The Trouble with Complex Adaptive Systems
Complexity in the natural world has always emerged via the process of evolution which is a massive, parallel search process that explores “design space” through trial and error and, because of the massiveness of the search mechanisms can afford to make many mistakes without the whole enterprise coming apart. As I wrote back in June, 2011, evolution has a trajectory in the sense that as long as free energy flows through the Earth system new more complex organizations will emerge and be positively selected. Reduce the free energy flow and the process will go into stasis or possibly a steady state equilibrium where no novel structures will emerge. Cut it back enough and evolution becomes devolution. We've never really experienced either of those situations on Earth thus far so this prediction comes from computer modeling of dynamical systems.
Evolution has produced a wealth of complex adaptive systems, e.g. plants and animals with huge numbers of component cells, tissues, functions, behaviors, you name it. Ecosystems are extraordinarily complex and are constantly subject to fluctuations in structures and organization. Biological units, like species or populations are adaptive over a sufficient time scale by virtue of many levels of adaptive mechanism. Over the longest time scales species are adaptive evolutionarily. On much shorter time scales, individuals may be physiologically adaptive to short-term and small amplitude changes in local conditions. But all such systems are able to change their chemistry, form, and behaviors to some extend in response to changes in their environments. Back to the evolutionary scale, such adaptations may prove useful in terms of fitness maintenance, but as often the environmental stresses caused by change acts as a negative selection force.
In general, all adaptive mechanisms operate on what seems like a simple principle. When the environment changes it demands some kind of new response from the system. As often as not this requires an increase in the complexity of the system to meet new demands (or also take advantage of new opportunities in an altered econiche). In other words, systems adapt to change by increasing their systemic complexity, sometimes, as in the case of speciation, becoming wholly new kinds of systems. This pattern works quite well in the biological world where evolutionary process is the final arbiter of what works and is kept for further testing.
But evolutionary process works slightly differently in the human-built world, the world of artifacts and institutions where humans actively direct the nature of responses to the problems generated by a changed environment. What we mistakenly take to be a completely “intentional” process of “invention” and “design” is still really just a massively parallel search through what might be called the “artifactual” design space (see my previous essay: What is the solution to all problems in which I discuss design space and the evolution of the human-built world). The difference is in the way novelty gets generated from human brains rather than from genetic mutations. The time scale is vastly different and the scope of change can be too. Sometimes in a flash of insight a human inventor can create something truly unique (and complex) in a fashion reminiscent of the “hopeful monster” theory in biological evolution (technically called ‘saltation’). Humans are able to introduce novelty at such a rapid rate that it is sometimes not possible for the forces of social selection (e.g. market acceptance) to operate in the same time scale. Saltation in the human-built world can be overwhelming (take, for example, the case of financial derivatives invented in wild exuberance and foisted on the world with absolutely no kind of selective testing until they all went down in smoke taking the whole financial system with them — something which is still playing itself out.)
Complexity generated (by humans) to solve social problems is not always subject to the kinds of normal evolutionary testing that allows an orderly progression from simple systems to complex ones. The situation has been exacerbated by the advent of high technology, particularly in energy and information. The rate of change in our societies today is overwhelming creating greater complexity in shorter time frames than could ever be reasonably tested safely in pockets here and there (as is the case in biological evolution). Everything is connected to everything else so tightly today that the evolutionary tests of any complex arrangement is now all-or-none. If a complex solution arises and seems to solve the problems on a short time scale, it arises almost instantly everywhere. There is no safe sandbox to allow it to play in. It simply gets adopted everywhere on faith. And if it actually fails the evolutionary test of validity it will crash everywhere. Ladies and Gentlemen, I give you modern capitalism as exhibit one, on this count.
So here is the problem. Natural biological evolution works because there are loose couplings between subsystems (e.g. populations and species) and long time scales over which selection pressures can work. The media for conducting the massive parallel search is relatively speaking cheap (think how many offspring the typical invertebrate animal produces at one time because only a very small percentage of them are going to survive!) But we don't exactly think of human life as cheap, although I have to argue that many of us act as if we really do subconsciously. The costs of mistakes in the human-built world are hard to bear. They are even more so and more damaging since our global society is so strongly interconnected. What this means is that the latent costs inherent in greater complexity are more easily hidden in the human-built world and so human created complexities end up accumulating defects and lead, eventually to failures (see comments later about complex societies and collapse).
In natural biological evolution (and other kinds of auto-organizing phenomena) adaptive systems deal with the problem of latent complexity-caused errors by literally restructuring their organization to accommodate new requirements for survival. The evolution of the animal brain and its culmination in the human brain is an excellent case in point. If we understood what transpired in the evolution of unbelievably complex brains, we might gain some insights that would be applicable to human-built structures, in particular to the governance of human society.
How Brains Got So Complex, and Dealt with It (And Why Societies of Brains Can't - Yet)
Arguably the human brain is the most complex subsystem on this planet, and as a result makes all super-systems in which it is embedded super complex. It is composed of trillions of components of thousands of types and functions. It is seemingly infinitely flexible in terms of what it can encode in the form of knowledge which can be used to make the human organism the most adaptable, as a single species, ever evolved. The success of the human brain, in evolutionary terms, is attested to by the success in procreation of the species to date (past performance is no guarantee of future performance!) From our studies of complex systems, especially those created by human endeavors, we know that as complexity increases in system there are diminishing returns with respect to functionality and reliability. In fact, at some point increasing complexity generates greater costs than benefits. When systems reach this level something has to happen. Either the system collapses (or is abandoned, like the space shuttle or the supersonic airliner), or the system undergoes a kind of phase transition or reorganization adopting a radically new architecture that preserves its best functionality while allowing yet newer forms of complexity to emerge.
This is what has been the case for the evolution of the brain. Over evolutionary history the brain increased in complexity not by simply adding on new circuits, but by inventing (so to speak) whole new kinds of circuits of neurons (as well as new kinds of neurons) that transformed the architecture. This happened at relatively distinct points in time after which brains having the newer structures were able to continue to add complexity in the form of better more elaborate functions until the next 'crises' of organization approached.
The earliest brains were just globs of neurons that processed simple sensory inputs and generated relatively simple motor outputs. These were clusters called ganglia which can still be found in animals like worms. Such a control structure is strictly operational in nature. There are no higher-level coordination functions needed since the environmental niche of a worm is relatively simple and only demands operational responses. Insects went a step further in developing something closer to a centralized brain at the head end of the animal. These brains provided an additional level of logistical coordination for the operational ganglia (still present in the nerve chords. They also provided a bit of tactical control for things like mating behaviors which were more elaborate due to the more complex environmental framework that insects live in. Of course the logistical and tactical controls were primitive and genetically determined. The amount of memory and learning in these animals was still limited to the basic neural plasticity in simple synapses.
Primitive chordates essentially inherited this very primitive brain architecture. As animals evolved into slightly more complex environments, the brains, say of fishes, simply complexified accordingly without any real architectural restructuring. But when animals had been well into invading terrestrial environments something had to give. Animals such as primitive reptiles and their predecessor amphibians needed to have more complex behaviors for more complex environments (which, by the way, included more other species to deal with and more niche opportunities to take into consideration). These animals could not merely further complexify the basic architecture (operational and primitive, that is merely instinctive, logistical/tactical controls). Something additional was needed. A new, more flexible computational platform was needed in order to provide more adaptive behavior. Patterns needed to be encoded in response to actual experience because the genome could not possibly contain enough information to control the wiring of the brain for these non-stationary patterns. Those patterns were more complex, themselves, and needed more refined pattern recognition and association (to action responses). A whole new kind of neural tissue was needed and that was the cortical sheets that provide a matrix for encoding spatio-temporal patterns. The new sheets (with some newly modified neuron types organized in new, modular substructures) provided not only a pattern learning and memory recall system but a control enhancement system for the logistical, but mostly for the tactical control of the animal. The old brain structures were not thrown out so much as modified to integrate the capacities of the new structures. Cortices provide the computational architecture for creating adaptive maps1. And these were, in evolutionary terms, an instant success. Reptiles rapidly dominated the terrestrial environment and some aquatic ones as well. They were highly successful at exploiting complex niches.
Of course early mammals inherited this same architecture. But latter mammals were able to take it a step further. What we call now the archicortex (or sometimes the paleocortex) emerged in mammals and gave them an edge as the age of dinosaurs ebbed. Using the same matrix-sheet adaptive map approach, but now extended to wrap around all of the more primitive structures, this platform not only extended the flexibility of (especially) the tactical control of behavior it introduced the opportunity for long-term memory encoding of far more complex patterns. It also added the capacity to not just encode patterns and provide for rapid recognition from sensory inputs, it also allowed some primitive model building in which mammals could consider more than a few options in choosing behavior. Mammals became able to “think” in a primitive way. Mental models of the components in one's environment allow one to consider possible futures with respect to how those components are going to behave. Animals with this architecture added the capacity to anticipate alternate behaviors to a small degree.
What we have been seeing in the evolution of these architectures is a movement toward greater hierarchical management of an individual's behavior in order to produce greater complexity in that behavior to address the greater complexity evolving (co-evolving) in the environment of those animals. At each stage, brain evolution involved increasing complexity in its structure to match the increasing complexity (or rather the opportunity to enter more complex niches where competition would be, for a while, less) of the environment. And this was accomplished not by just adding more of the same structure, but by reorganization and adjusting the control hierarchy to match.
The human brain represents the most recent restructuring. Actually the process began in the primates with the emergence of the neocortex (wrapped around the archicortex). This layer of newer cells and new substructure organization expanded and deepened the thinking capacity that the archicortex had bestowed on lower mammals. In addition another dimension of organizational structure, the role of the frontal cortex, had been taking on an increasing responsibility not only for organizing rational responses to the environmental situation but for planning further into the future based on using those mental models encoded in the rest of the cortex (and now made tremendously more elaborate and broader in scope by the advent of the neocortex). The prefrontal cortex added something that went beyond mere tactical decision making. It added the first real form of strategic control for behavior. By allowing humans to make more elaborate plans and organizing more tacit (experiential) knowledge to provide higher order judgment shaping of intelligent decision making, the human brain had become a wholly new kind of computational system providing seemingly infinite combinations of concepts and thoughts.
Of course, the brain's evolutionary restructuring to manage complexity was not a wholesale reorganization of every part. The older parts were retained and still performed their original duties. The brainstem nucleus (like a ganglion) responsible for controlling breathing still does so automatically. But the new brain can override parts of the older brain when circumstances warrant. One can hold one's breath if one wants. The reorganization has been by increasing the hierarchical control structures and improving as much as possible the top down coordination where needed.
There are other examples of restructuring architectures in biological evolution in response to increasing complexity. Indeed, most tissues and organs show this same pattern. Hierarchical restructuring can be found everywhere in nature and so it seems that we should consider it as something like a law of nature in this regard.
The law might be generally stated thus: As complexity emerges in a system (to respond to complexity in the larger embedding system) a point is reached wherein the system must restructure to enhance a hierarchical control network in order to improve coordination and avoid the diminishing costs of increasing complexity.
If Brains Could Do It, Why Can't Societies Do It?
As I have mentioned several times Joseph Tainter2, and other scholars, have begun to explain the collapse of societies in history as a phenomenon of increasing complexity (or what amounts to the same thing as demand for resources that require more work to obtain) leading to a situation of diminishing returns and eventual negative returns. Today we seem to be witnessing the collapse of the global society and a great deal of blame is being placed on the intricacies of subsystems like the financial markets. We know, of course that the problem is that we don't have enough usable net energy to continue growing our human built world (at the sacrifice of the natural resource world). It takes energy to run complex systems.
The question we should ask is: Why can't societies successfully restructure to a hierarchical system as the brain did in its evolution?
Well, they try to. The auto-organization of human institutions actually does follow the pattern shown in other natural systems' evolution. Current governments and economic management systems are quasi-hierarchical control systems. The problem is that the major operative components - human individuals - are not sufficiently empathetic to form strong cooperatives. Competition is still too much a factor in the relations of individuals. Rugged individualism sings praises to this fact. In short, unlike neurons that are highly cooperative with one another, humans are simply too autonomous and have wildly different perspectives from each other to be successful cooperators!
Loose coupling between components with competition is good for exploratory search of the design space. It offers an impetus to find better solutions or novel channels for energy dissipation. It is a natural and necessary stage of development of a complex system. But at some point, the need for efficient management of complex systems outweighs the need for exploration and the role of competition and novelty must be subjugated to that of cooperation and exploitation. If a complex dynamic system is to achieve stability the components must be able to form stronger linkages based on cooperation. The need for a hierarchical control structure overrides the need for self-interest.
“Freedom” is a luxury of non-complex systems in which the interactions between components are not so strong3. Individuals have more leeway in their choices of actions because the coupling strength between them is not so great that disturbances in any one relation will propagate destructively through the network of relations. As long as the component density is low and there is energy to use to redistribute component locations, the system can afford to have non-conforming activities among components without disrupting the whole system.
But when the density of components is high, when the interactions between components becomes sufficiently strong, the freedom of individual components with respect to their choices of action become more highly constrained if the system is to avoid disruption due to the propagation of disturbances due to one component behaving in a manner adverse to closely aligned other components4.
These principles apply to all complex systems. They especially apply to human societies. We humans, as interacting agents are still too autonomous (intellectually and emotionally) to form effective complex societies that actually provide for the well being of all members. Our governmental systems are hierarchical as we attempt to organize in the fashion nature has provided. But our capacity to form strong cooperative interactions under the population densities we now have is very limited. We are still very much driven by our animalistic limbic, self-interest-above-all brains and our cooperative higher-order regulatory (executive) functions are too easily overwhelmed by those drives. When push come to shove we will compete and seek to dominate rather than take time to find ways to win through cooperation. We are still motivated much more by personal profit than collective sharing. The latter is in our character. We had begun to evolve this aspect more strongly at the end of the Pleistocene era and we can see it in our ability to form groups, organizations, and attempts at collective actions. But we also see that it was only a partial evolution when we witness in-group fighting, personality conflicts that hurt the group, back-stabbing for personal gain, etc. As evolution goes, we were on the path to become angels, but we remain mostly self-interested animals on the whole.
We try. We achieve some really amazing organizations for mutual benefit. We have gained enough collective experience to see how to form truly large societies, states, countries, even international organizations and trade. But, we also had a great advantage in this development. We were always finding new sources of higher powered energy with which to create so much material wealth that push never really came to shove all that often. It did, of course, come every once in a while. And great civilizations rapidly collapsed when there were temporary resource limits that triggered the competition mode of behavior. Resource envy powered territorial grabs (imperialism and colonialism) that eventually became too complex to manage relative to the actual inflows of resources (particularly energy) as Tainter has related.
Now that resource limits are operating on a global scale the triggering of competition will not be a temporary decline followed by an overall upward complexification of global society. This time the drive to compete rather than cooperate will operate on a global scale for all of the future — of this species. And for that reason, the members of this species who are incapable of transcending the natural urge to self-interest even in the face of global civilization collapse, will kill one another off.
But recall, I said humans had been on the evolutionary path to greater empathy and tendency to cooperate? The genetic propensity for some people to have much greater capacity for empathetic and cooperative behavior is still incipient within Homo sapiens even if it is weak in most members of the species. The laws of probability distribution of traits guarantees that there are some members of the population who are capable of forming truly cooperative organizations in response to the stresses of resource depletion. And those who are capable will do so. They will be able to let the laws of complexity reorganization (through hierarchical management) operate on them so that they can succeed in surviving the future challenging environment by cohesion and adaptation at the group level rather than at the individual level.
Understand that the mind that is more empathetic and cooperative is not that of a mindless automaton (nor the colony mind of, say, an ant or a bee). We are still talking about human minds that are capable of being cooperative but also understanding the importance of being so. These are minds that can still generate independent thoughts and ideas but which will also subject those ideas to consideration by the larger group before adopting an attitude of being “right” simply because they originated the idea. In other words these are minds that can think for themselves, but also recognize the importance of consensus in arriving at conclusions.
Sapience is, in part, the capacity of a mind to both have independent perspective and motivation to find cooperative ways to meld perspectives. Sapience entails brain structures that help reduce the impacts of biases and heuristic thinking that plagues ordinary Homo sapiens5. Sapience improves the use of experience based models (in tacit knowledge) to override the limbic system influence over decision making (what Kahneman calls the System 1 decision maker — fast vs. slow or deliberative decision making. See: Thinking, Fast and Slow, Farrar, Straus and Giroux, 2011.) Strong sapience may be able to effectively eliminate or at least greatly reduce the effects of biases by “juicing” up the deliberative system (system 2 in Kahneman's jargon - call sapience system 2½ or even 3!)
Humans as currently constituted (mentally) are not really great observers. We are not unbiased or objective. Our perceptions are filtered and distorted by our already encoded beliefs, hence most of us form non-veridical models of the world to various degrees. Some are better than others, but we all suffer from misperceptions and misconceptions that cause our perspectives on things to not only be the result of seeing phenomena from different angles, but from different a priori beliefs. This is what we mean by ideologies causing us to have a distorted model of reality. Humans were clever enough to invent science precisely because we recognized our own failings. Empirical methods replaced simple observation. Repeatability replaced one-off conclusion drawing. Mathematical models replaced foggy mental images. We did this because the smarter ones among us understood that mere opinions were totally unreliable when it came to predicting future outcomes in nature. Each of us sees the world from a different point of view, but also from a different filter of assumptions and beliefs. Sapience helps to counter this effect by making the bearer more aware of his or her shortcomings as an observer and concluder. Because we, as individuals can never be certain that our ideas are right sapience pushes us to continually question everything. And it pushes us to try hard to grasp another's point of view when they differ from our own.
Empathy is stronger in more highly sapient people. And empathy isn't just a matter of feeling someone else's feelings. It is also putting yourself inside their head to see the world the way they see it. Communications is everything. The more sapient expositor will choose words and phrases with deep and shared semantics. The more sapient listener sill attend to every such word and phrase to maximize the transference of information. A collective of highly sapient people will be capable of a higher level of efficacious communication than a rabble of ordinary sapients. We have a remarkable example of the effects (and failures) of the latter when we watch our American congress in action. Even weakly sapient people will hang their heads and weep for what this supposedly deliberative body of decision makers has become in light of modern social complexities.
Which brings us to the reason that human governance institutions are failing in light of the complexities of modern society. Our institutions are only partially formed in the manner of nature's way to restructure as hierarchical control systems. They succeed only to the degree that humans are a little bit cooperative (and a little bit sapient). But they fail to the degree that we are still hampered by faulty decision making and poor perceptions and conceptions of reality. The complexities of the world that we have created are acting to accelerate the downward spiral of cooperation through effective communications. Today ideologies trump objectiveness and the capacity to continually question our basic assumptions. Opinions trump science. And the results are now seen to be catastrophic.
Human beings are neither like neurons nor like ants. They are not driven by mechanical programs to behave in a cooperative fashion. They are and will always be thinking beings with very complex models of how the world works, encoded in their heads. The ordinary human has very limited models to work with, and not very veridical ones at that. Even our so-called leaders, today, are plagued by these limitations. But more sapient beings have better, more comprehensive and veridical models to work from. The highly sapient brain includes the ability to guide learning through life, to attend to what is important and true, to note and question deviations from the expected (especially that expected based on what authorities tell us). This is how wise people end up with so much good tacit knowledge as they age. Even a young person with high sapience may be foolish in some respects, may make some foolish mistakes, but their sapient brains learn from those mistakes. Wisdom grows over time such that one makes fewer foolish mistakes later in life.
Now imagine a society comprised of humans whose average level of sapience is at what today is the rare high end of the spectrum. Imagine the wisest among them many orders of magnitude more sapient than even the wisest persons today. What kind of society would they have? Sapience and wisdom promote cooperative, empathetic attitudes even when perceptions or points of view differ. Sapience promotes effective communications as people attempt to work out the differences and come to a common understanding. Sapience in individuals permits sapience in governance through nature's organization of hierarchical decision making. People can still be individuals and appreciate individual differences and still work to cooperate for the good of the whole. This is the wonderful new framework for a society based on cooperation, not because everyone thinks alike or acts alike, or holds the same set of conceptions exactly. This is no society of automatons. Rather it is a society of people who are even more human than we are! Still autonomous in thinking, but capable of understanding one another and what is the good for society. This is the path that Homo sapiens was on before the advent of agriculture. We can only see it now, in retrospect, as the antithesis of competition and stubborn individualism is amplified by the overly complex society we have created.
Evolution is the solution. Our evolution as a genus is not yet over. As with other evolutionary events throughout the Earth's history we are in for a tumultuous, even brutal, selection event. But the potential for a new organization of human society (and a new kind of civilization) awaits. Will it be realized? Present humans cannot hope to succeed in becoming more sapient simply by learning. The brain has to further evolve. Perhaps with the active participation of those who understand, evolution will follow our desired trajectory. A future form of Homo will succeed.
1 The first recognizable cortical structures have been termed the archicortex. These structures appear to have evolved from sheet-like structures called mushroom bodies in more primitive animal brains. The story I am weaving here is very much a simplification of the evolution of brain structures, but follows the basic form. In general evolution finds ways to reuse older structures for new purposes through modifications. Hence legs and arms were once fins. The general pattern appears to involve something like a mutation that generates redundancy in certain kinds of structures. The two structures then start to diverge in form and function until whole new structures and functions emerge. That is a marvelous story in itself!
2 Tainter, Joseph A., The Collapse of Complex Societies, Cambridge University Press, 1990.
3 Or, as Janis Joplin said, “Freedom's just another word for nothin' left to lose...”
4 Oh can't you hear it now? Communism. Socialism. Screams of outrage. But actually I doubt that any right wing or libertarian reader would have made it this far. If you will allow me to be somewhat presumptuous, what proceeded to this point is too intellectually dense for the average ideologue to have been able to follow and they will have clicked away to a more mentally favorable site by now!
5 The study of intuitions and judgment has provided us with a truly eye-opening perspective on human cognition that exposes the fiction of humans as rational deciders! See: Gilovich, T., Griffin, D., & Kahneman, D. (eds.) Heuristics and Biases: The Psychology of Intuitive Judgment, Cambridge University Press, 2002.