Discussion 11C: The foundations of Attention

In the movie “Jurassic Park”, during the scene when Dr. Grant and Dr. Hammonds’ grand-daughter Lex first encounter the menacing Tyrannosaurus Rex, Lex instinctively lets out a scream, at which point Dr. Grant admonishes her, saying “Don’t move! He can’t see us if we don’t move…”

That particular scene has generated significant buzz in many circles, and there has been all manner of conjecture advanced on the scientific accuracy of Michael Crichton and Steven Spielberg’s screenplay characterization of dinosaur vision.

The Organon Sutra is contributing to that speculation only in passing, by adding that it would have been more accurate to say “He won’t eat us if we don’t move”, but such pedantic lines would never find their way into a blockbuster movie screenplay.

Tyrannosaurus Rex was technically not a denizen of the Jurassic era, and by the time of his appearance some 68 million years ago, in the upper Cretaceous period, his vision systems were well evolved from his Jurassic ancestors. The position of the cranial orbits placed the eyes closer together with a geometry fitting for most predators, and its narrow snout provided for an unusually wide binocular field.

With about 55 degrees of binocular overlap, the vergence vision system in Tyrannosaurus Rex had remarkable close-in perception, well adapted for predation. Now, certainly, any movement by the characters would have given away the presence of Dr. Grant and Lex in the movie, but in all fairness to the probable accuracy of Steven Spielberg’s screenplay, what saved the two characters from the non-motion based vergence imaging capability of the dinosaur was the close proximity to the overturned tour jeep they were standing just in front of. The image resolution of T. Rex vergence vision is based on binocular depth, and not motion, but Dr. Grant and Lex were situated so close to the jeep that T. Rex probably would not have resolved a difference between them and the vehicle.

The whole cerebellar dialogue in the last discussion was about egocentric locomotion, for which the vergence based vision system provided the image maps that was combined with all other sensory mapping to vector the real-time kinesthetic space needed for tetrapodal mobility.

But what has been going on with the other, motion-based optokinetic vision system during all this time?

For both the dinosaurs and the mammals, the two vision systems have been separate but co-existing in the optic tectum since the early tetrapods began their land occupation. The motion-based optokinetic imaging system has been predominantly an input to the emotive complex and the basal ganglia, developing imagery of things that move relative to a stationary background, while the binocular-based vergence imaging system has provided the cerebellum with the signaling needed for depth perception in egocentric mobility.

But because there is only a singular optical apparatus supporting both the optokinetic and vergence vision systems, the cerebellum must engage in some politics with the emotive complex to avoid monopolizing the gaze control over the eyes.

This vision system politic was worked out because the optokinetic process had millions of years of evolution producing the mechanism which warned the organism of the presence of predators, and so this system had to have priority access to the optical apparatus, including the visual orientation of the head and neck. But conversely, because any terrestrial organism would starve if it could not move and locate food, then the cerebellum required that the optical apparatus support the vergence imaging process while the egocentric motor system was engaged. For the primordial tetrapods, the “political” solution that was established, in what would become the attention mechanism in extant vertebrates, gave the cerebellum control of the optical apparatus while the perceptual diffraction circuits of the emotive complex were signaling a predator-free environment.

However, with the very first signal from the auditory or olfactory system that the environment had changed, or with the “pop-out” signaling of something in the visual peripheral field (even though the cerebellum has control of the gaze direction of the eyes at this point, the optokinetic imaging process in the optic tectum is still in operation), this attention mechanism would instantly engage the eyes, head and neck to center the gaze direction on the source of the “interrupt”. This forced engagement would halt the animals’ movement while the emotive complex evaluates whether to command the basal ganglia in an avoidance maneuver or not.

This attention mechanism allowed the two distinct vision systems to co-exist in the optic tectums’ of the dinosaurs and mammals through to the end of the Mesozoic era, working well for hundreds of millions of years, until a visit to their world by an extraterrestrial object about 66 million years ago.

The end of the Mesozoic era is demarcated by another mass extinction event similar to the Triassic-Jurassic extinction event of 201 million years ago. There is a significant amount of evidence that an asteroid fragment impacted the Earth at a point in what is now the Gulf of Mexico, about 66 million years ago. The impact was of such a magnitude that for a period, photosynthesis was globally halted in plants on land and with plankton in the sea, having after-effects resulting in the extinction of almost 75% of the Earths’ living plant and animal species, both terrestrial and aquatic. None of the dinosaur species survived, and among the mammals, only those mammalian species below a certain body mass survived.

But this mass extinction was also a grand opportunity, which allowed the formerly insectivorous and nocturnal mammals to come out from their night-time burrows and flourish in the vacated daytime ecological niches, many species of which demonstrated a marked trend toward a larger body size.

The dialog has touched on the development of the cerebellum in the dinosaurs and mammals for the two-hundred million year period prior to the asteroid impact, as the cerebellum came to excel in its ability to merge the dimensionally disparate maps and spaces of the exteroceptive and somatosensory systems into real-time sequential kinesthetic programs.

But the resultant architecture of the cerebellum was monolithic, and, although well-adopted to the transformations between maps and spaces, it could not create those domain maps itself, and as the surviving mammals came out of nocturnal habitats after the K-Pg asteroid event to assert control over the now vacated ecological niches they would occupy, these species would need the neural capability to construct maps and spaces whose invariant semantics were not just egocentric to the organism.

These developing species in the early Cenozoic era would need the capability to create domain maps that had allocentric frames of reference independent of the sensory or effective topologic domain that they represented. And this selective pressure would accelerate the development of the neural assemblies expressing an architecture which we call the cerebral cortex.

All of the maps and spaces being provided to the cerebellum at this point in neural evolution had topologic domains which were reflective of the sensory modus which produced them. The visual sense produced maps with spatial topological domains, the auditory sense produced maps with frequency and amplitude domain relationships, producing a harmonic topology, and the somatosensory assemblies created maps with a homuncular topology.

Where the Second Fundamental Precept developed the conceptualizations which allowed the dialog to express definitions for the abstraction of instantaneous signaling, the Third Fundamental Precept is intended to develop the conceptualizations to allow the expression of definitions for the abstraction of entire maps and spaces being created by that instantaneous signaling.

After the K-Pg asteroid event 66 million years ago, mammals diversified quickly in a new world with many of their natural predators removed, and the selective pressures which resulted in increased body size brought along a concomitant increase in the size of the evolving primordial central nervous system.

But characterizing the expansion of neural systems quantitatively consistently leads neuroscience research to inconclusive ends. A fundamental premise of connectionism in neural models is that individual units do not transmit larger amounts of information as the collective number of units increase, instead the computation increases as units are appropriately connected to larger numbers of similar units. Yet this model has the flaw of simplicity by omitting the consideration of state retention within an interconnected network.

In the act of biological adaptation, an organism will absorb some information from its environment. To do this, an organism must have a mechanism to sequester selected subsets of the continual flow of instantaneous signaling within its neural assemblies. Now, if an organism were to develop an ability to temporally retain all of its exteroceptive signaling, then it will soon find its behavior following the entropy of its environment, and by definition, in time will simply become inorganic.

So, certainly, the mechanism that selects the subset of signaling to be retained, the mechanism that “pays attention” to certain salient aspects of sensation while ignoring others, is key to an organisms adaptation.

However, the organism must have the machinery to assimilate the signaling in the first place. And the bottom-up engineer must maintain a clear separation between the concepts of the mechanism that selects subsets to the signaling to be assimilated, and the machinery that performs the assimilation itself, for a very crucial reason.

Consider this: An organism that can change its behavior based on the assimilation of some information in its environment has a certain survival advantage, but what would happen if that organism were to perform an assimilation which changed the very machinery that selects what information gets assimilated?

As this must occur at some point in the evolution of self-organizing entities, it is important to understand that the attention mechanism developing in our primordial mammals is a double-edged sword. The organism must be selective in its assimilation in order to change its behavior in a survival promoting manner, but what assimilation changes the behavior of the selection mechanism itself?

It would be grossly teleological to propose that the resolution to this question is the reason that Nature evolved cerebral cortex in mammals, but this dialog has not had any aversion to teleological arguments previously, and, although it is scientifically dubious, proceeding from this axiom will provide the bottom-up engineer with the proper perspective to understand Natures’ true intent, which was the development of mechanisms that expressed allocentric mapping of the environment.

In the primordial tetrapod, the optic tectum and auditory system and somatosensory systems had been developing all manner of maps for their specific sensory domains, for which Nature dutifully evolved the cerebellum, which excelled at sandwiching all of these maps to produce the egocentric motor behaviors needed for terrestrial locomotion. But now that the primordial tetrapod can move about, how will it go about adapting to its terrestrial environment?

Adaptations for our primordial organisms has so far been accomplished phylogenetically, but as the mammals emerge from their nocturnal burrows into the chaotic daytime environment following the K-Pg asteroid impact event, an environment now different than the one they were phylogenetically prepared for, they would need an adaptation mechanism that performed on a much shorter time scale than the one exhibited by the genome. Assimilation of information from the environment (as opposed to the mere abstraction of perceptions) must now be accomplished within the life-time of an individual organism.



Copyright © 2019 All rights reserved