Creativity: The Muse in the SystemMarco Polo describes a bridge, stone by stone.
"But which is the stone that supports the bridge?" Kublai Khan asks.
"The bridge is not supported by one stone or another," Marco answers, "but by the line of the arch that they form."
Kublai Khan remains silent, reflecting. Then he adds: "Why do you speak to me of the stones? It is only the arch that matters to me."
Polo answers: "Without stones there is no arch."
Historically, the study of creativity has been akin to diverse disciplines polishing individual stones, each vying to be the keystone, but each incapable of forming the complete structure on their own. While philosophers have been proposing theories for the origins of ideas for millennia, the modern academic interest in creativity can be traced to Gilford's historic 1950 address to the American Psychological Society. A new wave of psychosocial theories arose; some prominent examples are Csikszentmihalyi's sociocultural feedback model, Boden's H and P creativity, little c and big C creativity as well as de Bono's divergent and lateral thinking. Two more recent approaches have been the neuroscientific search for neural correlates of creativity and computer programs of algorithmic creativity. Each has brought a new perspective, but has failed to build the arch.
My work is about building an arch of creativity using stones from a different quarry. My thesis is that creativity may emerge from any system, not only the human mind, that contains within it precise and measurable attributes. As a first step, I have discarded an assumption that has infected much of the literature: only humans are creative. To make this distinction clear, I define creativity broadly as the ability of a system to: 1) Generate novelty, 2) Perform an action, 3) Detect the result and then 4) Adapt based upon the result. There is no bias toward the kind, degree or complexity of the system, no assumption about the magnitude of the novelty or the impact of any artifacts that might be produced, and no claim that creativity must arise from a problem. Defined in this way, many systems are creative: cities, immune systems, economies, ant colonies, ecosystems, minds, and groups of minds. On the surface, this approach appears antithetical to the humanistic approach! But, it is by looking outward, to the common characteristics of other creative systems that will allow us to better understand and appreciate human creativity.
As a starting point, I following a path that has been traveled before; novelty is the reassembly of old parts into new configurations. I follow this path farther by adding three supporting points. 1) Novelty arises from a system because of how the system organizes its internal processes; the Muse is built into the system. 2) Creative systems must contain within them non-random mechanisms for mixing old parts; the Muse is not blind. 3) Learning, hard earned from successes and failures, is stored inside the system; the Muse can become more creative. Now that the general shape of the arch has been sketched out, it is time to turn to the individual stones.
Network science studies how units (e.g. people, neurons, ants) connect and propagate information, materials, products, ideas or energy throughout a system. Using the tools of network science, we can uncover common patterns in creative systems. Creative networks are composed of smaller sub-units called motifs: groupings of units that perform the functions of collecting information from the environment, acting on the environment and processing information. Creative networks also fall into a unique class of networks called scale-free, or hub networks, a structure found in the human brain, genome of plants and animals, social interactions, and the organization charts at innovative companies such as GORE and IDEO. Networks form the container for mixing, and the bowl always has a similar shape.
A static scale-free network, however, will only find what Stuart Kauffman calls the adjacent possible: the obvious combinations that drive forward an incremental evolutionary process. For radical recombination to occur, the network must dynamically turn on unusual combinations in synchrony. A clue comes from a simple observation; creative systems cannot be creative all the time and expect to survive. They are stable and resistant to outside perturbations most of the time, but are punctuated by bursts of creative instability. Just such an idea is found in the physics of open systems. Creative systems are open to their environment, meaning that as energy and materials stream in from the outside world, the system is pushed away from stability. As the system pushes back, tension builds, eventually being released in what physicists call an avalanche. During a creative avalanche, motifs that are usually separated in the scale-free network can briefly communicate and perhaps stick. This burst of instability is the chaotic mixing bowl for trying out unusual combinations of motifs. After the avalanche, there is a return to stability, but with the possibility of a change in how motifs have stuck together. What is more, creativite systems self-tune to find a balance between stability and instability, a concept known as self-organized criticality. The theory of open systems even predicts that the magnitudes of avalanches vary in a way that has been documented in the creativity literature; while smaller avalanches are common, large avalanches are rare.
Random mixing during an avalanche would quickly become a combinatorial nightmare; finding an unusual but useful grouping of motifs would become prohibitively rare. But a slight skew could tip the balance; the house only needs a slight advantage to win in the long run. All creative systems have ways to constrain combinations without being deterministic. What is more, different creative systems can layer and tune these constraints to generate more or less radical jumps, with humans, and groups of humans, being the most adept.
Generative grammars define strict rules for how motifs can combine; many motifs simply do not stick together. Although the traditional generative grammars of linguistics are tree-like, there are classes of grammars that generate scale-free networks.
Bottom-up and top-down hierarchies are another common characteristic of creative systems; DNA prescribes the raw materials of cells, developmental rules guide cells to form tissues, tissues combine to create organisms, groups of organisms make a species, and ecosystems arise from interacting species. But these hierarchies run in reverse too; the ecosystem influences individual species, pressures on a species will impact individual organisms, and so on down the line. These bidirectional rules break the strict causal relationships assumed in most scientific experiments, and place even greater bounds on the recombinations that may occur.
Some creative networks also contain within them a model of the world; a bacterium has the simple model that moving up a concentration gradient will lead it to a food source. The advantage is that many motif combinations can be generated simultaneously and then allowed to compete against one another in the low-cost, low-risk internal world, with the winner being enacted in the real world.
Accompanying a winning solution is a prediction of the outcome. Creative systems often contain motifs that compare differences in the expected and actual outcome and then make internal changes to correct the difference, completing the feedback loop implied in my definition of a creative system. Motif combinations that function well are strengthened, building them into the system; an immune system stores a record of biological attack in antibodies, and a culture stores experiences in laws and customs. The system can learn, and in doing so it becomes more tightly coupled to its dynamic environment. With many experiences, deeply built-in responses can be grouped into heuristics, fast probabilistic guides that further constrain the combination of other motifs. This automaticity lowers demand on limited resources, allowing the system to bootstrap to greater complexity, and a greater range of function.
Building a creative arch may be useful to several fields. For scholars of creativity, new rigorous tools may be created to improve upon existing measurement instruments and dissect common creative techniques; De Bono's Six Hats technique positions a group on the edge of an avalanche. The framework may also suggest methods for tracking the development of individual creativity and monitoring team dynamics. I also envision a symbiotic relationship with scholars of artificial creativity, a subbranch of artificial intelligence. My view of creativity may offer a fresh perspective, while their rigorous algorithmic restrictions will help refine this creative framework. Certainly, there are pedagogical implications of designing and maintaining adaptive learning environments. Lastly, artists of all kinds may be provided with a framework for creativity that is more scientifically grounded.
Of the few threads that our academic cultures share in common, creativity runs throughout - sparking social and cultural change, powering the engines of technology and feeding our artistic souls. Understanding the creative arch is perhaps the most powerful way to appreciate the most unique of human of gifts: our extraordinary ability to create.