A Possible Physical Explanation For Hering’s Opponent Theory
November 18th, 2003 | No Comments »I have proposed that the retina of the eye is narrowly tuned to only three wavelengths and that these correspond to what have historically been termed the “primary colors”. Importantly, in my explanation, both ends and the exact middle of the visible band. are precisely and geometrically defined. The fixed geoemetrically-defined midband point forms the basis for the color constancy of vision. These three wavelengths - defined historically and incorrectly as “RGB” - then are the basis for the trichromicity of vision. The sensation of the hues of “Color” comes later! Separate optical or Fourier transforms seem to be determined at each wavelength and combine to form the final image.The area of the retina that is involved in processing the image perceived by the eye extends from the fovea to generally 20-30 degrees of retinal eccentricity . Beyond 20-30 degrees I propose that the total area (predominantly composed of rods) acts as an “integrating light meter” controlling the pupil of the eye and light entrance into the body of the eye. This explanation predicts that wavelength response… and specifically the short (blue) wavelengths…… will be control pupillary response. This is undoubtedly how evolution would have ordered things… wavelengths beyond this point into the ultraviolet region are the most damaging to the retina.
For the time being, I will separate the subjects of image formation and the synthesis of color and discuss here only the latter.
The three bands on the retinal surface that we propose are not the usual energy spectra but rather are peaks of the intensity of narrowly tuned optical antenna structures (which we define). The shape of these peaks follows from a simple counting of the number and distribution of cone-cone appositions in the fovea and similar calculation of cone-rod and rod-rod appositions at increasing retinal eccentricity. From this understanding it would seem that the eye performs a fundamental, what has been termed a “color” separation into the three defined wavelengths (again, the geometrically defined ends and middle of the visible band) as the first stage of the light detection and image formation process. Indeed, this may define the source of the “primary” color.
The trichromicity curves derived from receptor appositions and the accepted arrangement of cones and rods on the retinal surface are shown in the following figure. (these curves were obtained by simply counting appositions in small increments of angle and I am sure that they are to a first approximation correct. Mathematical methods will surely be applied, however, to further refine them but they will not be drastically changed)
* the color bands in the figure are added with trepidation. It is my assertion that “color” is a sensation percieved by the brain following a synthesis of wavelength signals processed by the retina and it’s underlying neural circuity as taught by Land.. I use color terms here as a “shorthand”.
One notes that these curves are not completely separated but have a considerable degree of overlap.. The long wavelength sensitive (foveal) peak slightly overlaps the mid-band curve from approximately one to three degrees At short wavelengths the mid-band and short wavelength peaks overlap to a far greater degree from approximately the defined mid-band “fixed geometric fulcrum” at 7.5 degrees to the defined short wavelength limit at somewhere near 20 degrees. One will note that the short wavelength limit is not really defined as a peak but rather as a region of high quantum efficiency extending on into the peripheral retina to form, at least I propose, the overall light intensity measurement function of the eye.
So.. what goes on here? This structure obviously results from the evolved distribution of rods and cones of the retina so it must have some meaning..It might be imagined that the distribution of cones and rods would have taken a different form to arrive at three cleanly separated peaks. It is difficult to see how this could have evolved, however, as the slope of the cone distribution curve as a function of retinal angle is already very severe. It seems a worthy hypothesis that the purpose is to effect some form of color mixing.
In studying Hering’s Opponent Theory it seems that this represents yet another obviously “ad hoc” attempt at an explanation for the behavior of the vision system, i.e., setting up a model to explain the internal workings of that system from external observations and experimentation. I do not mean this in a negative way - read on! This is much the same as Edwin Land’s efforts performing experiments external to the eye and then attempting to deduce what mechanisms must be involved within the eye to accomplish the results. I refer here to Land’s educated conjectures, for example, that a fixed, wavelength-defining “fulcrum” must be present somehow in the eye or that the eye must.detect multiple and separate images to arrive at his perception of color vision.
Hering’s effort was similar .. he was attempting to explain, for example, why no pure red could be obtained in vision but rather observing that this wavelength always came “mixed with yellow”. To explain this and other observations he created the concept of “opponent receptors” that had the effect of “balancing things out” and finally arriving at the primary colors. The “opponent” idea, as Land’s ideas , were valuable in the sense that they provided a tentative explanation for the observations. They might or might not, however, turn out to provide a factual explanation for the physical reality of the situation.
One must note that in Hering’s experiments one is “attempting to see the pure primary colors”… and to this end external stimulus is added to cancel some intervening wavelength effects. One must note that this addition of wavelengths is done external to the eye. The intent is to to separate the primary color peaks on the retina of the observer so that he/she sees the “purest” color.
Hering’s theory has found some resonance in apparent identification of a class of “opponent cells” in the neural network underlying the retina whose function is to effect the mechanism that he proposed. This may in fact be true… but such cells may, in reality, have another function… actually balancing (adding or subtracting color) the overlap inherent in the evolved primary color detection curves shown above. The basic elements of an “opponent theory” may therefore be very real. It may prove to be more appropriate to term “opponent “cells “balancing” cells..
I would introduce here the “opponent processing curves” of Hurvich and Jameson that were developed following Hering’s initial concept. (this figure was taken from the website “Opponent Processing Theory of Color Vision“).
Note that this curve is reversed relative to the curve above with 700 nm (red) to the right instead of to the left as above.
I quote from H&J’s observations: “Unique red cannot be obtained with a single wavelength. Even at 700 nm a small amount of yellow is apparent”. This follows from the convergence of the yellow and red curves at 700 nm in their figure.
I would propose that this is consistent with the overlap of the geometrically determined long wavelength (red) and mid-band (green) trichromacy curves introduced above. This small amount of overlap extends from 2-3 degrees of retinal angle.
A direct quote from the web page discussing opponent processing : “Hurvich and Jameson reasoned that when red and green are mixed together they produce yellow, not reddish green”.
That is exactly what seems to be occurring in this new model - the eye mixes red and green to form yellow. As a result of refraction in the eye yellow wavelengths should fall between the mid-band and red curves probably precisely where it does at 2-3 degrees.
The region of the retina from 1-3 degrees is characterized by a rapid decline in the number of cone-cone, long wavelength sensitive appositions and the infilling of rods that form an increasing number of “mid-band sensitive octagonal rod-around-cone centers. At first, in the 1-3 degree region, these centers are separated (in the “sea of cones”)but gradually become more numerous until at 7 degrees they become completely dominant. The neural “circuitry” underlying these centers will perform the necessary wavelength additive or opponent functions. I would hope that the retina plan identified here will give some clue to the organization of such circuitry.
In H&J’s construction the region of mid-band response balances the greatest magnitude of opposing signals. This is exactly the region in the above trichromacy construction where the two bandwidth-limiting and, the third, mid-band curves are most divergent .This is what seems to be represented in the H&J construction.
Note that in my curve mid-band is “pure”, i.e., not interfered with by any other overlapping curve. There is no need for any “Hering opponent correction” to see green - it is a “primary color”. This seems to corroborated by the H&J diagram where the waevlength identified by the observer as green does not seem to require an opponent wavelength.
The red and blue primarys are similar - nothing overlaps the red foveal wavelength. Blue will be similar at angles probablby around 20 degrees where the mid-band curve will not interfere (I calculated receptor appositions only out to 16 degrees of retinal angle).
The small “red opponent” of blue response at the short wavelength end of the visible band is difficult to explain. This would mean in the geometric model that there were long wavelength sensitive (i.e., cone-cone appositions) in the region of the retina from perhaps 8-20 degrees. This is not apparent in morphological measurements that I have seen. There are a small number of cones in this region but they seem widely separated forming ostensibly a few scattered cone-rod mid-band sensitive centers… in a “sea of rod-rod short wavlength sensitive centers”.
Involved here may be the conflict as to how we actually manage to see violet and purple? All other combinations of wavlengths within the visible band are additive while the formation of this color region is subtractive…?
I would believe that the geometric model in the process of defining the precise endpoints and middle of the visible band, actually, and perhaps fundamentally, defines the primary colors - red, green, and blue. The primary red is detected at from 0-2 degrees by the fovea, green at angles from 7-8 degrees, and blue at approximately 20 degrees (where the mid-band curve extrapolates to near zero).
This results from a biologically evolved distribution of cones and rods on the retinal surface that do not, as we demonstrate, cleanly separate the primarys.
Hering then, from external measurements, deduced the “corrections” that must be made in human viewing to obtain such primaries! I think that the term that he used - “opponents” was unfortunate - as noted above some term connoting “balance” would have been more appropriate.
THE UPSHOT OF ALL OF THIS IS THAT THERE MAY IN FACT BE NO DISCREPANCY BETWEEN THE “TRICHROMATIC” AND “OPPONENT” THEORIES OF VISION.
I would readily admit that this is not completely thought out..which I will proceed to do shortly. I thought it of some importance and so am including it on this date.
