Psi man mascot

Creativity: Variation with Selective Retention

So far we have described patterns in systems that already exist. They maintain stable states, they go through S-curves of growth, they show rhythmic cycles, and they reduce or amplify deviations from a goal-state or equilibrium state.

How do new systems come into being in the first place? That is the same question Charles Darwin faced as he tried to explain the origin of animal species.

Darwin suggested that species evolved through natural selection. The logic was simple. When animals produce babies, there are variations.

For example, every baby finch is different. This is due to combining the genes of two parents, mutation, and other factors. The key factor is that variation occurs in the reproductive process.

Some of these variations work better than others in producing finches who grow up to have viable young. Those winning variations eventually dominate the population.

Dad and two baby finches who look different
Natural variation in finches

That's Dad on the left, two baby finches on the right. As always, there is some variation in body shape and color, but there are also variations we cannot see. Some finches are adventuresome, some are shy, and so forth.

Up to about the 1990s, animal researchers resisted speaking of "personalities" in animals. However, accumulating data show that animal personalities are "stable and quanti­fiable" and make a difference in survival and reproduction (Pennisi, 2016).

For example, shy animals tend to do better in crowded and competitive situations, while bold and adventure­some animals tend to be better hunters. Inevitably the species is changed to resemble the animals that survive and reproduce.

Not all babies survive to adulthood and reproduce. Those that do survive and reproduce pass on their characteristics to the next generation; others do not.

When enough changes accumulate, animals in one population might no longer be able to breed and produce fertile offspring with another population (as happened with horses and donkeys). Then a single species splits into two species.

Formation of new species has been observed in modern laboratories with quick-reproducing organisms. Fruit flies were observed to produce a new species (that could not breed with the ancestral population) in 1971. Meyer et al. (2016) produced speciation of a bacterial virus in a test tube.

Darwin bred fancy pigeons as a hobby, so he knew about selective breeding, as did many other people in the mid 1800s. Darwin simply pointed out that the same process occurred naturally. Instead of human-directed selection, it was natural selection.

Darwin brought back samples of finches from different islands in the Galapagos chain. After examining them, he realized each species reflected conditions on an individual island.

This happens because island popula­tions are isolated. If an island has seeds or nuts with hard shells, a short, stubby beak is superior for cracking them.

Birds on that island that are born with shorter and stubbier beaks have an advantage obtaining food and reproducing. Gradually they take over the population of finches on the island. In Darwin's finch collection there was a range of different body types, each reflecting a different island ecosystem.

Generality of the Evolutionary Pattern

Many scientists pointed out that the pattern Darwin identified, variation followed by selective reproduction of adaptive variants, applies to more than just biological evolution. Gerald Weinberg (1965) wrote:

In the tumultuous development of Darwin's ideas in the century (following publication of The Origin of Species), natural selection has been revealed as a phenomenon not confined to 'living' systems, but explainable in purely abstract terms. All that is necessary is that a popula­tion exists under the following three conditions:

(1) Its individuals are capable of making reasonably exact copies of themselves.

(2) A certain amount of inexactitude is present in the copying process.

(3) An environment exists which selectively favors certain variations.

...All three must be present for Natural Selection to take place; and when all three are present, Natural Selection must take place. (Weinberg, 1965)

Donald T. Campbell noted "three essentials" for evolution of a novel adaptive form in any system.

  1. The occurrence of variations: heterogeneous, haphazard, "blind," "chance," "random," but in any event variable (the mutation process in organic evolution, and exploratory responses in learning).
  2. Consistent selective criteria: selective elimination, selective propagation, selective retention, of certain types of variations (differential survival of certain mutants in organic evolution, differential reinforcement of certain responses in learning).
  3. A mechanism for the preservation, duplication, or propagation of the positively selected variants (the rigid duplication process of the chromosome-gene system in plants and animals, memory in learning). (Campbell, 1969)

If you compare Campbell's formulation with Weinberg's, they are the same, but in different order. Campbell's first (occur­rence of variations) is Weinberg's sec­ond (inexactitude in copying).

Campbell's second (selective criteria) is Weinberg's third (environment favors certain variations). Campbell's third (a mechanism for duplicating) is Weinberg's first (individuals can make copies of themselves).

Campbell related this pattern to creativity in general (Camp­bell, 1959;1960). When a system produces something new and adaptive, it must use the process of variation and selective retention. There is no other way to do it.

You probably recognize (by now) the pattern underlying a general system principle. "Wherever you see behavior ABC, you will find XYZ under the surface." In this case, wherever you see creativity (general of novel, adaptive forms) you will find variation and selective retention under the surface.

An example of evolution outside the familiar context of biological evolution is cognitive development in children. Piaget observed the evolutionary pattern while studying his infant son, Laurent.

Laurent started with random move­ments and gradually combined them into larger, more complex units. If random arm movements produced noise from a rattle, Laurent repeated those movements. He also learned to wiggle in ways that produced noise from the mobile hanging over his crib.

These tiny accomplishments of infancy continue and get more complicated. This is the beginning of a lifetime of learning.

As Donald T. Campbell pointed out, learning shows the evolutionary pattern. Various behavior patterns are tested, and those that produce useful results are retained through memory.

Weinberg noted that, given the three necessary conditions (reproduction, variation, selection) evolution not only can occur, it must occur. The underlying inevitability is that only patterns passed forward in time (generated and selected) will be around in the future.

Generativity is often the hardest part of this process to explain, because (unlike the overall pattern) the details of generativity are different for every system. In every act of development or evolution, some sort of combinatorial system is at work.

In biological evolution, the existence of two parents, and the process of com­bining the mother's and father's DNA into a single combined DNA molecule, is the well-known mechanism by which new variations are produced. The variations are not random in a statistical sense, because that would mean all variations are equally likely.

That is almost never the case in adaptive systems. If nothing else, the variations are constrained by what components exist to be re-combined. When the word random used to describe variations, in evolution or other developmental pro­cesses, it really means unpredictable.

A creative system must use pre-existing components, and it must start from pre-existing configurations. Therefore not all variations produced out of those components is equally likely.

Children are more likely to look like their parents than not. But the details are unpredictable.

We never know what combination of traits from the parents will be expressed in a child. That is unforeseeable or blind to use Campbell's favorite word to describe the combinatorial process. The process must be blind in order to produce novelty.

We refer to long-term biological change as evolution. We refer to short-term psychological change (when it produces adaptive outcomes) as learning or crea­tivity. However, the same pattern (varia­tion and selective retention) underlies both, and the same pattern can be programmed into computers, to make them creative.

Evolutionary Computing

Computer programs can produce novel adaptive forms. Such computer pro­grams were originally called genetic algorithms and their descendants are often called evolutionary computing.

Genetic algorithms, invented by John Holland in 1975, proved useful to industry. General Electric achieved greater efficiency in its jet engines by using genetic algorithms.

In Holland's original conception, a genetic algorithm worked with a collection of candidate structures, let's say, 10 candidate designs for an engine. The best 50% were kept, while the worst 50% were removed from consideration.

Next the best 5 designs were varied in some way (by combining them, exchanging features, or adding new design elements) to produce a fresh set of 10 engines. Those 10 are tested and the worst 5 are discarded. The best 5 are used to generate another set of 10 candidates. And so forth.

In the case of improving jet engines, Homaifar, Lai, and McCormick (1994) explained that criteria to be optimized were thrust and overall efficiency. Four key parameters were varied by the genetic algorithm: compressor pressure ratio, fan pressure ratio, bypass ratio, and Mach number.

To locate the most efficient engine design, the parameters were varied and combined, then tested. The results showed "genetic algorithms are capable of optimizing a complex system quickly." New and more efficient engine designs were located.

The underlying pattern (variation with selective retention of adaptive variants) was the same as for evolution. It produced an adaptive outcome: greater efficiency.

The principle of variation with selective retention: Creativity, the production of novel adaptive forms, requires generating variations by combining available building blocks, then selecting and reproducing adaptive outcomes.

Natural Selection in the Nervous System

Our own nervous systems evolve during our lifetimes because of variation and selective retention. Nerve cells are produced in great numbers early in life, reaching a peak number around age 7.

Nerve cells shown thinning out but getting more complex with age
Neurons become less numerous but more complex with age.

After that, progress is made by growing some of the existing neurons and killing others. Programmed cell death, the process called apoptosis, removes neurons that are useless or counter-productive.

Another complementary selection process occurs with synapses of neurons. These are the places where neurons exchange chemical messengers.

Some synapses on a neuron are strengthened (those making positive contributions, presumably) and some are weakened. That results in evolution toward more competent neurons.

Neurons perform in combination with other neurons, in neural circuits, so combinations of neurons are tested as we go about our daily activities. Nerve Growth Factor (NGF) is then supplied to the "winning" combinations, and the brain evolves toward better, more adaptive circuitry over time.

Evolutionary Opportunism

The page on efficiency of modular construction described the point made by Herbert Simon in the classic 1962 article, "The architecture of complexity." Simon was both a Nobel Prize winning econo­mist and a prominent figure in cognitive science. With the parable of Hora and Tempus, he demonstrated that large pre-existing components enable a system to be assembled more efficiently.

That is true in evolution, too. Large systems are most efficiently built out of complex, pre-existing subsystems or components. When used in a new system, old compo­nents commonly take on new functions.

Components are constantly being re-pur­posed during evolutionary development. This must happen whenever a new function emerges (otherwise we would not call it new).

Evolutionary opportunism is one name for this process of using old components in new ways. It is so important, and so ubiquitous, that it has been recognized by numerous scientists and given several different names:

--exaption

--repurposing

--preadaptation

--borrowing a component

...Or you could just call it...

--using old structures for new functions

One example involves pre-flight move­ments in flightless birds. These are movements such as stretching the wings and fluttering, seen in birds before they take off to fly.

The same movements can be found in mating dances of flightless birds like ostriches and penguins. How did the movements end up in mating dances?

The answer is, "They were available." The pre-flight movements existed in the motor programs of ancestral birds. At some point a male bird fluttered a wing in front of a female and it charmed her, and they had baby birds who were inclined to use the same movements during courtship.

That happened in flightless species like penguins. Now they no longer use the pre-flight motor routines for preparing to fly; they use them in courtship displays.

Scientists who study animal behavior refer to such inherited behavior patterns as emancipated from the original function. The genetic benefit of flight-movement circuitry is now related to mating, not flight.

The mating function keeps the DNA in the species, because birds that do the mating dance well are more likely to reproduce. Birds that execute the movement clumsily (in the opinion of female birds) have fewer young. This keeps the movements shaped up in a flightless species.

The principle of opportunism: Complex systems are assembled most efficiently by using pre-existing parts or modules. This process may transform the meaning or function of parts included in a new system.

Creativity in Cognition

Creativity is the production of novel, adaptive forms. Cognitive productions include sentences, visual images, and solutions to problems.

These productions are novel in the sense of being new as a whole (never before existing in that exact form). They are adaptive in the sense of serving a purpose or solving a problem.

To most people, the word creativity implies social approval, not just novelty. A product is creative if it is (1) new and (2) virtuous (for example, people like it).

The test of time is also important for determining what innovations deserve to be called creative. If an artistic produc­tion is highly valued for a long time, we are more likely to call it creative.

If creativity is described as any process producing a novel adaptive whole, social factors like approval or the test of time are covered by that word adaptive. An adaptive thing, a successful outcome, is one that makes a contribution or receives a positive evaluation or contributes to reproduction.

In general, a thing is adaptive if it works. An adaptive product is selected over time because it functions, or people like it. Therefore it is valued and reproduced.

If we define creativity as the creation of novel, adaptive wholes, then creativity is everywhere in cognition. Noam Chomsky, the famous linguist, said in Language and Mind (1968) that creativity was the core challenge for scientists trying to explain language.

Almost every sentence we speak is new. We have never produced it before. Yet these novel sentences work; they function to communicate.

We produce these novel, adaptive forms in seconds. We also understand them when they are uttered by other people. That is also a creative act, because to understand a sentence a listener must construct a novel, adaptive form (the meaning).

Chomsky focused on language because he was a linguist. The same creativity is found in every other realm of cognition.

Novel adaptive wholes are generated in a split second by a perceptual system when it interprets a scene. Novel adaptive behaviors are generated by brain/muscle systems when we walk around, dance, or perform an athletic move.

Creativity pervades cognition. What is the underlying process? It is the same every time: variation with selective retention.

Variation must come from a combinator­ial process, some part of the cognitive system that puts things together. The details of that mechanism will be different for every system.

The combinations of elements making up a sentence (nouns, verbs, etc.) are com­pletely different from the combination of elements making up a motor movement (movements in a coordinate system) or the perception of an image (sensitivity to lines, edges, corners, etc.) But the underlying pattern of variation and selective retention is the same.

Components are put together in an atmosphere of constraints within the cognitive system, encouraging some combinations while ruling out others. In other words, a combinatorial process is followed by the selection of adaptive combinations.

The amazing thing about this process in cognition is that it is so fluid and so fast. New cognitive products are created in a split second (such as the chain of thoughts in your thought process).

Each moment of integration is the highly complex assembly of a cognitive product. The product, as Chomsky realized in the case of sentences, typically has never existed before in precisely that form. That is the everyday miracle of the mind: ceaseless creativity.

---------------------
References:

Campbell, D. T. (1959) Methodological suggestions from a comparative psychology of knowledge processes. Inquiry, 2, 152-82.

Campbell, D. T. (1960) Blind variation and selective retentions in creative thought as in other knowledge processes. Psychological Review, 67, 380-400.

Campbell, D. T. (1969) Variation and selective retention in socio-cultural evolution. General Systems, 16, 69-85.

Chomsky, N. (1972). Language and mind. New York: Harcourt, Brace, Jovanovich.

Homaifar, A., Lai, H. Y., & McCormick, E. (1994) System optimization of turbofan engines using genetic algorithms. Applied Mathematical Modeling, 18, 72-83. https://doi.org/10.1016/0307-904X(94)90162-7

Meyer, J. R., Dobias, D. T., Medinal, S. J., Servilio1, L., Gupta, A., & Lenski, R. E. (2016) Ecological speciation of bacteriophage lambda in allopatry and sympatry. Science, 354, 1301-1304. doi:10.1126/science.aai8446

Pennisi, E. (2016) The power of personality. Science, 352, 644-647.

Weinberg, G. (1965). Natural selection as applied to computers and programs. General Systems Yearbook, 15, 145-150.


Write to Dr. Dewey at psywww@gmail.com.


Search Psych Web including the General Systems Toolkit and the online textbook Psychology: An Introduction below.

Copyright © 2017 Russ Dewey