Insights into Innovation

See allHide authors and affiliations

Science  21 May 2004:
Vol. 304, Issue 5674, pp. 1117-1119
DOI: 10.1126/science.1099385

In his Theory of Economic Development (1), the economist Joseph Schumpeter distinguished between inventions—the creation and establishment of something new—and innovations, inventions that become economically successful and earn profits. In this distinction, Schumpeter echoes an earlier dichotomy in biology between the physical sources of genetic and phenotypic variability among organisms and those factors leading to the establishment (fixation) of a favored variant within a population. Schumpeter's definition of invention intentionally includes fixation, and thereby highlights the elusive nature of innovation with its connotation of influence and success.

The theoretical foundations of evolutionary invention and innovation were discussed at a recent workshop at the Santa Fe Institute (2). The meeting brought together biologists, paleontologists, technologists, and economists to consider the nature of evolutionary novelty and the similarities and differences between biological and technological invention and innovation.

Case studies of invention and innovation abound, from the Cambrian radiation of animals in biology to the telegraph, telephone, and Internet in technology, and some are sufficiently beguiling to obscure an evident lack of generality. Three explanations for the absence of robust, general theories of invention and innovation emerged at the meeting. First, “innovation” and “novelty” are two of the most overused and misunderstood words in evolutionary biology. For example, some meeting participants defined novelty as rare morphological transitions that result from breaching genetic or morphological constraints, exemplified by a developmental mutation in the Yucca moth that gave rise to a new antennal limb (3). Others defined novelty as changes that have important consequences for the environment, the classic example being the origin of oxygen-dependent photosynthesis that led to an oxygenated atmosphere. Still others defined novelty as changes resulting in the generation of abundant taxonomic diversity, such as the cichlid fishes of East African lakes or the diversification of flowering plants. Second, scale is a problem: morphological innovations in the fruit fly Drosophila challenge developmental biologists studying mutations in homeobox genes that affect embryonic development (Nipam Patel, University of California, Berkeley). Yet mutations in homeobox genes and associated morphological changes may be dismissed as unimportant by paleobiologists interested in larger scale changes. Finally, many discussions ignore the distinctions made by Darwin and Schumpeter between invention as origin and fixation, and innovation as consequence and success.

Evolutionary invention and innovation.

In general, genetic or phenotypic variation in organisms starts to accumulate in the center between two developmental barriers that are difficult for selection to overcome (constraint 1 and constraint 2) (A and C). Once selection overcomes these constraints and the variation becomes fixed in the population of a particular organism, it is called an “invention.” (B) Sometimes an invention is of immediate value to the organism in its current environment and so can be defined as innovative. (D) Alternatively, an invention may become innovative only later, when there is a change in the environment. In both cases, the invention becomes an innovation because it forms the basis of a series of subsequent adaptive radiations. Invention need not imply innovation, which often depends on additional environmental events (D). Often inventions are recognized after the fact—sometimes long after they have emerged.

In biology, invention covers the generation of variant structures through mutation, gene duplication, and horizontal gene transfer. By contrast, in the world of technology, invention depends on the construction of new devices through human ingenuity. In both cases, invention is coupled to the selective mechanisms of fixation within either a population of organisms in the case of biology or the marketplace in the case of technology. Yet the persistence of some new feature is a weak predictor of ecological, evolutionary, or cultural impact. Enter innovation, which encompasses the testing of inventions in the ecological, evolutionary, or technological marketplace. Such testing leads to an increase in abundance of a particular organism carrying the trait, the generation of new species diversity, the delivery of ecological services such as the recycling of nutrients, or the worldwide success of a technological gadget (wireless networks, for example). The question is, How do innovations arise?

David Jablonski (University of Chicago) posited a critical question based on his analysis of the increase in new orders of marine organisms in nearshore regions during the post-Paleozoic era. Was this pattern driven by increased generation of new morphologies due to environmental disturbance or by the preferential survival in a disturbed environment of inventions developed elsewhere? The folding of RNA to form secondary structures is a useful model for analyzing the constraints of phenotypic variations. Walter Fontana (Santa Fe Institute) showed that the difficulties associated with overcoming constraints can be quantified by examining the topology of RNA structures. He pointed out that the most difficult phenotypic transitions (inventions) are neutral networks that share small boundaries with adjacent networks. (A neutral network is the set of genotypes connected by point mutations that produce a single phenotype.)

Pursuing the invention question further, there was discussion of several frequently overlooked epigenetic contributions to evolution, such as the methylation of genes and the posttranslational modifications (such as phosphorylation) of proteins (David Krakauer, Santa Fe Institute). Such modifications may generate proteins with new functions more rapidly than mutations, while reducing deleterious consequences. Martin Fajnik (University of Maryland) discussed the extensive gene rearrangements that enabled jawed vertebrates to evolve antibodies and an efficient adaptive immune system, an excellent example of how organisms can exploit adaptive, hierarchical selection (4, 5).

The success of new inventions is as important as how they got there. Economists have long debated whether innovation is driven by demand or supply, mirroring a similar, if more diffuse, debate among evolutionary biologists about whether the environment is the cause of evolution or whether organisms construct their own environments through evolution. The discussion by John Odling-Smee (Oxford University) of environmental niche construction crystallized this debate (6). He argued that organisms actively contribute to the construction of their own environments and influence their own selection regimes. Earthworms consume up to 82 tonnes of soil per hectare per year—this ecosystem engineering increases carbon and nitrogen compounds in the soil, benefiting earthworms and other species. Such modified environments persist well beyond the lifetime of the creatures that constructed them. Hence, invention and innovation feed back upon each other in a complex dynamic that blurs the familiar boundaries between environment and organism, development and selection.

Biologists, technologists, and economists are all concerned with innovation, but meeting participants questioned whether the similarities go beyond metaphor or simple utility. Biological evolution provides a powerful source of ideas for engineers. Engineers seeking “open-ended” design solutions to resolve the inherent trade-offs among modularity, repetitive structures (regularity), and hierarchy (recursive composition) look to biology—for example, the modular structures of kinase proteins—for inspiration (Hod Lipson, Cornell University). Population-based optimization techniques developed by computational biologists studying evolution may help engineers to design better robots or computer algorithms (Ken de Jong, George Mason University). Likewise, analyzing the directed biological evolution of protein structures (Jesse Bloom, Caltech) has helped bioengineers to design more efficient enzymes. The biologists, however, pointed out that evolution goes beyond mere optimality. Evolution points the way to the efficient exploration of combinatorial state spaces where the size of the search space increases exponentially with the number of components. Many meeting delegates suggested that invention and innovation are about developing new grammars, that is, developing new rules for combining biological and technological modules. A good example in biology is the way that DNA methylation expands the “grammar” of gene expression. Richard Lenski (Michigan State University) has analyzed the evolution of bacterial populations in the laboratory and of computer microbes to discern “genetic” trajectories that produce new traits (such as drug resistance in living bacteria and evolutionary patterns in digital organisms).

Talks by economists and technologists revealed other similarities between biological and technological inventions. A good example is economic niche construction where one technology facilitates another through positive feedback, as exemplified by improved chip design due to chip-based automation (W. Brian Arthur, Santa Fe Institute). Other examples include recombining existing technologies into new inventions (W. B. Arthur and Lee Fleming, Harvard Business School) and connecting biology and engineering by providing more inclusive definitions of machines (Doyne Farmer, Santa Fe Institute). In addition, Eric Bonabeau (Icosystem) proposed using genetic algorithms to generate a suite of engineering inventions that could then be assessed by humans in a filtering step during the transition from invention to innovation. Although human inventors purposefully combine features across technologies—and this may be the most important source of new technological inventions—many engineered recombinants, like biological recombinants, are defective. The redeployment of elements of the developmental tool kit, such as transcription factors in new settings, is analogous to the sharing of technologies.

There emerged at the meeting an appreciation of the importance of reciprocal exchange of principles between biological and technological innovation. Such principles have arisen sporadically, however, and their success has been hard to predict. More useful would be a general theory of innovation within which to organize existing case studies and models. Meeting participants agreed that this theory requires some new combination of the dynamics of development (or construction more generally) and selection. A theory that unites technology with biology probably also requires nontraditional models of computation coupled to a better understanding of the complex feedbacks present between individuals and their environments.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
View Abstract

Stay Connected to Science

Navigate This Article