Discussion 10F: The emergence of vertebrate vision

Since the dialog will have more to say about how Nature implements these instrumental neural maps, our imagination scenario will chronicle those important progressions as our primordial metazoan evolves into the vertebrate stage, and the dialog notes the end of the Cambrian period with an almost singular visualization.

Our now familiar primordial metazoan, with the neural sophistication developed so far, supported by the mechanical and electrochemical exteroception of its contact sensory complex, and the photonic sensation of its compound eye sensory apparatus, all driven by a locomotive axis having tens of millions of years of phylogenetically developed releasing mechanisms, is now at an evolutionary crossroads.

On one branch of that evolutionary fork, certain species of our motile metazoan will remain on the bottom of coastal waters as they evolve into the arthropod taxon of crustaceans, and terrestrially, the insects. These progressions will require only morphological adaptations for the most part, as the sophistication of their neural arrays have developed sufficiently to support any new morphological specializations. And since intelligence has not subsequently emerged in any arthropod species, the Organon Sutra will have little more to say about this taxonomic line.

On the other branch of the evolutionary fork, we will find the environment continuing to task Nature with lessons that will shape not only the morphology of the non-crustacean subphylum other metazoan species will be developing into, but in many ways will accelerate the elaboration of their neural arrays. And the elaboration of the neural array certainly is the focus of the Organon Sutra.

Just as the phylogenetic success of ramified antennae allowed the locomotive morphology of the early metazoans to develop species with increased mobility, the adaptation of metabotropic state synthesis and neural maps will provide many species possessing compound eyes to develop morphological traits leading to increased mobility for those clades.

And just as the increased mobility of the early metazoans outran the sensory capabilities of their antennae, forcing Nature to evolve an exteroceptive ability beyond that of contact sensation, the increasing mobility in some species at the end of the Cambrian will force Nature to adapt an even more elaborate photonic apparatus than that of the Cambrian period compound eyes.

Although the locomotive morphology of the developing crustaceans need only promote changes which favor adaptations to niche habitats of their bottom dwelling world over increased mobility, the adaptations leading to the vertebrates will be all about increased mobility, and that new mobility emerges as certain Cambrian species rise up off the bottom of the coastal shallows and become free swimming pelagic organisms at the dawn of the Devonian period.

Since the locomotive machinery of this new mobility will not require much more neurological sophistication than the motive complex our late Cambrian metazoan already possesses, the dialog will not elaborate on that morphological evolution.

However, because of this new mobility, the photonic capabilities of Cambrian compound eyes will not provide the necessary signaling for a free swimming organism to abstract a world which is not only changing on its own, but will now be speeding by the organism at an ever increasing rate. Nature will face another selection pressure to evolve our organisms’ sensory exteroception, and along with it, even more sophisticated adaptations to the neural array of our metazoan as it develops into the pelagic vertebrate of the post-Cambrian period.

There are no fossil traces of free swimming vertebrates with compound eyes, probably because that combination would have been structurally and neurologically untenable, and species possessing both of these traits would not have survived the millennial generations needed to leave a fossil record. But these interim species did provide an evolutionary bridge for Nature to evolve the final form of what Charles Darwin himself considered to be one of the chief challenges to his theories of natural selection, even saying that the idea of the evolution of the eye by natural selection was “absurd in the highest degree”.

As bottom-up engineers, we have seen how crafty Nature can be as she responds to the many lessons that the environment demands, and with time on her side, the response to these new selection pressures will be no different. Once the mechanisms for the implementation of neural maps had absorbed the topological mapping of the visual field into neural array itself, there was no longer a need for the huge numbers of individual ommatidia that the arthropod compound eye had evolved, and so it is not hard to see how, over time, their numbers would dwindle in favor of a construction with only one orb possessing a single lens, which focused light onto a two-dimensional carpet of many photoreceptors.

Now, the structural evolution of the early dual-lateral vertebrate eye is but a small part of the story, although it is typically the focus of most researchers. The real story is in the many new neural adaptations that Nature will conjure up to complete the functionality of that structure, and that neurological story will be the focus in the current discussion of the dialog.

As the movement of the world past the eye of the free swimming vertebrate was the chief selection pressure for the adaptations abandoning the compound array, predictably, we should expect that Nature would use that phenomenon as the functional basis for its new adaptations of vertebrate vision. These functional mechanisms comprise many singular adaptations, but the entire collection of neural adaptations to further evolve the vertebrate eye can be grouped into four functional classes, only two of which we need to consider in this current discussion. (The third and fourth functional classes will be discussed as the dialog progresses through the Third Fundamental Precept). The two present classes can be characterized as visual field gaze stabilization, and of course, the all-important image formation processes themselves.

One of the distinctive memes in the early evolution of vertebrate species was the induction of the organisms’ electrochemical exteroception structures and the photonic apparatus into the cranial housing of the vertebrate skull, which became a singular fuselage encasing these sensory devices along with the organisms’ evolving brain, and also providing an attachment for the jaw, and the organisms vertebrate structures. This cranial evolution also brought along with it the development of a distinct system of musculature, distinct in the sense that this collection of muscles did not contribute to the locomotion of the organism, and therefore did not participate in the phylogenetically elaborated releasing mechanisms of the neural motive complex.

One part of this evolving complexity of muscles was the group that supported jaw movement, and these muscles were controlled by reflex centers in the organisms brainstem, an aspect of the new neuroanatomy which also brought about nerve myelination, that was necessitated by the evolution of vertebrae, the dorsal supportive structure for which the subphylum is named. (Although the dialog still characterizes the entire neural structure of early vertebrates as a ‘neural array’ at this stage of post-Cambrian evolution, the term ‘brainstem’ will be introduced for this anatomical feature.)

Also within the evolving cranium were two other distinct sets of musculature, distinguished by their functionality and the center of their neural control. With the cranium providing an encasement of the vertebrate “brain”, which the post-Cambrian neural array will be evolving into, it also developed cranial orbits to house the devolved compound eyes, now in a form of the dual-lateral orbs with much of the anatomical features found in vertebrate eyes today. Within the eyes can be found the next set of muscles, implementing the adaptations for the visual functionality that is termed ocular accommodation, and because the neural centers that control their contraction and relaxation comprise an emerging control assembly in the developing vertebrate neural array, their activities will be described in a later discussion.

Although the anatomical evolution of the vertebrate eye has been the subject of a tremendous amount of speculation by researchers, the bottom-up engineer should by now be comfortable with the visualization of its emergence after tens of millions of years of Natures’ experimentation with the crystalline lenses and ganged photoreceptors of the many ommatidia in arthropod compound eyes. In discussion 9F, the dialog pointed out that the vertebrate eye would not evolve until the neural mechanisms which could abstract its complex signaling had evolved, and in the discussions since then, the progressions which developed that complexity has been chronicled, allowing that post-Cambrian neural sophistication to exploit the signaling resulting from a newly evolving vertebrate eye.

Once the early vertebrate eye had evolved along with the vertebrate cranium, the whole assembly required only one additional mechanism to close the loop of functionality. With its many arrayed ommatidia, the construction of the compound eye had provided a “wide-angle” reception of its visual space, but the implementation of a single lens in the early vertebrate eye resulted in a much narrower receptive field of the visual space surrounding the organism, which then necessitated a mechanical process to adjust the direction of each eyes’ focal axis. Now this mechanical process can be seen in the development of ocular muscles, which coordinate movement of the eye globe within the cranial socket, but the neural controls for these ocular muscles require more detailed explanation.

Just earlier in this discussion, the neural control of these ocular muscles was characterized as visual field gaze stabilization, but this label is actually a muscular simplification of three separate  neural control processes, each of which have complexities which require elaboration.

The first of the neural control cycles needed for ocular muscle control is typically called the vestibulo-ocular reflex, and this refers to a compensation mechanism which results from the new form of pelagic vertebrate locomotion. Fishes swim with a simple lateral motion, and as their tails produce thrust, they also apply torque to the vertebral column resulting in some lateral movement of the head. Although much of this torque is absorbed by the dorsal and pectoral fins, the remaining head movements displaces the cranial orbits in their visual field, which requires a compensation mechanism to stabilize the gaze angle of the eyes relative to a fixed point in the field. Now, there are vestibular sense organs which provide a simple reflex circuit for constant stabilization, but this stabilization must be modulated because, certainly, there are also reasons why the remaining neural control processes of gaze stabilization want to change the gaze angle of the eyes.

To explain the remaining two neural controls for the intentional movement of the eyes’ focal axis, the dialog must digress a great deal and detail the neurophysiological processes by which the evolved vertebrate eye begins to produce an “image”.

In the human vision system, there are about 1.5 million retinal ganglion cells sending axons to the thalamus from each eye, and the brain produces a complete analysis of their patterns of illumination about 10 times a second. If this were done by a passive, hierarchical, “pixel-by-pixel” examination of each of the 3 million illumination levels every 100 milliseconds, even the massive processing power of the human cerebrum would be overwhelmed. Surely Nature learned early on that even with the parallel processing capabilities of the brain, a complete analysis of every illumination signal in terms of color, intensity, depth and motion 10 times a second would be neurologically unfeasible.

Nature has accomplished the extraordinary feat of vision by going to some extremes in order to reduce the data at the front-end, in the very retina itself, by evolving several techniques which determine what not to look at.

If an area of uniform illumination is surrounded by an area of a different degree of illumination, the constant information within the uniform area is redundant. And if the perception machinery had a mechanism that could detect only boundaries between different illumination levels, the signals from the center of a uniformly bounded area can be discarded. At some later point in the analysis process, the information about this redundant field could be accurately reconstructed if needed in the imaging process, by simply extrapolating the illumination level on the inside edge of a boundary across through the region of discarded signaling. And the key to this imagery magic is the “if needed” part.

This boundary detection strategy is similar to the boundary resolution process that was described as the neurophysiological form of Natures’ first step in moving from local phasic activity to global associations in the previous discussion. In that discussion, neural maps implemented a boundary resolution process that was developed by Nature for discriminating the signaling in the late Cambrian arthropod compound eye, and it was described as the second part of a two-part technique, with the first part characterized as the contrast extraction function between individual exteroceptive signaling elements themselves. And perhaps the bottom-up engineer might expect Nature to continue this two-part process in any strategy to reduce the data loading for imaging in the vertebrate eye.

However, there was a unique development in the phylogenetic transition from the compound eye to a vertebrate photonic structure. Curiously unlike the almost universal expression of reciprocal connections among neural assemblies, the neural connections from the ganglionic cells in the vertebrate retina to their targets in the thalamus are not mirrored with reciprocal signaling from the thalamus back to the retina. When Nature evolved the vertebrate eye, she developed a neural channel between the retina and thalamus that signaled in only one direction.

Light is a necessary, but not sufficient condition for the initial analysis of an image. What is needed in image analysis is not the signaling of a discrete level of illumination itself, but the contrast between a discrete level of illumination and its topologically adjacent neighbors. And at this point, the bottom-up engineer should recall that the first-stage contrast extraction operation of the two-part boundary resolution process for the compound eye was conducted with metabotropic state synthesis to abstract the temporal component in the contrast function, a synthesis process that requires metabotropic feedback signaling.

But perhaps because of the ontological particulars of the optic chiasm, Nature did not evolve the optic channel between the vertebrate ocular retina and the thalamus with reciprocal feedback signaling. Without feedback to modulate any metabotropic physiology in retinal ganglionic cells, how could Nature capture the temporal aspects of contrast extraction in order to make it useful for the second-stage process of boundary resolution?

Nature undoubtedly did experiment with connected neural ensembles to effect both the phasic and temporal products of the contrast extraction process. But as hierarchically minded designers who have attempted to engineer this process in a connectionist manner have discovered, modeling temporality explicitly, and elegantly, is as difficult as engineering emergent behavior, and for the same reason.

So how did Nature implement the temporal functionality of the contrast extraction process in the vertebrate eye without the aid of metabotropic feedback?

Researchers have identified several movement modalities of mammalian eyes, such as intentional gaze fixation after attention shifts, the subliminal saccadic motions of the eyes and the vestibulo-ocular reflex. In addition to these motions, there is another motion which is not detectable by ordinary means. When the eyes seem to be at rest, and none of the previously mentioned motions are active, there is still a very fine cyclic motion of both eyes with a frequency of about 10 cycles per second.

This cyclic motion is rather erratic, although its amplitude is sufficient to move the image from a fixed object back and forth on the retina by a distance equal to a few times the average separation between the reception field of individual retinal elements.

Because of this cyclic motion, those retinal elements near a boundary of uneven illumination levels are swept back and forth continuously from the lighter to darker sides of the boundary at about 10 times per second. This produces a changing signal to the retinal ganglion cells, and it is this alternating signal which forms the temporal component used with the static phasic function of contrast signaling by the center-surround ganglion cells.

At the same time, for receptive fields of the retina being illuminated by patches of constant illumination, their receptive elements are not swept with changing levels, and therefore do not contribute an alternating temporal “signaling” to the contrast function of their retinal ganglion cells.

Because of this arrangement, the information transmitted to the thalamus from the center-surround retinal ganglion cells is reduced to the contrast levels for receptive elements having a different level than their topologic neighbors, drastically reducing the data load presented to the image formation assemblies of the brain, while still providing for its reconstruction when needed. (Although this mechanism describes the contrast extraction of light intensities, the dialog on the Third Fundamental Precept will relate how the same mechanism is used for encoding the color extraction process in retinal ganglion cells as well.)

Nature has evolved several additional techniques in its strategy to determine what not to look at, but the dialog will focus on this process as the discussion returns to the imagination scenario of our evolving Devonian vertebrate.

Because of its ontogenetic nature, the neural pathways from the early vertebrate eyes to the primordial organisms’ optic tectum probably developed the same characteristics which precluded a feedback path from the optic tectum back to the ganglionic cells in the organisms’ retinas. Indeed, it has been observed that the optic nerves in extant fish and amphibians regenerate when severed, growing from severed ganglionic axons back to their original targets in the optic tectum, an ability that would preclude the genesis of reciprocal connections. This uni-directional signaling would necessitate a similar mechanical solution to the first-stage contrast extraction process needed for the boundary detection in image formation.

And although the early vertebrates probably did not utilize the 10 Hz micro-movements of the eye that is observed in extant primates today, the post-Cambrian organisms would still utilize mechanical methods to implement the contrast extraction “data compression” process that their neural arrays (less sophisticated than the extant mammalian central nervous systems of today) would need even more.

This discussion had to digress from the narrative explaining the intentional movements of the gaze angle in free swimming vertebrates in order to illustrate how Nature had substituted mechanical movements in vertebrate eyes for the metabotropic processes previously used for contrast extraction, but now the dialog can return to the description of the neural mechanisms that closed the loop of functionality first developed by Nature for the imagery of light.

It was the movement of the world as it changed on its own that spurred the development of the contrast extraction/boundary resolution process for the arthropod compound eyes in the first place, and it was the translation of that already changing world past the free swimming vertebrates that engendered the development of the vertebrate eye, so it should come as no surprise (again) to the bottom-up engineer that Nature would take the complication of imaging a world that is moving past her new photonic apparatus and turn the complication into an advantage. Free swimming vertebrates did not implement the 10 Hz fine movement of tetrapod eyes because the world that is being imaged was already moving past them.

In addition to all of the other mechanical methods just described to reposition the gaze angle of free swimming vertebrates, there is an eye movement known as the optokinetic response in fish, which is traditionally considered to be a reflexive behavior, but like the vestibulo-ocular reflex, it must be considered in the overall context of free swimming gaze fixation.

The optokinetic response is also found in the many movements of tetrapod eyes as they follow a moving object in smooth pursuit, but in fish, the optokinetic response is a combination of both smooth pursuit and rapid saccadic eye movements because, as opposed to land animals, their entire world is always in motion, and this gaze tracking mechanism is used to visually counterbalance the forward motion of the fish as it swims in the water. And just like most of the phylogenetic adaptations we have witnessed so far, to fully understand how it contributes to the developing imaging processes in fish, it must be considered from the overall perspective of the organisms’ neural array that is directing this behavior.

To take advantage of the translational motion of the world as a fish swims forward, Nature engineered the optokinetic response in the form of a two-step, cyclic behavior. The cycle of the response is synchronized with the rhythmic phases of the fishes’ fin strokes, and the first step of the primordial optokinetic response was keyed to the fin-stroke phase which generated the most propulsive acceleration, during which the optokinetic muscles would shift the gaze angle of the eyes toward their forward most looking position in a rapid, saccadic trajectory, where the gaze is then held momentarily steady.

The second step of the optokinetic cycle would be synchronized with the refractory phase of the organisms’ fin stroke cycle, when the rigid curl of the thrust stroke relaxes and the organism coasts on its forward momentum. During this second step, the optokinetic muscles would again re-assert the positioning of the eyes, and in this phase the gaze angle would be commanded in a smooth pursuit to track the world in its retrograde progress past the visual field of the fish.

During this second step, the organisms’ neural array would begin the analysis of the contrast extraction signals produced by the first step in order to resolve those global boundaries which might be present in the topologic field of contrast signals, much like the process for Cambrian “vision”. But certainly it didn’t take too much experimentation on the part of Nature to discover a curious effect as this second step of the optokinetic cycle progressed.

Recall that one of the tricks that Nature had developed to engineer the data compression functionality needed for visual processing resulted in ganglionic cells that required mechanical motion of the visual field across individual receptive elements of the retina, in order to elicit the necessary alternating signaling for their center-surround contrast extraction.

Now, during the second step of the optokinetic cycle, the gaze angle of the eyes are moving in a smooth pursuit, tracking the motion of the world as it passes the visual field of the fish, and even though the eyes themselves are moving, their gaze angle is fixated on a particular point in the world.

During this fixation, the levels of illumination of the world in visual space would no longer be sweeping across the individual receptive elements of the retina, and without an alternating contrast, the visual world would, in effect, “disappear”.

The bottom-up engineer might at this point wonder how this could be productive in the image analysis process. But a moments reflection might suggest the enormous utility in this arrangement. As the visual world disappears due to the gaze fixation of the optokinetic response, those “things” in the world that are moving in relation to the point of fixation would not disappear. And their continued movement could represent attractive prey or mates for our organism, or more importantly, dangerous predators, the very conditions for which Nature has spent tens of millions of years evolving our organisms’ neural array to resolve.

Even before a single neuron beyond the retinal ganglion cells has begun an analysis of the image, our primordial vertebrate has already learned to resolve “things” in the world separate from their background.



Copyright © 2019 All rights reserved