Color vision out of the corner of the eye

Scientists from Sydney, Göttingen, and New York have now elucidated how color is perceived in the peripheral visual field. In most humans, color vision is best in central vision and is far less sensitive in the periphery. Is this due to a mixture of color signals at the retinal level or does it reflect suboptimal processing of color signals in the central nervous system? As the team now reported in Nature, peripheral ganglion cells are just as color sensitive as central cells. Thus, our loss of color vision in the peripheral visual field must have a central origin. The results also pose intriguing problems for the nature of the retinal wiring which provides the color signal. (Martin, Lee, White, Solomon, & Rüttiger, Nature 410, 933-936 (2001) )

The human visual system contains two chromatically opponent channels - red-green and blue-yellow - which encode the dimensions of our color experience. The blue-yellow system is phylogenetically ancient in mammals but a red-green system is only present in monkeys and humans; other mammals are red-green color blind. The channels derive from the three photoreceptors responsible for daylight vision, the cones, which absorb preferentially at short (S), middle (M) and long (L) wavelengths; other mammals only possess one M-cone type. The chromatic channels calculate a difference of the cone signals, +S-(M+L) for the blue-yellow channels, +M-L (and +L-M) for the red-green channel. This occurs in the retina soon after the photoreceptors.

The anatomical origin and wiring of the blue-yellow channel is understood but the basis of the red-green channel is unsure. It is known that the retinal pathway responsible is the so-called midget system, in which in central vision a single M- or L-cone connects to a single midget bipolar (the second-order retinal neuron) and then to a single midget ganglion cell, which sends signals to the brain. There are two possibilities; either the 'private line' ensured by the midget wiring is enough to generate a red-green signal for our cortex, or there are additional, highly specific, synaptic mechanisms in the retina which refine the red-green signal to a much higher degree. Such mechanisms have been sought anatomically but not found.

A critical test between these hypotheses is color vision in far peripheral retina, at eccentricities beyond 20-30 degrees. Here, the midget wiring is lost, and the dendritic tree of each ganglion cell of this type makes potential contact with 20-30 cones. With no 'private line', on the basis of this hypothesis, these peripheral cells should no longer be red-green opponent. Our ability to detect color differences is severely restricted in far peripheral vision, and it has been proposed that this is because the cellular basis of red-green color vision is lost at the retinal level. On the other hand, if synaptic mechanisms are involved, then these cells may maintain red-green sensitivity.

In the study now published in Nature, the scientists from the Max Planck Institute for Biophysical Chemistry in Göttingen, from the University of Sydney, Australia, and from the State College of Optometry in New York sought red-green opponency in far peripheral retina using detailed quantitative methods. The results were unequivocal; the properties of the peripheral red-green cells were very similar to those found in central midget cells. Therefore, the loss of chromatic sensitivity observed psychophysically must have a cortical origin.

Figure: Dendritic trees of ganglion cells are usually circular and receive input from all photoreceptors within their domains. However, dendritic trees of peripheral midget cells are often anisotropic; the figure shows an extreme example with a very 'patchy' dendritic tree. The cell has been fitted by eye onto a low-pass filtered cone matrix; the blue blobs represent the S-cones (which are regularly arrayed) and the red and green blobs regions of L- and M-cones (which are random). The shape of the dendritic tree strongly suggests that the dendrites seek out midget bipolars connecting specifically to a given cone type, in this case the L-cone. A mathematical simulation based on anatomical anisotropy data confirmed this supposition. (Reprinted by permission from Nature (410:933-936) copyright (2001) Macmillan Magazines Ltd.)

This finding poses an anatomical puzzle; how can these signals be generated? The dendritic trees of these peripheral ganglion cells are often highly irregular in shape, and a modelling approach showed that this irregularity was consistent with dendritic trees seeking out either M- or L-cones, which are thought to be randomly distributed within the cone array. Two mechanisms for the development of this specificity are possible; either, during early visual development, Hebbian learning occurs in response to the colors seen by the infant eye; or some biochemical marker aids the specific connectivity. The latter possibility might be more plausible, but the sequence differences for the genes encoding the M- and L-cone visual pigments involve just a handful of amino-acids, and it is difficult to see how such a marker could be generated. To elucidate the mechanisms involved provides a future challenge.

Max Planck Society for the Advancement of Science
Press and Public Relations Department

Hofgartenstraße 8
D-80539 Munich

PO Box 10 10 62
D-80084 Munich

Phone: 49-89-2108-1276
Fax: 49-89-2108-1207

E-mail: [email protected]
Internet: http://www.mpg.de

Person responsible for contents:
Dr. Bernd Wirsing (-1276)

Biology, Medicine:
Dr. Christina Beck (-1306)
Walter Frese (-1272)

Chemistry, Physics, Technology:
Eugen Hintsches (-1257)
Helmut Hornung (-1404)

The Arts:
Susanne Beer (-1342)

Online:
Dr. Andreas Trepte (-1238)

ISSN 0170-4656

Prof. Dr. Barry B. Lee
Max Planck Institute for Biophysical Chemistry
Primate Retina and Color Vision Group
37070 Goettingen, Germany
present address:
SUNY, State College of Optometry
33 West 42nd St
New York, NY 10036-8003
Phone: 1-212-780-5149
E-mail: [email protected]

(You may use the text at your convenience. If you want to reproduce the picture in your report you must state the copyright held by Nature: "Reprinted by permission from Nature (410:933-936) copyright (2001) Macmillan Magazines Ltd.". For electronic purposes, the word Nature must be hyperlinked to the Nature homepage at www.nature.com).