to avoid these problems enabled us to record electrical responses from the S-cone pathway at retinal and cortical level using surface electrodes. Blue sensitivity is reduced in glaucoma and we have applied these techniques to measure the reduction objectively. Statistically significant attenuation of signal amplitude was demonstrated in groups of patients with primary open angle glaucoma and ocular hypertension, but not in 9 subjects with normotensive glaucoma when compared with an age matched group of normal controls. Pattern electro-retinograms were attenuated in all three groups. The off-response of the visual evoked cortical potential from the same S-cone stimulus was also shown to be significantly attenuated in primary open angle glaucoma, compared to age matched controls. The significance of these findings is discussed in relation to neural damage.

Bursts, edges and junctions in the LGN and area V1 of the awake primate

What aspects of the visual scene are sampled by the visual system? Current thinking supposes that edges are sampled by the early visual system, and that these edges are then combined cortically, in hierarchical stages, to form junctions. Thus edges are considered the most fundamental visual feature. Here we reexamine this idea by using eye-position corrected reverse correlation methods in freely viewing awake monkeys as they view illusions in which corners seem brighter than edges, even though both features belong to the same surface and have equivalent physical luminance. Recordings in area V-1 reveal that corners evoke stronger neural responses than edges do, in correlation to perception. Recordings in the lateral geniculate nucleus show that subcortical visual neurons also respond better to junctions than to edges, suggesting that junctions, and not edges, are the most fundamental visual feature. These results also suggest that the visual system does not perceive junctions solely as a function of cortical mechanisms that combine edges in hierarchical stages. These results represent a major change in the way we think about the early visual system, and have far reaching implications for theoretical models of early visual processing.

Motion Vision: the brain, a model, and the real world

Because animals live and act in a 3D-world full of moving objects their retinal images never remain still this makes visual motion processing one of the most important functions of sensory systems. So it is not surprising that motion perception has a long tradition as a research topic, starting from classical observations of the waterfall illusion, and leading to highly sophisticated psychophysical experiments on the analysis of complex motion distributions such as present in optic flow or motion transparency. Starting from such examples of how the brain extracts meaningful motion information, a simple and biologically plausible model of motion processing (the 2DMD model) is developed, which is utilised as research tool to study the distributions of local motion signals that are available to the brain as input for higher-level processing stages. The use of this model is demonstrated for the analysis of motion processing two typical situations:

1. investigating the effect of small involuntary eye movements when looking at simple black and white patterns such as used in Op-Art paintings, and
2. looking at the patterns of image shifts that are experienced when an observer moves through a natural environment.

Optical and electrophysiological correlates of olfactory and visual perception

I have examined the aspects of the visual scene and the neural responses that correlate with visibility. I have also asked which parts of the brain must be activated for the stimulus to become visible. My results show that spatiotemporal edges generate the strongest neural responses (bursts) and percepts. These results are confirmed by their ability to predict several previously unknown visual illusions. The activity in the early parts of the visual system, furthermore, does not always correlate with perception, suggesting that higher cortical areas may be necessary for visual awareness of visibility. The optical methods used to study these effects rely on signals related to changes in blood flow. The exact nature of the blood flow dynamics underlying optical signals, however, is unknown. I will discuss my work in the olfactory system of rodents to examine the underlying physiology of the vascular system that generates optical and BOLD signals.

Adaptation, chance, and constraint in the evolution of color vision: bees as a model.

We commonly assume that the sensory systems of animals, including their colour vision, are precisely tuned to the environment. Yet there are many cases where such tuning is not obvious. Here I present evidence from the world of bees, to explain when we might expect evolutionary tuning and when not. Bees are an excellent model system to study the evolution of colour vision, since the most biologically relevant coloured targets - flowers - are comparatively easily pinpointed. It is shown that evolutionary chance processes may have a more important role than previously thought, even, suprisingly, in promoting adaptation.

The representation of the visual world in the primary visual cortex of the common marmoset revealed by optical imaging.

Retinotopy, the fact that nearby cells in the visual cortex encode nearby positions in visual space, is perhaps the most fundamental characteristic of the visual system. Even species such as rodents, that seems to possess little in the way of organisation for orientation, possess a rough retinotopic organisation of space within their primary visual cortex.

Recent studies have shown that optical imaging of intrinsic tissue properties can be used to measure local retinotopy in a variety of species and in particular can be used to analyse anisotropies in cortical magnification factor (CMF) as noted in earlier electrophysiological and anatomical studies.

We examined the retinotopic mapping of the visual world in the primary visual cortex of the marmoset monkey using differential optical imaging. Two sets of complementary stripe-like locations were visually stimulated in turn. Their difference depicts the cortical representations of continuous bands of visual space. By rotating the sets of stripe-like locations it is possible to map different spatial axes. Analogous to the macaque we found that the V1/V2 border represented the vertical meridian while horizontal, 45, and 135 degree angled stripes of space were also represented in a continuous manner. We developed a new automatic method of calculating local measures of cortical magnification from our optical retinotopic maps. Using this method we found no evidence of any local anisotropies in cortical representation. Overall our results indicate that space is mapped isotropically in the primary visual cortex of the common marmoset.

A bird's eye view of the vertebrate near response

Shifting focus from a far to near object involves a change in the curvature of the intraocular lens via constriction of the ciliary muscle. In primates, this accommodation response is coupled to a convergence of the eyes (to eliminate disparity) and pupillary constriction (to increase depth of focus). The neural circuitry underlying the different components of this synkinesis, or near response, has been studied in a number of vertebrate species, including monkey, cat and bird. A comparative perspective on current findings suggests that general organisational principles can be identified. In all vertebrates, accommodation and pupilloconstriction are controlled by parasympathetic preganglionic motoneurons within the nucleus of Edinger-Westphal [EW] or its equivalent. Function-specific EW cells project to either accommodative or pupillary postganglionic cells within the ciliary ganglion that in turn innervate the ciliary muscle or iris sphincter. Using the transsynaptic tracer, WGA, injections of the iris and ciliary muscle were made to identify the pupillary and accommodative cells within EW. Initiated in the pigeon and the chick, these studies have been extended to the cat and various species of monkey. The results indicate that the preganglionic cells are usually in close proximity to the oculomotor complex and appear to be organised into spatially segregated subdivisions of EW. Quantitative analysis reveals that about 10-20% of these preganglionic cells are involved in controlling pupil size. Vergence of the eyes involves the medial rectus muscles. In the primate oculomotor system, the motoneurons supplying the medial rectus muscle are subdivided into three subgroups (A, B & C). The C-subgroup is located at the dorsomedial corner of the oculomotor nucleus, near EW, leading to speculation that these motoneurons play a particular role in the convergence component of the near triad. Several lines of evidence indicate that the organisation of the medial rectus nucleus is similar in birds.

Evidence suggests that, in birds, the mesencephalic reticular formation projects to both EW and the isthmo-optic nucleus [ION], which is reported to mediate changes in retinal activity and may be involved in the control of visual attention. Our recent studies in chick reveal that a unilateral lesion of ION produces a hyperopia in the ipsilateral eye and disrupts the balance of growth of the two eyes.

The directional sensitivity of cones and rods, why?

Stiles and Crawford showed in 1933 that the sensitivity of the retina varies with the place of entry of the light in the pupil, and thus with the direction of the incidence of light upon the retina. This phenomenon is called the Stiles-Crawford effect of the first kind (SCE I). The SCE I is accompanied by a hue shift, discovered in 1937, usually called the Stiles-Crawford effect II (SCE II). The main features of the latter can be explained by selfscreening of the visual pigments.

A survey will be given of further experimental data, among which the dependence on wavelength, and the directional sensitivity of the three types of cones.

A simplified model will expand the existing theoretical framework, in which the receptors are considered as waveguides. This will describe the data mentioned above.

The main reason for the existence of the SC-effect is probably the straylight from the ocular media. The straylight will decrease the contrast in the images on the retina. The effect of the Stiles-Crawford effect is a decreased sensitivity of the cones to stray light. The near non-existence of directional sensitivity of the rods, when straylight at night-time conditions is almost negligible is in accordance with this hypothesis. How a picture looks like if the Stiles-Crawford effect would not exist, will be illustrated in a qualitative way. A quantitative presentation of the effect of straylight, including the effect of age, is needed.

City University Home | Applied Vision Research Centre Home | Optometry Department Home