A.I. Circuits vs. Life Fields.
This article is an excerpt from my upcoming book Dynamics of Life. It is a bit longer and not an easy read, but perhaps worth reading to obtain an alternative view to the prevalent A.I. doctrine. On the plus side, it is easier read than my previous article on Medium, Merleau-Ponty and the A.I. though it is not nearly as easy read as AI, the Hoax of the Century :)
Dynamics of Life goes beyond just the A.I. and the Neural science. The central idea of this book is that there are similarities between the dynamics of a population of electrons, the dynamics of a population of neurons and the dynamics of market participants. They all resonate and form wave-fields, rather than static hierarchical or network structures A.I. researchers love so much. New phenomena are not created by Fregean compositionality, i.e., syntax, but by the phase shift resulting from the criticality which systems enter when fields on neighboring scales mutually entrain each other.
In his critique of Cartesian dualism, Prigogine quipped that “triumph” of modern science was the elimination of time. Elimination of time also eliminated evolutionary thinking from scientific arguments. Reductionistic method, the dominant scientific principle today, reduces macro structures to micro-level elements and structure, or syntax, of their timeless composition. This introduces linear causality, which Western science is so fond of, and one-way constraints where macro level syntax completely constrains lower level micro elements. In Selfish Gene, Richard Dawkins proclaimed that phenotype (our bodies) exists for the benefit of genotype (our DNA) and not the other way around. This view is rather similar to the view of Keynesian economics, which sees the individual passively responding to a demand curve and macro level government monetary policies. They are both wrong. In both cases, interacting individuals constrain macro-level structure just as much as it constrains them. In the words of Austrian economist Von Mises: “Common man …is like a sheep in the herd…yet the common man does choose”. What Mises says is that entrainment between neighboring spatial and temporal scales is mutual. This concept is difficult to grasp for a Western mind, used to the explanation of the world in terms of linear causality such as input/output computer metaphor, often applied to human perception in the form of stimulus/reaction model. A good example is a difficulty modern linguists have with grasping the Whorfian idea of language influencing perception and much greater difficulty still of modern biologists to understand Esther Thelen’s generalization that language and perception simultaneously influence each other.
It turns out that the consequences of this mutual entrainment are far-reaching. In a Cartesian and Keynesian world, all systems strive for equilibrium. In Prigogine’s mutually entrained world, on the other hand, populations at all scales operate in a “far from the equilibrium” mode, resonate and form waves, i.e., fields, demonstrating that static structures such as AI type neural networks are insufficient for generation of intentional behavior. Frege’s compositionality principle, stating that the meaning of a complex expression is a function of the meanings of its constituents and the way in which they are combined ceases to apply to large populations at any scale. When it comes to an understanding of the brain and other complex systems, tried and proven reductionism just does not work. It did not work for Walter Freeman when he was researching neuron populations, Esther Thelen’s research of synergies in the development of human reach and walk, David McNeill’s study of the evolution of speech/gesture or for Ludwig Von Mises analysis of markets.
Once we dispense with Cartesian dualism and with its corollary concept of representations, a new paradigm emerges. It is the paradigm of the population dynamics operating on many scales of the universe, from electrons to neurons, to brains, to financial markets on a spatial scale, and from real-time behaviors to developmental and evolutionary time on a temporal scale. These boundaries are always human-made, according to our needs, i.e., they are not hard “objectively existing” boundaries. Rather, we humans select spatial and temporal scales from a continuum of oscillating waves, i.e., fields. For example, person-family, family-nation and nation-world form four distinct scales of social reality with three adjacent boundaries. I further argue that these interacting populations mutually entrain each other between the adjacent scales. Novel phenomena appear via phase shifts, which are fast but continuous changes, not discontinuous “digital” jumps. Thelen states that while evolution appears structured, there are no fixed structures, just dynamic assemblies, i.e., wave-fields.
I postulate that there are clear analogies between the dynamics of a population of electrons, the dynamics of a population of neurons and the dynamics of social interaction as in financial market transactions. They all resonate and form waves-fields, rather than static hierarchical or network structures. This does not mean of course that there are no differences between electron fields and neuron fields, but it does mean that we can apply the same organizational principles to all spatial and temporal scales.
A corollary idea is that adjacent fields interact, on both spatial and temporary scales and thus they entrain each other. Reductionists see only one-way causality, from micro-to-macro direction. Physicist Brian Green in The Elegant Universe concludes it is only a matter of time till we have enough computing power to explain properties of tornados in terms of the interaction of electrons and quarks. However, as we add top-down leg of constraints to reductionistic bottom-up leg, we will see that it makes no sense to ask how molecule interaction or butterfly moving its wings can cause a tornado because there is no linear causality between events on the scale of individual air molecules and the scale of tornados. This is what Carver Mead meant when he stated that “Properties of each electron depend on the state of the entire ensemble.”
The fascinating question, however, is when and how does the continuous interaction of a large population of elements change its behavior and form a new, qualitatively different, macro entity, a new phenomenon. If all matter is composed of electron fields, how come that chair (or diamond lattice) seems solid to our eyes? On most places on earth water molecules usually assume their liquid form, but when they reach a certain temperature, liquid water can turn to ice or steam. How many water molecules must change their behavior before liquid turns into ice? Is this process discrete or continuous? How do electrons “jump” between orbitals? How many atoms does it take to form a molecule? How many neurons are needed to form a cortex? Is helium atom a molecule? Is lungfish still a fish? The answer, just like the question about Jerusalem, depends on whom does one ask. Scientists have difficulties seeing their role in the definition of boundaries between spatial and temporal scales. The majority of scientists do not doubt that, at least in a spatial dimension, nature really is divided into a myriad of scales and that they are merely discovering these objectively existing boundaries.
What are the emergent phenomena? In the last few decades, widely diverse researchers described the role of genes, CAM’s and neuromodulators in the organism development, the role of motor system in the language, and even investor decisions in the free market, not in terms of causal agents, rules or programs, as the majority of scientists still believe today, but in terms of modulators i.e. distributed control parameters which are inducive to the emergence of population identity. Is it not remarkable, that all these independent writers in widely disparate fields all found the same bi-directional interaction between adjacent space-time fields? This bi-directional interaction between the scales has been described by several different names which I believe refer to the same phenomenon. ”Reentrant feedback,” circular causality,” “dialectics,” and “autocatalysis” refer to in many, albeit not all contexts, to the phenomenon I will call “mutual entrainment.” We will encounter this concept in many parts of this book.
Wikipedia defines entrainment as 1. flow of air bubbles during pouring of concrete; 2. brain waves. Bubble concrete example is an interesting metaphor as air bubbles have little to say where they move since they are stuck inside liquid concrete being poured on the sidewalk. Swarming insects are also “stuck” inside the swarm and are following the overall direction of the swarm, but unlike the concrete’s air bubbles, insects have a degree of freedom which bubbles do not. Individual insects constrain swarm just as much as they constrain each other. Insects are the swarm, and at the same time, the swarm is the insects. Seemingly, a single entity forms two different phenomena at two different spatial scales. To describe it, we say that two phenomena mutually entrain each other.
Neural scientist Gyorgy Buzsaki defines the mutual entrainment as “a measure of the stability of two or more oscillators that they would not have on their own. Mutual feedback is the key to entrainment of oscillators of various frequencies and stabilities. When multiple single-cell oscillators with different intrinsic frequencies are connected, they may produce a common intermediate global frequency. This frequency is not simply a linear sum of the frequencies because each neuron fires phase-locked to the global rhythm….The emergent population rhythm enslaves the behavior of individual units”. This formulation is identical to that of the phenomenon described by Haken for the laser.
In effect, Buzsaki states without perhaps realizing it, that mutual entrainment of micro and macro levels forms a field. He was not alone. With an open mind and open eyes, we can see mutual entrainment on all spatial scales. Here is a short selection of quotes from researchers in different fields, all documenting the mutual entrainment between neighboring spatial scales:
Electrons and Atoms — On the sub-atomic scale, Carver Mead concludes that “Properties of each electron depend on the state of the entire ensemble,” adding, “the mechanism for initiating an atomic transition is not present in an isolated atom; it is the direct result of coupling with the rest of the universe”.
Bacteria and Cells — Evolutionary biologist Lynn Margulis describes how the populations of interacting bacteria deposited their separate DNA into a common nucleus and through a phase shift formed higher life form, the eukaryotic cell.
Cells and Organisms. Biologist John Tyler Bonner describes phase shift from cells to organisms through primitive chemical signaling within a population of unicellular organisms such as slime mold. The “Control system” for slime mold is literally outside of bodies of the individual amoebas. Multicellular organisms exhibit mutual entrainment too. Gerald Edelman writes: “As cells bind into specific collectives during ontogeny, the binding itself changes the form of the cell, by signaling back to the genome (phylogeny). The interactive cycle is established where the alteration of CAM binding alters morphology, and where changes in morphology alter CAM expressions.”
Neurons and Brain. Neural scientist Walter Freeman discovered the mutual entrainment in neuron populations in the brain. Instead of static structures such as those assumed in the prevailing cognitivist view of Hebbian assemblies, he sees “Sensory stimulation, neuromodulators, and growth causing neurons to come together and form a mesoscopic pattern of activity. This pattern simultaneously constrains the activities of the neurons that support it.”
Organisms and Swarms. I have described the mutual entrainment in the movement of a swarm of insects, a school of fish and a flock of birds. Psychologist Kurt Lewin: “The person and his environment have to be considered as one constellation of interdependent factors. Life space is a field of forces — attractors and repellers with varying strength”. On the temporal scale, J.J. Gibson sees the mutual entrainment in the perception and action: “We must perceive in order to move, but we must move in order to perceive.”
Language and Consciousness. For a philosopher George Mead and psychologist Julian Jaynes, language emerged before the human consciousness and made the emergence of consciousness possible. For social communication, David McNeill, inspired by Vygotsky, observed: “Gesture output abruptly increases between three and four years, and speech and gestures become inseparable. This is a reentrant process. First, the abundance of gesture makes dialectic more accessible. Second, dialectic itself can cause the upsurge”. Psychologist Esther Thelen claims convincingly that “Perception modulates language and language modulates perception. It makes no sense to ask whether one determines the other.”
Individuals and Societies. Even the structure of society itself is not fixed by some external factors; rather it is a dynamic process on two scales. Sociologists Berger and Luckman in Social Construction of Reality write: “By playing roles, the individual participates in a social world. By internalizing these roles, the same world becomes subjectively real to him. Society is a human product, and at the same time man is a social product”. Edward Bernays writes in Propaganda: “The interests of the railroad and the communities through which it passes mutually interact and feed one another.” John F Kennedy also weighed on this issue: “Ask not what your country can do for you, ask what you can do for your country.”
Transactions and Economies. Austrian economist Ludwig Von Mises was critical of Keynesian economists focusing solely on a macro level of government and overplaying the role of central regulators. He argues instead that: “Common man …is like a sheep in the herd…yet the common man does choose. Society is not merely an interaction. It always involves men acting in cooperation with other men in order to let all participants attain their own ends.” Not only would Burke agree, so do smart investors and economists. George Soros does: “Not only do market participants operate with a bias, but their bias can also influence the course of events.” Alan Greenspan concurs: “Stock prices are not merely a leading indicator of business activity but a major contributor in that activity.”
Each of the examples above describes results of a lifelong search of some of the world’s most profound thinkers for the fundamental principles of sciences, particularly the sciences of life. Taken together, they form a compelling formulation of a new research paradigm. It turns out that “fundamentals” of life are found in relations between the “elements” of life, rather than in the elements themselves.
Limits of Modularity
Although we typically interact with the nature on the scale of centimeters and meters, we have no problem extending the spatial range with a microscope to millimeters and micrometers and with a telescope to miles and light-years. While Lakoff’s “base level interaction” of folk categorization agrees rather well with scientific categories, as we move up and down the spatial scales, science increasingly takes over and creates a plethora of specialized domains which are not readily accessible to common man; physics, chemistry, biology, sociology, or astronomy.
In those domains, we listen to expert scientists accredited by specialized social institutions to be the trust bearers for the rest of us. Their words are the science itself in our eyes, and we gladly submit to their judgment. Over the centuries, these learned men divided nature into many slices, starting with subatomic particles, atoms, and molecules, to cells, organs, organisms, societies. While it is true that the neighboring science disciplines sometimes overlap, as in the discovery of DNA, it is also true that the chemistry and physics use different languages, organize different conferences and form different cultures, sometimes fiercely clashing with each other.
__________________
It is equally true that much of science is based on the reductionistic principle. Cells are comprised of molecules in specific structural configurations; molecules are comprised of specific structures of atoms; sentences are comprised of words adhering to a particular syntax, and so on. Each time someone reduces a complex phenomenon into simpler components connected with a particular structure, he/she gets awarded a prize, and we chalk up the achievement on the altar of never-ending scientific progress. But how do we decide where to make the boundaries between different levels of nature?
Humans love containers. They are everywhere we look, starting from the envelopes, mailboxes, trashcans and shipping containers which revolutionized oversea transport, to biological cells, encapsulated software modules of object-oriented programming and language container metaphors such as “He is in love,” which are fundamental to everyday speech according to George Lakoff.
Containers are also called modules, and they simplify our everyday life. They are only accessible through public, standardized interfaces, so for example mailman can deliver your mail without invading your privacy. Large software projects involving hundreds of developers would not be feasible if everyone had to track changes in everyone else’s code, instead of just a few interfaces of modules directly relevant to their work. Cognitive scientists love modularity as it allows them, just like it allows software designers, to break a complex problem into a bunch of simpler ones and link them together into an encapsulated structure using Fregean compositionality principle. Next, the scientist or the designer writes a bunch of rules to put inside of each box, and if external interfaces connecting the boxes are chosen well, the whole system works.
Fodor developed a modularity thesis into the architecture of cognition. He concluded that cognitive mechanisms that mediate visual size perception must be informationally encapsulated. His theory was widely accepted, and much empirical work has been directed to researching the modularity of human perceptual subsystems in the mature language and mature vision. Unlike Fodor, Thelen was primarily interested in the dynamics of development of cognitive functions. She admitted that “informationally encapsulated” units may emerge with learning, but she pointed out that in a live system there must be deep correspondences between modules. For example, the long necks of giraffes, which are adapted for eating, require special cardiovascular support. It is not enough that each piece does its job well; they must fit together. Nature of one piece constrains the others, and because of that, there will be developmental interactions between the modules. While the modules may seem stable at a particular time, over the developmental time the internal structure of modules, even their existence, may change as a product of interactions in the whole organism. For example, larvae may be very different from an adult organism. There can be no complete encapsulation in a developing dynamic system because there is no cognitive theorist or designer in the brain to draw a new box or write a new rule.
In social life, modularity is hard to enforce by a new law, as can be seen from the politics of immigration everywhere in today’s world. Walls Street bankers, on the other hand, striving for global order, are very keen on dissolving containers on many spatial scales, from the scale of family and tribe to the scale of nations. There are always dependencies between the modules such as the above case of giraffes neck and heart. In the case of manufactured artifacts, some years ago Consumer Report found out during annual testing of cars that an 8-year-old $20,000 Toyota was more reliable than a brand new $50,000 Mercedes, even though Mercedes components were individually more reliable than corresponding Toyota components. Japanese engineers take a more holistic, “mutually entrained” view to the design process, including living on the factory floor and frequent visits to customers.
In Thelen’s view, during the development, anything can influence anything else. In dynamic systems, order, discontinuities, and new forms emerge precisely from the complex interactions of many heterogeneous forces. Thus, the power source of human cognitive development is not located in the separate modules, but their mutual interaction.
In summary, on a macro level looking from above, we see fixed structures, particles, solids, modules, grammar, symbols, and rules. However, looking from below, they are all dynamic fields of micro-level elements. They are not hardwired; they are just temporary assemblies and swarms of interactive populations, whether they are molecules, cells, organisms or societies. Actual separation of levels is entirely the product of human imagination, a sort of useful illusion. As Anaxagoras said 2500 years ago, modules and their boundaries are all products of the human mind.
The emergence of New Phenomena
Surprisingly, most scientists think they are just discovering the objective structures “out there” in the world, rather than considering the possibility that they are actively creating these boundaries themselves, according to their own beliefs and adherence to beliefs of their colleagues via imitation. Some scientists indeed reject reductionism between these levels and argue instead that individual scales can be studied in isolation. Fodor, in particular, contends that there is nothing to gain by reducing a macro phenomenon to a micro level. He is right, but then he proceeds to argue for the autonomy of individual scales, in his case the linguistic scale. Chomsky adopted this approach and concluded that neural science could not possibly contribute to linguistics. David Marr uses the same reasoning to conclude that high-level computer vision does not need to be “reduced” to neuron level operations in the brain. The problem with this view is that it replaces reductionist boundaries with phenomena wholly isolated from each other. In my view, neither reductionism of Chomsky nor parallel modular theory of Fodor and Marr explain the emergence of new structures in nature. To come up with a better explanation, we need a more drastic change in our outlook.
It takes courage to question the accepted wisdom. Esther Thelen was brave to argue that the “power of the explanation is in the dynamics of micro and macro processes, in the view from below examined from above.” Thelen believes that perception and language simultaneously influence each other. Her position brings her close to Whorf but with a significant difference. In her view, language does not directly “determine” perception, but rather language and perception influence each other through continuous interaction. Using the terms of mutual entrainment, we see the language and biology as two neighboring spatial scales, where language emerged at some point out of biology via phase shift when biology reached criticality during the evolution of human brain and formed Wernicke’s and Broca’s areas in the left brain hemisphere. Similarly, in the economy, Von Mises, while still working within the dualist framework dared to reject top-down reductionism of Keynesian economic theory. Friedrich Hayek, working at the philosophy department of Chicago University, abandoned dualism and completed Mises work. Curiously, even after he earned Nobel prize for his work, School of Economics at Chicago University did not allow Hayek to teach their students within their hallowed classrooms.
Cartesian dualists explain the creation of new phenomena as a Fregean composition. They effectively reduce the macro level object to its components and rules of assembly (syntax) which encapsulate this assembly into a module. If the Fregean composition explains how we create the new machines and the new computer programs, presumably the same Fregean composition can explain how the new molecules, new cells, new organisms, new songs, new stories or new currencies are created in the same manner. But do they? Let us turn our attention to the emergence of novel organisms.
If one thinks that the brain’s internal “representations” “mirror” already existing and timeless structures in the “objective” world, one is more likely to focus on the static structures inside the brain. It is this dualism that impels the AI researchers to see the static symbols being sent as “messages” through the conduit of the static axonal network. Walter Freeman, on the other hand, focuses instead on brain function, which forms a dynamic e-field, possibly “ignited” by the static neural network, but which forms a new phenomenon, just as “real” as the neural network producing it. In Freeman’s view, live, working mammalian brain forms and maintains an ephaptic e-field, not so different from a large flock of starlings circling overhead. It is this field, rather than the underlying neural networks, which is essential to understanding the working brain. We thus arrive at a very different concept of novel phenomena, which are the product of subjective human imagination, rather than mere mirrors of “objective world structures.
Swarms, Flocks, and Schools
One afternoon, sitting in the garden, I noticed a large swarm of tiny flying insects. The swarm was quite visible against the background in the crisp, late afternoon light. Its motion was very intriguing, it formed a dynamic, always morphing shape, moving seemingly at random but with strange fluidity and rather beautiful grace of abstract dance. This “abstract dance” and “organic shape” was, of course, the active construction of my perception, no different the image of the rainbow I see after the rain or the reflection of a ship in the water. In each case, these “images” were actively constructed by my perception and may not exist for other people in different locations with different viewpoints. Merleau-Ponty would answer the proverbial question about whether a falling tree in the forest makes a sound when no one is there to hear it, with a categorical no.
For a few minutes the swarm was hovering there in the middle of my backyard, and then it disappeared. Now, I know that the swarm shape was composed of moving and interacting individual insects; there was no leader coordinating this undulating motion. The swarm shape was built bottom-up from a myriad of small movements of individual insects, all moving for a short time in the same direction, perhaps for mating or forming a feeding “wolfpack” or just being moved by a gust of wind. At the same time, it was clear that just as the shape of moving swarm was a result of all these interactions, so too the micro movement of individual insects was constrained by the overall macro shape of the swarm, like dancers in a crowd on a dance floor. One could say that micro and macro phenomena were mutually entrained, without any single central process “regulating” the behavior of swarm.
This kind of behavior is not unusual for a group of animals: ant swarms, with individual ants communicating chemically via their antennas, school of fish, using sensitive lateral organ, or a flock of birds using vision, behave similarly. My claim is that electron fields, neuropil, organisms, and market behaviors are not that different. We do not usually think about the ephemeral temporal swarms as “real” objects, let alone “organisms.” After all, they are not tangible, they do not have a purpose, and they dissolve just as readily as they assembled into brief macro existence, much like cloud formations in a storm. It would be hard to believe that during its existence, a swarm is an animal, say like a dog. However, what if there is more to swarms than meets the eye?
Slime Mold
Then there is slime mold comprised of individual amoebas, which spend part of their life cycle in a multicellular form, bound by a sheath. When the food is abundant, individual amoebas exists as a single-celled organism and feed individually. When the food is depleted, however, some of the amoebas will emit pulsating puffs of a simple cAMP molecule to form gradient signaling to the other amoebas — the message being “come to this place.” The other amoebas move towards it and at the same time release a puff of cAMP of their own. Eventually, the population of amoebas congregates towards the center and starts moving as a single body, a multi-cellular slime mold slug. In this form, the population of amoebas is sensitive to airborne chemicals and can better detect food sources. It can readily change the shape and function of its parts and form stalks that produce fruiting bodies eventually releasing spores that become single-celled amoebas, thus completing a reproduction cycle.
Bonner points out that if the amoeba moves near the apical hot spot, they are stimulated to secret more cAMP creating a self-feeding, autocatalytic process. Remarkably, the gradient is maintained in the surroundings of the cells and not within the cells themselves. Thus, the motor control of the slug, its “neural system,” is literally outside of the bodies of individual amoebas. The resulting fruiting body is a sort of a “frozen swarm,” without a sharp boundary, with external gradient shaping the movement of individual amoebas.
In Donner’s view, cAMP in slime mold and chemicals surrounding ant swarms can be considered autocatalytic. The difference between slime molds and ants seems to be that the amoebae are always bound by a sheath, while the ants roam free. That is precisely why we classify slime mold as a single animal. What if this difference is, however, more one of a degree rather than one of a kind?
Sea Pens
Next, there are Sea pens. Sea pens live at the ocean bottom and look like beautiful underwater flowers.
Sea Pens are colonies of polyps. Each “individual” Sea pen is comprised of about 2000 polyps who spent about 15 years together and in the process are rather specialized to provide several important functions for the colony, like a feeding polyp, the primary polyp, and siphonozooid polyps on the stalk. Wikipedia classifies them as “colonial marine cnidarians belonging to the order Pennatulacea. It seems Wikipedia has no problem in considering Sea pens regular, ordinary animals, even though they “really” are consortia of polyps.
Placozoa
Trichoplax adhaerens is the best-known member of the phylum Placozoa, one of the earliest-diverging metazoan, i.e. multicellular phyla. Trichoplax is a small, disk-shaped animal with 1mm diameter that glides on surfaces in warm oceans to feed on algae. Like the single-celled amoebae, which it superficially resembles, it continually changes its external shape. Trichoplax lacks tissues and organs; there is also no manifest body symmetry, so it is not possible to distinguish anterior from posterior or left from right. It is made up of a few thousand cells of six types in three distinct layers: dorsal epithelia cells and ventral epithelia cells, each with a single cilium, ventral gland cells, syncytial like fiber cells, lipophils, and crystal cells. Lacking sensory and muscle cells, Trichoplax moves using cilia on its external surface.
Between the two layers of cells is a liquid-filled interior space, which, except for the immediate zones of contact with the ventral and dorsal sides, is pervaded by a star-shaped fiber syncytium: a fibrous network that consists essentially of a single cell but contains numerous nuclei. The individual fibers can relax or contract and thus help determine the animals’ shape. In this way, the fiber syncytium may assume the functions of nerve and muscle tissues.
There are no neurons present but in the absence of a nervous system the animal use short chains of amino acids known as peptides for cell communication, in a way that resembles how animals with neurons use neuropeptides for the same purpose. Individual cells contain and secrete a variety of small peptides, made up of between four and 20 amino acids, which are detected by neighboring cells. Each peptide can be used individually to send a signal to other cells, but also sequentially or together in different combinations, creating a vast number of different types of signals. This allows for relatively complex behavior such as crinkling, turning, flattening, and internal churning.
Roundworms
As we step up the evolutionary ladder, we progress to roundworms. C. Elegans are very well studied; indeed they became a sort of test bed for evolutionary biologists. C. Elegans look like “real” animals with “real” organs. Remarkably, every worm grows in essentially the same way — from a single cell, the fertilized egg, to an adult containing precisely 959 somatic cells. Out of these cells, 302 are neurons forming 8,000 synapses, but all cells form a well-defined structure of differentiated organs.
The motor behavior of the worm is particularly interesting, given the animal’s fully mapped nervous system. However, the worm’s locomotion dynamics derive, not only from the neuromuscular and chemical control systems, but from mutual entrainment of internal control and the physical properties of the body, and the environment (water, dry surface) in which it moves. More on that in Chapter 7 focused on the evolution of the brain.
I presented several intermediate steps between temporary swarms and fully multicellular persistent animals. But where is the boundary between an “ephemeral” swarm and a solid, “durable” animal like a worm or a dog? Lynn Margulis is convinced that all organisms are temporary “consortia” of smaller life forms. Who decides the “proper” time scale for the form duration? On a closer look, there is a continuity where before, it seemed, were clear boundaries.
However, there is an even bigger issue here. What is the “structure” of a swarm? It certainly can not be a Fregean structure defined by the reduction of a macro-level object to its components and rules of assembly. Language sentences analyzed with Chomskian grammars or Hebbian neural networks in cognitive science are pressed into such a static network structure. But generative grammars and cognitive science approaches did not produce good working systems. In these networks, the symbolic “information” flows only in one direction, between the micro level “elements” which essentially entrain the macro level. In reality, these “swarms of life” forms are entrained in both directions, from the bottom up and from the top down, all at the same time. Freeman calls this phenomenon “circular causality.” I call it the “mutual entrainment.”
Field Continuity and Boundaries
Continuity of fields does not mean there are no abrupt transitions. Water, ordinarily liquid, “suddenly” changes its state into ice below the freezing point and gas above the boiling point. Electrons “suddenly” jump to a new orbital. Neural scientist Walter Freeman and physicist Carver Mead claim that these transitions, while very fast, are nevertheless continuous and can be better described as phase shifts of large resonating assemblies. Carver Mead does not think that electron transition is discontinuous; it only seems so to us due to self-reinforcing, non-linear nature of two small dipole resonators. Quantum jump may be fast, but it is continuous. Mead views nature as being continuous both in time and space; in his view electron is just a wave, never a particle. Quantum mechanics, universally accepted at universities today, postulates, on the other hand, that electron is simultaneously a solid particle and an electromagnetic wave. This duality is ultimately grounded in the Cartesian dualism of body and mind. Einstein never accepted this duality, and for that, he was criticized by the Copenhagen clan. Younger Schroedinger fared worse as the entire generations of young physicists were subjected to the harsh bullying tactics of Niels Bohr, Werner Heisenberg, and John Von Neumann. Not until the last year of the 20th century did Carver Mead muster the courage to notice that the emperor had no clothes:
“One must solemnly affirm one’s allegiance to the Quantum God before one may be admitted to the physics clan. It is my firm belief that the last seven decades of the twentieth century will be characterized in history as the dark ages of theoretical physics.”
Similarly to Mead, Freeman describes gamma wave packets generated in the cortex as a gas to liquid transitions of a large interacting population of resonating neurons. In their 2006 paper, Freeman and Vitiello started with observation based on Freeman’s experiments in which he recorded EEG waves:
“The dominant mechanism for neural interactions by axodendritic synaptic transmission should impose distance-dependent delays on the EEG oscillations owing to finite propagation velocities and sequential synaptic delays. It does not.” In plain English, this means that unlike a computer, which consists of billions of discrete, digital circuits sending bits of information to each other, human brain functions on a very different principle. It’s 200 billion neurons in neocortex considered by traditional cognitive science to be transistor-like switches, could not possibly account for the incredibly fast spread of EEG oscillations in neocortex if they were “sending” information to each other via rather slow axons in the form of an action potential.
The answer to this puzzle may be provided by quantum field theory, which is rather similar to Carver Mead’s Collective Electrodynamics, both of which differ drastically from the quantum mechanics. Freeman speculates that his observed wave packets act as a bridge from quantum dynamics at the atomic level through the microscopic pulse trains of neurons to macroscopic properties of large populations of neurons. The wave packet is a collective mode that sustains a field of neural activity in the form of a boson wave, i.e. Bose-Einstein condensate. This approach is radically different from the traditional view of the brain as a “digital, information processing” system. Physicist Vitiello shows that in addition to neuron fields, other examples of macroscopically observed patterns are phonons (elastic waves) in crystals and the magnons (spin waves) in magnets. Freeman and Vitiello believe that cortical boson condensate, or its wave packet, ignited with collective e-fields emanating from neuron ion channels acting like electric dipoles, may explain the rapid course of perception. Their theory explains the fact that neocortex can respond to the impact of photons from a face in a crowd on a handful of retinal neurons mixed among many impulses elicited by light reflected from the crowd. The phase transition from a disorder to an order emerges suddenly: the neural vapor as it were, condenses into neural droplets, the first step in recognition, within a few tens of milliseconds, which is insufficient for classical models. It also conforms to the decades-old dictum of Wolfgang Kohler: “A theory of perception must be a field theory.”
Boson condensate forms a field which enables an orderly description of phase transition that includes all levels of macroscopic, mesoscopic, and microscopic organization of the cerebral patterns that mediate the integration of animal with its environment. Microscopic levels go down to electric dipoles of a myriad of proteins, amino acid transmitters, ions and water molecules that comprise the quantum system. This hierarchical but non-modular system, extending from atoms to the whole brain and outwardly into the engagement with the environment in an action-perception cycle is the essential basis for the emergence and maintenance of meaning during the development through successful continuous interaction within the environment. For this process to work, it is crucial that neighboring micro-meso and meso-macro levels of the brain influence each other bidirectionally and concurrently.
To sum up: Fregean compositionally does not work outside of Dualist theories. We already defined the mutual entrainment as a bi-directional dependency between the micro and macro levels of the same phenomenon. It is also implied by the recognition of live organisms as open systems. All animals exchange thermodynamic energy with their niche, which is called context in the case of human behavior. All phenomena are relative to the activity and interest of the observer and, like Wittgenstein’s language games, they have no hard boundaries.
Criticality and Phase Shift Form New Phenomena.
For a phase shift to occur, a system must reach criticality. Criticality can be seen everywhere in daily life, at all scales. We can see it in microscopic electron transitions between orbitals, and we can observe it on a macro level as water suddenly freezes at zero degrees Celsius. While temperature drop approaching zero is continuous, the behavior of water is dramatically discontinuous. Esther Thelen describes these nonlinearities, or phase changes, as highly characteristic of non-equilibrium systems. In her 1994 book, A Dynamic System Approach to the Development of Cognition and Action, she quotes Farmer and Packard: “Adaptive behavior is an emergent property which spontaneously arises through the interaction of simple components. Whether these components are neurons, amino acids, ants, or bit strings, adaptation can only occur if the collective behavior of the whole is qualitatively different from that of the sum of the individual parts. This is precisely the definition of nonlinear”.
As Prigogine noted, most theoretical systems studying change have a teleological core. This approach presumes the end state before the developmental process even began and thus negates the “arrow of time.” Both he and Thelen wanted to study systems that “change over time, where novelty “emerges,” and the end state is not coded anywhere and where behavior at the macro level can, in principle be reconciled with behavior at the micro level.” Their approach is different from the reductionist approaches grounded in Cartesian dualism, which seek to find the essence of a system in a unique and privileged component of that system, such as Federal Reserve in the US economy. Examples of such complex emergent systems are weather, laser beams, liquid flow patterns, slime molds, nerve impulse patterns, heart rhythms, perceptual systems or economic patterns. All these systems are “open systems,” existing in a “far from equilibrium,” yet stable states. Far from equilibrium conditions can be maintained only by the continuous flow of free energy into and out of the system. These emergent organizations are different from the elements that constitute them. Details of one such a system are described by Herman Haken, an inventor of laser: “In an ordinary lamp, electrons of individual atoms make their optical transitions independently of each other, but in the laser, the electrons make their optical transitions cooperatively. Such cooperative activity can be brought about by external orders, but in the laser, there is nobody to give orders. Thus the natural cooperative behavior of the laser is an act of self-organization”. Incidentally, two high priests of quantum mechanics, the physicist Niels Bohr and the computer scientist Von Neumann, argued against the possibility of the laser until the very day Townes showed them one in operation. Haken’s genius was to understand how certain collective actions of individual elements increase until they appear to dominate and govern the behavior of the whole system. Haken called these parameters “order parameters” as they allow to describe the system with a few collective variables, rather than by many individual elements. New patterns fall out strictly as a result of the interactions among the elements that compose the system, the constraints on the system and the energy flux. No separate “central controller” or “master plan” is needed.
As a psychologist, Esther Thelen studied learning of motor activities, i.e. how children learn to walk or grasp objects. Her discoveries dismantled the old concept of development based on pre-existing “central pattern generators” which were presumably encoded in DNA and replaced it with a novel, more dynamic approach. In her study of children’s acquisition of reaching proficiency, Thelen demonstrated that each infant acquired reaching proficiency by solving a different set of problems. She illustrated how infants limbs exhibit remarkable self-organizing patterns that emerge from the physical characteristics of bones, joints, and muscles and the energetic and metabolic characteristics of an excited baby. Thelen quotes from Kurt Lewin, an important system theorist whom we will encounter again in Chapter 7: “the person and his environment have to be considered as one constellation of interdependent factors,” i.e., an open system. Any physical setting acquires meaning to the individual only as a function of the state of the individual in that setting. Lewin characterizes the life space as a field of forces — attractors and repellors forming a landscape which our life journey both forms and traverses at the same time.
Thelen was very impressed with Haken’s concept of order parameters. “What is remarkable that parameter change (temperature in this case) is entirely nonspecific to the actual behavior of the system. While attractor causing the behavior was assembled through slaving of its order parameter, there are no codes, prescriptions, schemata or programs orchestrating the nature of the attractor. In biological systems, any number of external variables can act as control parameters; for example in photosensitive animals light intensity determines locomotor vectors. The energy level is another common control parameter, as can be seen readily from the gait of horses. As horse continuously increases the speed, its gait shifts discontinuously from walk to trot, to a gallop”. Oxygen gradient or gravity gradient are additional examples.
Thelen generalized her research from motor activities to all cognitive activities and, much like Merleau-Ponty proposed that symbolic thought itself is emergent in an activity and the products of those activities in the physical world and on us. In that, symbolic thought is no different from interpreting novel words or walking up and down slopes. Thelen: “As we act and explore our world, our actions will produce concrete physical changes in that world that we may perceive. By perceiving the products of activity and the reentrant mapping of that perceiving onto the ongoing activity from which it emerged, we create external symbols and discover their value. Development does not unfold according to some pre-specified plan. There is no plan. Developmental change is caused by the interacting influences of heterogeneous components, which are not encapsulated modules; indeed development happens because everything affects everything else”. But time-locked patterns of activity across heterogeneous components are not building “representations of the world” by connecting temporary contingent ideas. “Indeed, we are not building representations at all.”
Contrary to Thelen’s view, the overwhelming majority of cognitive scientists and AI scientists love representations, and without the slightest hesitation, they concluded that brain “must be” a larger and more complex version of digital computers invented in the second half of 20th century. This conclusion, made without any prior epistemic analysis, is simply taken for granted, as chapter xx quotes from a well known AI researcher, Jerry Feldman show rather clearly. My own experience of this happened in 2015 when I talked in person to Ilya Sutskever, a Google Brain scientist and a student of neural network guru Geoffrey Hinton. I approached Ilya at one of Stanford University VLAB meetings, attracted by his T-shirt with a large picture of “brain as a computer circuit.” This image, quite popular on the web and also shown in the middle of the picture below, was proudly displayed on Ilya’s chest, so I asked: “Are you sure that brain looks like that?” — “Completely sure” was Ilya’s quick answer. I did not give up: “If you looked at any neural science 101 college course, you might learn that it may not be the case”, receiving a quick retort again: “Why would I do that? There is nothing I could learn from neural science”. He then added: “Let us just agree to disagree” and walked away.
This reaction I knew only too well, the vast majority of computer scientists today simply assume that brain “must be” a glorified computer, much like 400 years ago Rene Descartes contemporaries assumed that brain was a clock-like mechanism, replete with spinning cogwheels found in mechanical marionettes of the day. The same mechanical metaphor can be found in the early attempts of Charles Babbage and Countess Ada Lovelace at building a “difference engine” and Alan Turing’s early effort at ENIAC, a mechanical precursor of modern digital computers. The reason this assumption comes to computer scientists so naturally is not just theoretical, even though CS curriculum at Western universities today is based on a hardcore set of Cartesian beliefs. More importantly, young computer scientists today grow up in a Heideggerian world they are “thrown” into, without having much choice in the matter. Today, this world is inhabited by the population of millions of Von Neumann’s computers, most of them built by Intel. By programming these machines daily from their childhood, computer scientists constructed, in Berger and Luckman sense, an “external” world into which they later externalized themselves. In the final step, they merged their humanly constructed world with the world of nature, reifying it into “objective reality,” forgetting that they and their fathers created it in the first place.
Is there an alternative? If the computer scientists metaphor is misleading, what is a better metaphor for our brain?
In the image above there is a modern neural network on the left and its crude expansion into AI view of the brain, depicted on Ilya Sutskever’s T-shirt, so irresistible to Silicon Valley culture, in the center. On the right, there are two graphs. On top, there is the time graph of values of the harmonic oscillator and below the same for chaotic oscillator based on Leon Chua’s nonlinear Double Scroll attractor. Both right side images illustrate not the static circuits, but the temporal, dynamic fields. I argue in this book that these fields are not only far better models for the understanding of the human brain; they are also far better models for the understanding of the modern physics, chemistry, biology, linguistics, and even the economy and culture.
In the realm of intentional human behavior, neuroscientist Walter Freeman described the brain as a dynamic, far from equilibrium, open system. “All that brains can know has been synthesized within themselves… Every neuron participates in every experience and behavior… Perception and action result in global amplitude modulated pattern, as the cooperation carries the entire hemisphere from one global chaotic attractor to the next”. Freeman concluded that consciousness is not an organ, i.e., some part of the brain, rather it is a higher level of self-organization. Social self-organization is an extension of the micro-meso interactions we saw between neurons and populations, and between meso-macro populations and global AM patterns. At each level, the individual retains autonomy but accepts constraint with respect to the embedding surround. “In human societies, we experience empathy; we confirm our solidarity by shared gestures and words of joy, threats, and invitations to cooperate. We respond by making ourselves similar to each other using mirror neurons in the process of assimilation”. George Mead would expand this assimilation into the emergence of language and its companion, self-awareness, resulting in the emergence of “I,” “me” and the world of culture. In the world of economics, Von Mises and Hayek understood this dynamics of life very well, unlike Keynesian economists, whose static systems are striving for equilibrium. Federal Reserve misguided attempts to eradicate “income inequality” will result in the inevitable crash, an equivalent of death in living systems. 21st-century economics may yet vindicate Austrian economic theory as well as Freeman’s concept of the brain as a field of forces produced by mass action of a vast population of neurons in neocortex, cerebellum and limbic system. Evolution on all scales is an ever-expanding spatial and temporal progression of dynamic fields, resonating with each other at the same micro-level scale, mutually entrained between neighboring scales and forming, via phase shift, novel, macro-level phenomena.
