Swarm Intelligence and Dynamics
The article below contains excerpts from L.K. Samuels’ new book, In Defense of Chaos: The Chaology of Politics, Economics and Human Action.
Column by L.K. Samuels.
Exclusive to STR
Intelligences take many forms, but the particular ways in which social animals—ants, bees, wasps, termites, birds, fish, and so forth—network into highly coordinated units without supervision has always mystified and fascinated scientists. As it turns out, the unfolding patterns of insect and bird behavior have provided far greater scientific insights than a mere collection of dead bugs in a display case could.
Despite their interest in the behavior of insects, early researchers were not too keen on studying the bugs up close. So when scientists grudgingly pulled out their magnifying glasses to scrutinize the world of “swarmettes,” it was to improve the properties of artificial intelligence. Originated in 1989 by Gerardo Beni, Suzanne Hackwood, and Jin Wang, the field of swarm intelligence sought to study self–organizing agents in cellular robotic systems. The objective was to design unintelligent robots capable of performing intelligent tasks, inside swarmlike groups. The discipline soon gained popularity and became an integral part of complexity science.
Eventually, swarm researchers took a serious look at the creatures they were trying to emulate. They found that the science of swarm intelligence offered a powerful computational methodology, something that could compare the actions of animals, humans, and robots to how decentralized, self–organizing structures emerge. For instance, computer models based on swarm dynamics have predicted the behavior of vehicular traffic and the movement of pedestrians on streets by looking at how army ants conduct raids. Similar studies showed how communal goals are achieved by purely decentralized means, where dumb individuals cooperate to accomplish complex, difficult tasks. In other cases, scientists had found that swarms engage in “stigmergic assemblies,” a condition in which participants communicate with each other mostly by indirect means. Swarm intelligence was getting complicated.
Complexity With No Central Processor
The pressing question for early chaologists was obvious: How could mind–numbing complexity arise from social species imprisoned by limited intelligence and capabilities? After all, insects are barely conscious. Jesper Hoffmeyer, a biosemiotic researcher at the University of Copenhagen, broached this subject when he examined the swarmlike behavior of cells and tissues in the human body. He drew parallels between the intelligence manifesting itself in insect swarms and the cognitive workings of the human mind. Defining swarm intelligence as a “distributed problem–solving capacity,” Hoffmeyer entertained the idea that, when it comes to intelligence, the brain works like a “swarm of swarms” that overlaps within a “hierarchy of swarms,” sometimes likened to a “floating brain.” In this way, swarm dynamics might help solve what Hoffmeyer calls the “ever–returning homunculus problem,” which holds that “there seems to be nobody—no homunculus—inside our brain who does the thinking, there just is no central processor to control the activities of the mind.”1
If, as biologists claim, the human brain has no central processor, then what is controlling what? Could swarm entities, in a sort of networked–mind intelligence, be setting all the rules? Could higher order emerge from mindless voids—similar to how evolutionary processes might operate? These questions elevated the importance of swarm dynamics, putting it on the front burner. Most researchers now subscribe to the theory that the aggregate actions of dumb ants responding to local stimuli led to completion of complex tasks, making many human endeavors appear almost simple–minded. Christian Jacobs, associate professor of Bioinformatics at the University of Calgary, Canada, detected an impressive decentralized intelligence within swarms. He wrote: “For some types of applications, a collection of small, simple agents with limited intelligence, local decision–making capability, and a communication path to nearby peers can outperform a large centralized processor.”2
One scientist who believes that humans could learn a thing or two from biologically inspired systems is Eric Bonabeau. In his judgment, swarm intelligence is an evocative modeling tool that could help human beings better organize themselves and society—which could help in developing optimization algorithms and problem–solving methods. A theoretical physicist from Paris–Sud University in France and author of Swarm Intelligence: From Natural to Artificial Systems, Bonabeau became obsessed with complexity theory and adaptive problem solving by watching insect colonies in the Rocky Mountains. From those observations, he came to the conclusion that “swarm intelligence offers an alternative way of designing ‘intelligent’ systems in which autonomy, emergence, and distributedness replace control, preprogramming, and centralization.”3
In a 2003 interview, Bonabeau was unabashed about what he had observed. He had noticed that concentration of authority is a poor organizational technique. He commented: “Human beings suffer from a ‘centralized mindset’; they would like to assign the coordination of activities to a central command.”4 Bonabeau contended that simple systems can reach high levels of complexity without taking big, deliberate steps.
Like genes, social animals do not respond consciously. And yet, the ant colonies Bonabeau had scrutinized operated seamlessly and efficiently with no central command whatsoever. “There’s no individual in charge,” explained Bonabeau. There’s no top–down supervision. Yet these insect colonies have arranged coordinated activities of extraordinary achievement, which also serves the best interest of their whole community.
So the next question becomes this: How do dumb units—in this case, bees and ants—respond so quickly, accurately, and uniformly? For many researchers, this is where dumbness becomes an advantage. Lower life forms are too simple–minded to alter reality toward a particular bias. In sharp contrast, human behavior is more complex—the smarter the person, the greater the temptation to lie and cheat in order to achieve imagined goals.
The dumbness–versus–high–intelligence disparity cannot be overstated. Dumbness prevents social insects from distorting reality; they are too dumb to deliberately lie. In contrast, humans are too smart not to. Since human intelligence is based on abstract thought, this mental capability allows people to reshape reality. This gives humans the ability to rationalize and theorize in which ways to disregard the truth, imaging reality not as it is but as it should or could be.
Socially organized insects (eusociality) don’t have the intelligence to make up falsehoods. They all witness what is actually happening, which allows a colony instantly to react as one. In their world, truth has no agenda, but for highly evolved life forms, agendas are geared to override the truth.
But could the principles of swarm dynamics work in human society? Kevin Kelly, the founding executive editor of Wired magazine and an expert on the digital culture, has given a high–five affirmation. In New Rules for the New Economy, Kelly went to great lengths to exhibit the incredible nature and accomplishments of swarms. First he jumped on the self–organization bandwagon and stated, “Dumb parts, properly connected into a swarm, yield better results.”5 Next he equated unintelligent computer chips and personal computers with the hive–mind intelligence that has been so successful with networked systems, suggesting that the “surest way to smartness is through massive dumbness.”6
As a concrete example, he pointed to the human body, where dumb cells swarm together to propagate a smart immune system—still somewhat of a mystery to medical doctors. Obviously, swarm intelligence has established the fact that individual parts do not need to be sophisticated for their system to work well. But Kelly envisions a worldwide interconnectivity of cooperation that will liberate mankind by creating a real–time networked society.
Given recent advances in wireless technology and real–time GPS location signals, Kelly predicted a well–connected and coordinated society, robust in prosperity, attributable to the wonders of swarm dynamics and unguided learning. Examples of wide–ranging applications are everywhere. For instance, robotic arms at a GM auto–paint factory in Fort Wayne, Indiana, operating independently with tiny brainlets, beckon unpainted cars to come to their station if they need a particular color, saving time and money. The robots effect their own scheduling, without a central system telling them what to do. Japan uses the same, decentralized, mini–brain swarming system to schedule its famous, high–speed, bullet trains. Other applications include routing optimization, traffic signal control systems, communication networks, simulated crowd scenes in animated movies, military applications for controlling unmanned vehicles, optimizing shipping schedules, Particle Swarm Optimization (PSO) algorithms, NASA’s planetary mapping, and possible methods for medical “nanobots” to seek out and destroy cancer cells.
Now consider an even more intriguing finding. At the University of Florida in 2004, researcher Thomas DeMarse put together a collection of 25,000 living rat neurons, or nerve cells, inside a glass dish, and tried an experiment to produce a sort of “living computer.” The experiment was originally set up to study neural disorders, such as epilepsy, but instead, DeMarse arranged 60 electrodes in a grid to see whether this experimental brain could fly a simulated F–22 fighter jet. Amazingly, it could—and did.
The team of researchers discovered that the cortical neurons were self–organizing and connecting inside the glass dish, creating a living neural network. Next, through a specially designed multielectrode array, DeMarse was able to communicate with the living network. At first, the cluster of neurons made the aircraft drift randomly, but eventually modified the use of the data and gradually learned how to operate the aircraft. The maneuvering was limited, basically controlling the pitch and roll of the simulated flight. But DeMarse’s experiment with neural cooperation showed how swarm dynamics operates on the micro level.7
Medical scientists have realized that both chaos and swarmlike dynamics can drive the human brain. In a revealing article in New Science, David Robson concluded that “Disorder is essential to the brain’s ability to transmit information and solve problems.”8
A number of neuroscientists now recognize that near–chaotic states are vital to how the brain stores memory. Although the brain mostly appears orderly, it can unpredictably lurch “into a blizzard of noise.” And it is this swarm of disorder perched on the edge of chaos that is responsible for sparking man’s intelligence.
Researchers point to a clustering effect known as self–organized criticality, which somehow activates spontaneous avalanches of instability in a healthy brain. But nobody can figure out how the brain reaches its critical tipping point, where order bursts into chaos. Scientists did discover that collapse of order is tied to the power law of distribution—which says that bigger avalanches happen less often than smaller avalanches.” Yet the triggering device for the neural avalanche remained elusive.
When computational models were applied to mimic the brain under early theories of chaos, researchers were unable to imitate the behavior of the brain. Something was missing.
The problem lay with chaos theory itself. In the1980s, most chaologists were convinced that dynamic, nonlinear systems were simply hiding order from within—“deterministic chaos.” Most systems were not viewed as being dependent on randomness. Rather, complexity was seen as too deeply elaborate to understand, at the moment. But that mindset changed by the 1990s, when scientists discovered that the brain generates random “noise.” This fact alone made researchers realize that the brain depends on random fluctuations, meaning that the brain’s network of neurons requires both “stable phase–lock states and the unstable phase–shifting states.” In fact, there appears to be higher intelligence among brains that spend a good deal of time in a balancing act at the edge of chaos.
In this sense, the human brain represents the emergence of collective intelligence, with dumb neurons and synapses swarming in self–organizing clusters, to create intelligence.
1 Jesper Hoffmeyer, “The Swarming Body,” paper presented to the 5th IASS congress in Berkeley, June 1995, later published in Semiotics Around the World, Proceedings of the Fifth Congress of the International Association for Semiotic Studies, Berkeley 1994, Berlin/New York: Mouton de Gruyter 1997, pp. 937–940.
2 Frank Lacombe, “Modeling Swarm Behavior,” Science website at the University of Calgary. Posted on Physorg.com, Feb. 21, 2006.
3 Derrick Story, “Swarm Intelligence: An Interview with Eric Bonabeau,” O’Reilly Network, posted Feb. 21, 2003.
4 Story, ibid.
5 Kevin Kelly, New Rules for the New Economy, Chapter One: Embracing the Swarm, New York: Viking, 1998, p. 13.
6 Kelly, ibid., p. 14.
7 Ray Carson, “Brain in a dish acts as autopilot, living computer,” University of Florida, Oct. 22, 2004, part of a $500,000 National Science Foundation grant.
8 David Robson, “Disorderly Genius: How chaos drives the brain,” New Science, June 29, 2009.