Rethinking the Process of Vision

My general writing style is to try quickly to outline the complete thought or thoughts and then to go back later and ‘fill in the blanks’ putting the ideas into hopefully reasonably readable text.. In the following you will be treated to the original outline because as I think that it is a rather important summary I will publish here and complete the thoughts in future days.
GCH

The text:

I believe that we should begin to view the eye as Roger Penrose (and I think Koch) proposed as the ‘extension of the neural network of the brain where this network interfaces with external reality’ and not as we have historically considered as the ‘singular organ of vision’. In other words, the eye and brain operating on the same – quantum – principles.

One’s thinking here should start with the points about which I have been writing lately, namely the statement made by Albert Rose that the vision process has evolved to the quantum limit. To quote him exactly:

“It is not conceivable that the sensitivity of the vision process can be significantly enhanced. The visual process is at an absolute terminal point in the evolutionary chain. To the extent that the visual process now succeeds in counting each absorbed photon, there is little possibility, outside of increasing absorption, of a further increase in sensitivity. The laws of quantum physics impose a firm limit which has been closely approximated by our existing visual apparatus”

(emphasis in the original, from, Albert Rose, “Vision Human and Electronic” Plenum Press, pp 29-30)

This thought has truly fundamental meaning that seems not to have been seriously considered.

To begin, the ‘retinal signal’, i.e., the isomerization of the retinal molecule following the absorption of a single photon , is of a very small magnitude indeed. In electronics terms it is very difficult to see how this signal, competing against the thermal noise background of the temperature of the body, can be amplified to a useful level (…for transmission to the visual centers of the brain etc.). As I have written, we technologically require either: a.) application of electronic ‘gain’ of values approximating a million or, b.) reduction of thermal conditions (temperature) to a few degrees above absolute zero ( - 273C). Rose with his background in electronic image technology could only assume that the required gain must be present somewhere in the biological system but, he could not even begin to imagine where or how this might come about.

Another group (F.Rieke, D.A. Baylor, “Single-photon detection by rod cells of the retina” Rev. Mod. Phys. 70, 1027 – 1036,1998) make essentially the same point that Rose makes. Their conclusions are so interesting that I will again quote their text exactly:

Abstract: “At low light levels, the visual system detects and counts photon absorptions with a reliability close to limits set by statistical fluctuations in the number of absorbed photons. Thus the rod photoreceptors that provide the input signals to the dark-adapted visual system act as nearly perfect photon counters. This elegant performance is possible because light detection in the rods satisfies four functional requirements: high quantum efficiency, sufficient amplification to produce a measurable response, low dark noise, and low trial-to-trial variability in the elementary response. The rod meets these requirements using biochemical reactions rather than the solid-state reactions of silicon detectors, yet its performance equals or exceeds that of man-made detectors in several ways.”

I have emphasized their last point which might indicate that the subject of biological reactions is understood. But the last paragraph of their conclusions:

“Although our understanding of visual transduction and signal processing has advanced rapidly during the past 15 years, several fundamental questions remain: (1) What mechanisms are responsible for the reproducibility of the rod’s elementary response? (2) How are the rod’s single-photon responses reliably transmitted to the rest of the visual system? (3) What is the molecular basis for the differences in kinetics, sensitivity, and dynamic range of rod and cone photoreceptors? The answers to these and related questions will deepen our understanding of the strategies and capabilities of biological detection”

I believe that the thoughts expressed in their conclusion might lead one to think that the biological reactions mentioned in their Abstract are perhaps not so well understood!

Now back to my line of thought.

Albert Rose and the above referenced investigators both use the construction (erroneously, I believe) that an ‘integration time’ of a fraction of a second or so is inherent in the vision process. Rose comes about this thought naturally as times of this magnitude are involved in photography and electronic imaging as necessary to provide statistically significant signals to form an image. But, as I have written, I believe that this ‘integration time of the eye and vision’ does not correspond at all to the eye alone (or more, precisely the retina alone) but rather encompasses signal transmission from the initial light interaction with the retina over the entire brain/ nervous system to ‘pushing a button’ to record a visual event.

The realm of quantum physics implies consideration of space and time..and very small increments of each. How can the eye detecting at the quantum limit as it does be reconciled with the macro-increments of time that have been assumed?

The answer is that it cannot. The vision process must be operating in very fast time indeed…in a time domain heretofore not considered.

As my thinking has progressed the following model seems reasonable:

The central point is that each of the hundred million or so detection sites on the retina must be considered individually…as a vast array of sensitive devices acting in parallel to detect (or ‘count’) single photons. This is the only way that one can assume that single photon counting can come about…individual, fast counting devices sufficiently sensitive to respond to the energy of a single photon.

And in seeming correspondence, the time scale of operation of these individual devices must be very fast indeed with signal- obscuring noise (see my previous reference) being a time integrated function.

And…I have outlined the probable structure of these receptor-appositional devices.

This line of thinking leads inexorably to the concept of a fleeting ‘quantal image’ impressed on the retina in very short time (probably less than 10-15 seconds). We know very little about such an image as it is buried in the area of physics termed quantum statistics.

And then..how does it come about that the human system perceives and processes such an image. It must convert this image data into a slower (or integrated) form compatible with the human nervous system and human comprehension.

The mechanism I believe is contained in slow transmission through the coherent bundle of fibers that constitute the optic nerve. This nerve bundle of finite length serves the function of ‘slowing the retinal acquired quantal image to human nervous system-responsive time’. It is also possible that this transmission performs a time integration function to allow image information to reach a level of statistical significance.

‘Looking outward’ one must not lose sight of the idea that the retina ‘interrogates the realm of (‘outside’) quantum reality’. I have written and speculated about this.

‘Inward looking’ again, I have been fascinated with the seeming fact that the retinal image is transmitted to the visual centers of the brain as the image itself, i.e., there seems to be a 1:1 spatial correspondence of image pixels or elements existing between retina and brain. The image itself appears in the brain!

If this is so then the geometric rule (my geometric “Rosetta Stone”) that I propose is at the basis of formation of the retina itself must be transmitted to the brain. Might this give some insight into the structure of neural networks contained therein?

Does all of this bring together the elements of the eye and vision, the neural structure of the brain, the realm of quantum physics and, finally, pure geometry? It seems to me that this explanation makes a case for this..

Paraphrasing Einstein: ”Geometry may be at the heart of everything”.

GCH/1/28/07

No visual system ’sees color in the dark’! The basis of my explanation for the vision process is that the diameter (or lateral dimension) of retinal receptors controls the response of the retina to wavelength and not pigment molecules within each receptor (rhodopsin acts as a structural component to position the retinal  electronic signal-producing molecule within the differing diameters of cones and rods). It is receptor diameter (of cones and rods) that determines the center-to-center appositional distance between adjacent receptors. A new form of ‘optical antenna’ that I describe is formed where the dimension (length) of these antennas determines wavelength response. Importantly, there are only three geometrically determined lengths. The sensation of the ‘hues of color’ requires that two receptor sizes be admixed on the retinal surface. This sensation is not the result of pigment molecules or complexes contained within individual retinal receptors! What we term the ‘hues of color’ follow from the eye receiving sufficient light intensity on either side of a geometrically-defined mid-band wavelength to determine a valid ratio of the two intensities. The hues of color can only be determined if two different sizes of receptors are present. Vision otherwise is monochromatic. The geometrically defined mid-band reference is exactly the ‘fulcrum’ point the presence of which was brilliantly deduced by Edwin Land. Thus, the biological retina needs no laboratory instrumentation to define wavelength - it is determined by light refraction within the eye and geometry!

On the subject of nocturnal vision it is crucial to note the unassailable measurements (see my previous Comments on this subject) demonstrating that evolution has brought vision to the ultimate sensitivity, i.e., that the vision process is able to ‘count’ single (or at least a few) photons. This statement has fundamental meaning (at least in physics) but seems to have been disregarded by vision science. The sensitivity of vision cannot be further increased! Any idea therefore that such statements as ‘the hues of color can be sensed in the dark’ are meaningless. Vision cannot discern the hues of color in the dark for the simple reason that too few high energy (blue, for example) photons interact on either side of geometric mid-band to determine a useful ratio. The only photons available are of lower energy in what is termed the near infrared. An analogy - water cannot run uphill!

Also, low light level night vision systems developed extensively by the military knowlegably employ technology that extends the response of such systems beyond the visual bandwidth into this near infrared (or NIR) region that extends roughly from 700-1000 nanometers (or from 0.7-1.0 microns).- using the only photons that are available. (one will remember that the human visual band extends from 400-700 nanometers or 0.4-0.7 microns). Such night vision systems generally use what is termed an ‘S-1′ photosurface that is sensitive to wavelengths in the NIR region but is to a large degree insensitive to visible light.The NIR signals are amplified and converted to a visible phosphor display for human viewing.

It follows therefore from my explanation of the vision process that the retina of any specie that evolved to possess the capability for nocturnal vision would be found to be composed of receptors having a larger diameter than the human variety (the diameter of cones found on the human retina is about 2 microns). I recently came upon the following paper describing the retinal morphology of the Gecko family of lizards - and it goes further in validating my concept than I ever could have expected! The dimensions of retinal receptors of both nocturnal and diurnal species are measured with, in summary, nocturnal retinal receptors being large (very large) with those of the diurnal species being close to the dimensions of human ‘daylight seeing’ vision. What could be better!

The abstract of this paper follows with the emphasis being mine:

_________________________________________

Biomedical and Life Sciences and Medicine
Issue Volume 29, Number 7 / July, 2000
(1) Lehrstuhl für Tierphysiologie, Fakultät für Biologie, Ruhr-Universität Bochum, D-44780 Bochum, Germany

Abstract: Geckos comprise both nocturnal and diurnal genera, and between these categories there are several transitions. As all geckos depend on their visual sense for prey capture, they are promising subjects for comparison of morphological modifications of visual cells adapted to very different photic environments. Retinae of 22 species belonging to 15 genera with different activity periods are examined electron microscopically. Scotopic and photopic vision in geckos is not divided between ldquoclassicalrdquo rods and cones, respectively; both are performed by one basic visual cell type. Independent of the activity periods of the individual species, the visual cells of geckos exhibit characteristics of cones at all levels of their ultrastructure. Thus, gecko retinae have to be classified as cone retinae. Only the large size and the shape of the photoreceptor outer segments in nocturnal geckos are reminiscent of rods; the outer segments are up to 60 micrometers in length and up to 10 micrometers in diameter. The visual cells of diurnal geckos have considerably smaller outer segments with lengths ranging from 6 to 12 micrometers and diameters ranging from 1.3 to 2.1 micrometers. Nocturnal and diurnal species differ in the structure of their ellipsoids. One type of visual cell in nocturnal geckos has modified mitochondria with either rudimentary cristae or no cristae at all, and one type of visual cell in diurnal geckos possesses an oil droplet. The visual cells of Phelsuma guentheri and Rhoptropus barnardi are intermediate between those of nocturnal and diurnal species.

___________________________________________

One sees that the receptors of the nocturnal species are measured as being 10 micrometers in diameter while receptor diameter of the diurnal varieties are smaller being from 1.3-2.1 micrometers. Retinal receptors of the nocturnal gecko are thus some 5X larger than the human variety. I wold project that this dimension extends the visual range of the nocturnal species into the near infrared or NIR region of the spectrum.

The smaller receptors of diurnal gecko species are found to be similar in size to the long wavelength-defining cones on the human retina. This leads one to believe that the vision of these species possess a ‘daytime seeing’ visual band much as we humans. That the receptors are slightly larger than human variety by perhaps 2x which would indicate a shift toward sensitivity at longer wavelengths - but not nearly to the 5x increase of the nocturnal specie.

To go further in defining the visual characteristics of these (or any) specie I would require a more detailed morphology (’retinal map’) of a distribution of receptors - assuming that more than one size were present. This would be very similar to the retinal map of the human retina published by Osterberg (and referenced in my paper). If only one size of receptor is present vision as the above reference seems to indicate, vision will be to a large degree monochromatic, i.e., centered around the wavelength determined by the single receptor apposition dimension. In antenna terminology this is a “high Q” antenna geometry. As noted above, any possibility of sensing the ‘hues of color’ would require that two different sizes of receptors be present and admixed across the retinal surface.

Now, what exactly would all of this conclude that the nocturnal Gecko sees? Given only the above information I would conclude that vision of the nocturnal Gecko lies in a narrow band somewhere in the infrared region of the electromagnetic spectrum beyond the visible. The retina would not be sensitive to visible light! Thus under no circumstances could it discern the, what we term, ‘hues of color’.

Let’s be more precise in defining the wavelength that it does see. Assuming that the long wavelength (red) limit of the human visible band is 700 nm (or 0.7 microns), and that this corresponds to a wavelength-defining receptor dimension of ~ 2 microns, the Gecko receptor being 5X larger.( assuming a linearity of response that may be reasonable) would be responsive at ~ 3.5 microns. This is somewhere between what is termed the Short Wave Infrared SWIR band (1-3 microns) and the Mid Wave Infrared MWIR band (3-5 microns).

Since I have no means of imaging at these wavelengths I will illustrate what I believe the Nocturnal Gecko sees using the Near Infrared NIR band ( from 0.7-1.0 microns).These images can be obtained using an NIR filter (that passes NIR and blocks visible wavelengths) and a state-of-art digital camera. It is not quite this easy …but almost. . A website that contains teaching and instructve NIR images is:: Infrared (IR) Basics for Digital Photographers-Capturing the Unseen.

I believe that these NIR images make the point that I am trying to make, namely, that what the Gecko sees are contrasts in black-and-white that correspond to the blue-to-red visual band that we sense in the human visible band as the ‘hues of color’. Again, I would contend that the species does not at all, and cannot, see ‘color’ as we define it.

At NIR wavelengths the following light reflective characteristics are observed: a.) light reflection from green foliage, a probable background comprising a Gecko’s vision, is greatly enhanced appearing as bright white, b.) surprisingly, and perhaps counter intuitively, red colors at the long wavelength end of the visible band appear even whiter, and c.) blue colors at the short wavelength end of the visible spectrum appear dark even to black.

From this one might imagine the characteristics of nocturnal Gecko - a (even enhanced) black and white gray scale corresponding to light reflected from objects in the visible wavelength (RGB) band. This would be set against the enhanced white reflection from green foliage.

The following series of images illustrate this effect. The first is a normal visible range, unfiltered image of spots of blue and red paint together with a bit of green ‘foliage’ (one knowledgeable of leafy things might recognize parsley). The image shows exactly what one would expect in the visible range. In the next image the same scene is photographed in the NIR region using a Hoya 72 filter that blocks essentially all of the visible band. One sees that the red is almost invisible against the white background (Indeed, the above referenced NIR website indicates that the color red can be viewed in the NIR as whiter than a white background). The blue spot in contrast appears a very dark gray even to black. Thus, in the NIR region there is a grayscale across the visible band extending from red to blue. The green ‘foliage’ is as expected highly reflective in the NIR showing a high contrast white.

Please note that these images are quite un-optimized with, for example, the color spots having some shine or reflectivity. Ideally a matte surface would yield even more black and white contrast.

Next, we attempted to simulate ‘what a Gecko might sees in a usual (to him or her) outdoor setting using my desert backyard at night after sunset. The same two bottles of blue and red paint were set into some local foliage (?) together with a blue colored tweezer that was available. Note again that these two photos are not optimized with the blue and red colors changed by viewing through the reflective glass of the bottles and the objects themselves were placed deeply into the vegetation. If they had been on the same plane as the vegetation their reflectance would have been more pronounced. Note the intense white representation of the green foliage.

But here they are:

Outside NIR image of the above scene …note blue tweezer and red and blue bottles of lacquer:

I expect a rejoinder to my proposal to be that “microspectrophotometric measurements of Gecko (or other nocturnal species) retinae show a pigment response that belies your NIR prediction!”. I happened to come across a site today that used this rationale in measuring the response of another nocturnal specie - the owl. In that work the retinal pigment response was in the same (always the same!) wavelength region as measured human pigments, i.e., at ~ 500 nanometers or the approximate middle of the human visual band. Please see my entry on this site of January 18,2006 ” More on Retinal Pigments” for my thoughts on this type of measurement. In summary,it is my belief that retinal correct response can only be measured with the organ in the living state, i.e., where wavelength-defining spacing between receptors is maintained. Any single receptor - and pigment or pigments thought to be contained therein - represents the ubiquitous (i.e., to all receptors) spatially quantum-confining electron central space. The energy gathering ’surround’, provided in the living state by specific retinal apposition dimension, defines wavelength. When a receptor is ‘teased’ from a section of dead retinal tissue and flattened for microspectrophotometric measurement on a microscope slide. The milieu in which it finds itself (some kind of preservative solution) will effect some dimension of lateral index of refraction gradient….but this will NOT be the state that the receptor was in on the living retina. This refractive index (that will probably vary a bit) gradient always seems to produce very nearly the same ‘pigment’ wavelength response - this type of measurement always yields almost the same curves! Why hasn’t anyone noticed this? Please see my Comment noted above.

The rhodopsin protein component of the retinal/rhodopsin complex that resides within receptors serves in this explanation the function of a variable geometric spacing element assuring energy transfer within the different sizes of receptors - that we have termed cones and rods. I believe that it is a misconception to believe that they are independent ‘pigments’ within themselves.

Just to remind again, the three ‘lengths’ that derive from an admixture of two diameters correspond to the exact ends and geometrically determined center of the band of vision. The size of each component determines the placement in the electromagnetic spectrum of the band.

GCH

1/21/07

Harris (”Rewiring Neuroscience - What Does a Memory Look Like”) in proposing that the retina forms the Fourier (focal or diffraction) plane of the eye points out correctly that Figure 4 of my original paper “What the All-Cone Fovea Images Solely at Long Wavelength” is misleading. I used an optical transform (the only means that we poor humans have of visualizing a Fourier transform!) from Caulfield to illustrate the point that the transform of an outline sketch is primarily (the key word!) a small central spot - that I associated with long wavelength interactions and the central fovea.

What I did not note was that the high spatial frequency information associated with edges of the line drawn sketch was encoded in the outer reaches of the transform. Since there is not much information in the sketch itself (most of the field is uniform white) this part of the transform is almost invisible.

Harris goes on to note (again correctly) that the bright central spot encodes low spatial frequency information associated with intensity-only term of the Fourier equation. He terms this “throw away” information but I don’t believe that this is true. Nature rarely seems to deal in useless information. Rather I think that this might add something to my thoughts.

First, I have proposed that the visual process takes place in the area from the central fovea to approximately twenty degrees of retinal angle. This would be approximately the extent of the high frequency information encoded in the transform noted above. Remember that beyond this angle, i.e., the entire peripheral retina, acts as a short wavelength-defined, wide angle ‘light meter’ controlling pupillary constriction and overall light entrance into the eye and the sensitive retinal Fourier plane. Even with this situation there would still seem to be a need for a brightness (i.e., ‘intensity’) controlling signal for processing the visual image. I might believe that this is the role played by this central part of the transform.

I have probably been misleading in other statements to the effect that this transform (to be precise henceforward: ‘the long wavelength transform’) encodes the ‘outline sketch’ (that I have termed the ‘Marr sketch’) of the perceived image on the fovea. It does encode such a sketch but not directly on the fovea! This was misleading!

GCH

1/16/07