Regardless of how long creative algorithms may take to exceed human performance in any one domain, they have already taught us two important lessons. The first is that the right building blocks are crucial to producing creative works. These building blocks are obvious in domains like electronics or music, but less obvious in domains like the visual arts. The second lesson is that creativity requires the conquest of landscapes as complex as the adaptive landscape of evolution, the cost landscape of vehicle routing, and the energy landscape of bucky-balls. In these landscapes creativity must avoid the fatal attraction of nearby hills or shallow valleys, which harbor the obvious and mediocre. They are death traps on the path toward the original and superior.
The universe has managed to sidestep these hills since even before life arose—in creations as different as bucky-balls forged in a sun and snowflakes drifting down from the sky—all thanks to the power of thermal vibrations. Life has added its own tricks—genetic drift, recombination—and computer scientists have added yet others, all in the service of finding the highest peaks or lowest valleys in a landscape of creation.
The books of nature are still easier to read than those of our brains, and that’s why we have not yet mapped the landscapes that our minds explore. What we can safely say, however, is that landscapes must be as central to human creativity as they are everywhere else. That’s because creativity is about solving difficult problems, and—a key insight of twentieth-century computer science—a problem’s difficulty is encapsulated in its solution landscape. No matter whether a problem is solved by an evolving organism, a computer, or a human mind, a landscape of solutions needs to be conquered. What is more, the landscapes of our minds must have many peaks and valleys—mediocre stories, corny poems, and lousy compositions—otherwise anybody could write like Tolstoy or compose like Mozart. But before we explore how human minds avoid the trap of mediocrity, it will be useful to learn that Darwinian evolution and the mind’s creative process are more similar than one might think.
Chapter 7
Darwin in the Mind
On April 26, 1937, the tiny Basque town of Guernica was overflowing with people attending its Monday market, when a squadron of German bombers and fighter planes attacked. Three hours, forty tons of bombs, and thousands of machine-gun bullets later, hundreds of unarmed villagers were dead.
Words fail to capture the horror of Guernica, but art might be able to. If any painting can capture human anguish, Guernica, Pablo Picasso’s contribution to the 1937 Paris World Fair, is the one. A woman wailing over her lifeless child, a dismembered warrior with a broken sword, a terrified figure engulfed in flames, and a screaming horse with a gaping wound in its flank suffer amid a jumble of broken bodies. Guernica is a stark testimonial to suffering and pain.1
Guernica not only reveals the horrors of war, it also reveals a great deal about the creative mind. That’s because Picasso dated and numbered the forty-five different sketches he drew to prepare the painting. Some of these sketches explore different arrangements of people, animals, and body parts, others show variants of the mother with the dead child, the head of the dead warrior, and the gored horse. Some sketches are easily recognizable in the final painting, others are transformed almost beyond recognition, and yet others Picasso tossed out altogether. What is more, Picasso’s lover, Dora Maar, photographed various stages of progress in the painting’s creation.2
These sketches and photographs have been a boon to students of creativity. One of them is psychologist Dean Simonton, who has used the sketches to study the nature of human creativity, specifically the idea that human creativity is a microcosm of Darwinian evolution, played out inside the human mind.3
This idea is not new. It is, in fact, much older than Picasso’s painting and arguably even older than Darwin’s Origin of Species. In 1855, four years before Darwin published Origin, the Scottish psychologist and philosopher Alexander Bain espoused it when he pointed out that trial-and-error is important for creativity.4 Twenty-five years after Bain, the philosopher and psychologist William James described the creative process as “a seething caldron of ideas, where everything is fizzling and bobbing about in a state of bewildering activity, where partnerships can be joined or loosened in an instant, treadmill routine is unknown, and the unexpected seems the only law.” To him, “the genius of discovery depends altogether on the number of these random notions and guesses which visit the investigator’s mind.”5
But the notion that creativity resembles Darwinian evolution only gained serious traction in 1960, when an article by psychologist Donald Campbell introduced the term blind variation and selective retention—often abbreviated as BVSR—which is still widely used to describe it.6 Its essence is this: just as mutations blindly create genetic variation, we humans blindly create variants of images—or texts, concepts, and ideas. And just as natural selection retains some organisms and discards others, we retain those ideas that are pleasing, useful, or simply apt—like those Picasso selected to convey the horrors of war. One important consequence of Darwinian creativity is that it faces the same towering obstacle as Darwinian evolution: successful creation can require much more than what shortsighted selection can provide.
It is easy to accept that selection matters for Darwinian creativity, because we obviously select some ideas and reject others. In contrast, the blind variation part is harder to swallow. We do not know exactly how new thoughts, concepts, and ideas—useful or not—originate, a bit like Darwin, who did not know how variation originates. The comparison is useful, but it is also misleading. Unlike Darwin, who knew nothing about DNA, we do already know the ultimate substrate of new thoughts. It is the firing of neurons in our brain. The firing rate of neurons fluctuates randomly and spontaneously because neurons sometimes release neurotransmitters at random, exciting other nearby neurons and causing them to fire. Ultimately, these random firings are caused by the same thermal vibrations seen in molecules and atoms—heat—that are responsible for the folding of proteins, the catalysis of chemical reactions, and the self-assembly of crystals.
Here is how Stanislas Dehaene, one of the world’s leading neuroscientists, describes this process:
There is nothing magical behind the notion of spontaneous activity. Excitability is a natural, physical property of nerve cells.… Fluctuations [in neural activity] are due in large part to the random release of vesicles of neurotransmitters.… This randomness arises from thermal noise.… What starts out as local noise ends up as the structured avalanche of spontaneous activity that corresponds to our covert thoughts and goals. It is humbling to think that the stream of consciousness, the words and images that constantly pop up in our mind and make up the texture of our mental life finds its ultimate origin in random spikes sculpted by the trillions of synapses laid down during our lifelong maturation and education.7
So we already know the root and random causes of new thoughts. What we do not know yet is how exactly the firing of one (or a trillion) neurons translates into a new thought. Perhaps it would console neuroscientists that those biologists who study DNA mutations still struggle with a similar problem. Whenever a single DNA mutation brings forth a mouse with a new coat color, a fly with a crippled wing, or a plant with larger leaves, it can take geneticists years to find out how the DNA change causes the new phenotype. Any one gene cooperates with hundreds or thousands of others, and disentangling their role in the phenotype remains challenging. Disentangling how neurons produce conscious thoughts will be no less challenging.
Another reason why people resist accepting creativity as Darwinian originates from a common misunderstanding: blind or random variation is often taken to mean creation from scratch. But that is not its true meaning, not even in biological evolution, where DNA mutations modify already existing DNA. Mutate the DNA of a fish, and you get another kind of fish, not a bird, reptile, or dinosaur.8 (Darwin did not know about DNA, but he recognized the essence of this principle when he coined the phrase “descent with modification.”)
/> As in evolution, so too inside our heads—blind variation does not mean creation from scratch there either. A creative mind does not create arbitrary images that are unrelated to what it is trying to express. To prepare Guernica, Picasso’s mind did not turn to a playing child, a blooming flower, or a rising sun, which might not have worked to visualize the unspeakable. Nor were Picasso’s sketches completely unrelated to previous works of art. The horse, for example, occurs not only in Guernica, but is a frequent theme in Picasso’s paintings, and the woman with the dead child resembles one of Francisco de Goya’s etchings in the Disasters of War. In other words, the capacity to create major paintings, novels, or theories requires training and experience so as to create the right kinds of variation. Picasso’s mind contained a huge stockpile of imagery before he created Guernica—it had to, or a painting as powerful as Guernica could not have emerged. Likewise, physicist Paul Dirac had to be steeped in physics and mathematics, or he could not have predicted the existence of antimatter. Dostoyevsky had to be immersed in telling complex stories, or he could not have wrought a masterpiece like The Brothers Karamazov. And Beethoven needed to have absorbed innumerable musical phrases before he could compose his symphonies.
In the same vein, because blind variation builds on past experience, Paul Dirac could not have come up with Beethoven’s ninth symphony, nor would Dostoyevsky have discovered vaccination. Louis Pasteur’s famous saying that “chance favors the prepared mind” applies to much more than scientific discoveries. Preparing a mind requires a life’s worth of learning and experiences. These experiences lay down a pattern of neural wiring that guides which new thoughts, images, or melodies can emerge spontaneously—just like a genome’s existing DNA restricts what new mutations can create.
So neither creators nor their ideas are blind to the past, but they, like evolving organisms, are blind to something else: the future. Just as nature cannot foresee how a mutant organism will fare, Picasso could not foresee how a particular sketch would fit into the whole painting.9 Had Picasso been clairvoyant, he could have created Guernica in one fell swoop and saved himself all those sketches. No need to paint the warrior’s upraised fist, which appears in an early stage of the painting, because it will get modified and later disappear again. And no need to produce sketch number nineteen, the head of a man with bull’s horns, nor sketch number twenty-two, a bull with a human head, because neither of them will make it into the final painting.10
Other creators are no more farsighted. When University of New Mexico psychologist Vera Steiner-Johns examined the lives of more than one hundred eminent creators—including painter Diego Rivera, chemist Marie Curie, and composer Aaron Copland—she found the telltale signs of such blindness: their minds spouted a profusion of novel ideas before they selected those worth keeping.11
Many creators are aware that their ideas are hit or miss. The French poet and essayist Paul Valéry said that “it takes two to invent anything. The one makes up combinations; the other chooses.” The English poet and playwright John Dryden described a nascent play more floridly as “a confused mass of thoughts, tumbling over one another in the dark; when the fancy was yet in its first work, moving the sleeping images of things toward the light, there to be distinguished, and then either chosen or rejected by judgment.” The physicist Michael Faraday said that his ideas had “been crushed in silence and secrecy by his own severe criticism” and that “in the most successful instances not a tenth of the suggestions, the hopes, the wishes, the preliminary conclusions [were] realized” in his final work. The chemist Linus Pauling said it more succinctly when he argued that a successful scientist must “have lots of ideas and throw away the bad ones.”12 Jacob Rabinow, an inventor of devices as different as optical scanners and pick-proof locks, as well as a member of the National Inventors Hall of Fame, said that “you must have the ability to get rid of the trash which you think of. You cannot think only of good ideas.… And if you’re good, you must be able to throw out the junk without even saying it. In other words, you get many ideas appearing and you discard them.” And John Backus, a computer scientist who helped create the widely used scientific programming language FORTRAN, said that to work successfully, “you need the willingness to fail all the time. You have to generate many ideas and then you have to work very hard only to discover that they don’t work. And you keep doing that over and over until you find one that does work.”13
Such testimonials are only a sliver of the evidence for the Darwinian nature of creativity. Other evidence comes from the serendipitous nature of many scientific discoveries. DuPont chemist Roy Plunckett stumbled upon Teflon when he tried to create a new refrigerant gas. The British inventor Thomas Newcomen discovered the atmospheric steam engine when a broken seam in an engine’s outer envelope accidentally injected cold water into the engine’s steam cylinder. And Louis Pasteur discovered an important principle of vaccination when he found that a spoiled culture of chicken cholera can immunize chicken.14 (Such creative accidents also help us see that we tend to overrate human brilliance.)
Spectacular successes like these make history, but at the price of the many failures on which they are built. We can get a glimpse of these failures and their numbers wherever written publications immortalize a creator’s work. Such publications—and how often others cite them—provide further evidence for Darwinian creativity. Each citation of a publication pays an intellectual debt to the work. The greater this debt, the greater is the number of citations, and the greater is the work’s influence. For this reason, citation counts can quantify not just the influence of a single publication, but that of an entire body of work—and of a creative person. That’s also why citation counts are a popular currency of influence when academic or government committees award grants, prizes, and jobs.15
At one extreme of the influence scale are publications cited by thousands. They are the best predictors of scientific distinctions such as Nobel Prizes.16 At the other extreme is work that never gets cited. It is like the proverbial tree that falls in the forest while nobody is looking. And it is astonishing how many trees fall unnoticed, how much effort of creative people goes to waste. Year after year, thousands of newly appearing publications are completely ignored by others, another case in point for creation by trial and—mostly—error. The number of misses is especially egregious in the humanities, where more than 80 percent of papers are not cited even once five years after they have been published.17 Numbers like this might bring shoddy research to mind, but even eminent creators, those whose work revolutionized their field, also produce inconsequential work. A lot of it.
For example, let’s look at the publications of ten influential psychologists, including luminaries like B.F. Skinner, a father of behaviorism who showed how to manipulate the behavior of animals and humans, or Wolfgang Koehler, who taught us that chimpanzees can solve problems creatively. Of all the publications that these psychologists wrote, 44 percent were not cited even once within five years after publication.18 That’s almost half the work of the best scientists in their field, completely ignored. And eminent artists do not fare better—most of their work seems not to be created for eternity either. For example, only 35 percent of the work of ten famous composers, including Mozart, Bach, and Beethoven, is still performed or recorded.19
Not only do the greats produce widely ignored ideas, but they can also make blunders so stunning that they are still remembered a century later. Lord Kelvin, for example, a towering figure in nineteenth-century physics whose name is immortalized in the scientific unit of temperature, underestimated the age of the earth by a hundred-fold. His estimate disturbed Charles Darwin greatly because it meant that evolution would not have had enough time to create life’s diversity. (The estimate was proven wrong by Ernest Rutherford in the early twentieth century.) No less a genius than Isaac Newton maintained that an achromatic lens—one that focuses light of different colors on the same plane—cannot be built. A few centuries later, such lenses are commonplace in microscope
s. And Albert Einstein, who disavowed quantum theory because he felt that “God does not throw dice,” stubbornly developed a unified theory of physics that was doomed for that very reason.20
If outstanding creativity is hit or miss, yet another historical pattern should follow. Dean Simonton calls it the “constant probability of failure”: the more dirt you pan, the more worthless gravel you will find.21 Poet W.H. Auden expressed it this way: “Chances are that, in the course of his lifetime, the major poet will write more bad poems than the minor”—simply because major poets write many poems. The flip side is a constant probability of success: the more dirt you pan, the greater your odds are of hitting gold—the more creative works you produce, the greater should be their success. This is indeed the case, as one can prove for scientists by studying citation patterns. To begin with, a scientist’s total number of citations rises with the total number of publications. That much is perhaps obvious. Less obvious is that the total number of publications also predicts the number of citations that the scientist’s top three publications receive. What is more, the most prominent US scientists—Nobel laureates—publish on average two times as many papers as their less prominent peers. Also remarkable is that this pattern does not just exist for the recent past. It holds all the way back to the nineteenth century, where the total career output of a scientist predicts his name recognition to this day.22 Exceptions do exist, like the Austrian monk Gregor Mendel, who published little and was ignored for half a century, but whose experiments on pea plants eventually triggered the genetics revolution of the twentieth century. But the small number of such exceptions confirm the rule.
Life Finds a Way Page 13