5
Communication and the Evolution of Cognition
James A. Evans
One common way in which social interaction scaffolds cognition is through communication. This chapter explores the role of communication and shared communication protocols in shaping the cognitive institutions of knowledge and culture. Natural language is the defining human communication protocol, but there are many others: scientific and technical dialects, telephony and network protocols like TCP/IP, even electrochemical signals such as pheromones through which human and nonhuman agents signal one another. Indeed, communication protocols may not seem to be about communication at all, and include intentional or unintentional standards that facilitate communication. For example, shared measurement systems facilitate trade (e.g., metric system, container sizes [Levinson 2006]), and shared classification systems enable data transfer (e.g., medical records systems, database schemas).
Communication also includes the networks that emerge through interaction via communication protocols. This includes the topology of hyperlinks on the World Wide Web, the pattern of mobile phone conversations, or the swarm of discussion among scientists at a conference.
Cognitive institutions of knowledge and culture emerge, circulate, and evolve through human communication. Even tacit knowledge, which may not be articulated or even articulable (Polanyi 1958), passes in the slow transmission of complex gestures via apprenticeship and socialization. In this chapter, I argue that by examining the process through which communication scaffolds knowledge and culture, we can understand how shifts in the (1) quantity and (2) quality of communication in a system can influence the knowledge or culture in that system.
My argument rests on the axiom that communication is not merely a transparent conveyor of ideas, practices, or products. Communication leaves a trace on the things it conveys. This is a soft form of Marshall McLuhan’s thesis that the medium is the message (McLuhan 1994). The social process of communication passes to the receiver information about the sender. At the least, this suggests the message is compatible with and possibly even important to the sender. Moreover, the social structure of communication leaves a pattern that itself may be evaluated as compatible or incompatible with a sent message by the receiver. Because the medium constitutes part of the message, communication scaffolds knowledge and culture in ways well beyond simply facilitating the circulation of ideas themselves.
Quantity of Communication
Influence of the quantity of communication on the world’s knowledge and culture can be illustrated with a simple thought experiment. What would happen if everyone shared a common language? The potential for communication between any two people would increase. As a result, even if the number of conversations remained stable, the global current of information would escalate. An idea would be more likely to travel from a student in Yemen to one in Peru if they both spoke English or Esperanto or both coded in the computer language Python. Moreover, because the process of communication conveys some endorsement by the sender, communicated information takes the form of persuasion. Communication primes the receiver to do more than merely receive information but to consider it and to use it.
I will initially develop my argument about communication quantity in the context of two cases. The first explores the influence of the World Wide Web on knowledge and culture shared across it. The second considers how shared scientific language and theoretical commitments influence the accumulation of knowledge and the rate of advance in science. Then I will extend this to other communication contexts through several additional examples.
How the World Wide Web Scaffolds Knowledge and Culture
The World Wide Web hosts billions of web pages—trillions when one includes “invisible” pages unlinked, under password protection, or generated dynamically by online databases (Illinois Mathematics and Science Academy 2003). Many of these are connected via hyperlinks through which one online document references another and the reference can be directly followed. More recent social innovations like wikis (e.g., Wikipedia) and collaborative tagging—so-called Web 2.0 applications1—enable widely distributed multiparty conversations. Together these developments have ushered in a wave of optimism about the collaborative production of knowledge (Sunstein 2006), technology, and art (Weakley and Edmonds 2006).
The combination of individual Internet search and collaboration practices together unleash increased quantity of global communication across the World Wide Web. The power to increase the reach and volume of global communication has long been lauded by Internet optimists and evangelists. My own previous work provides empirical support for this view by demonstrating that scientists and humanists in poor countries are much more likely to cite research freely available on the Internet (Evans and Reimer 2009).
Despite the vast number of pages on the web, any two pages are only about twenty links away from each other because of the uneven distribution of hyperlinks (Albert, Jeong, and Barabási 1999). As a result, web surfers can reach virtually any site on the Internet with relatively few clicks. This expanding reach of Internet users can be confirmed by anyone who has known just enough about a subject to misspell it in a search engine’s text box, then instantly retrieve multiple, superb in-depth treatments (and some superficial or ill-informed ones as well). Even as the Internet extends the information reach of individuals, however, it narrows the global scope of things considered (Evans 2008, 2011).
The World Wide Web and other dominant systems of interconnection lure individuals to ignore information not readily accessible through them (Castells 2000; Evans and Reimer 2009, 71). As the cost of accessing information on the web drops, the relative cost of information beyond it is much greater. There is substantial evidence that systems of interconnection also facilitate convergence on limited information within those systems. Research on markets from sociology (McPhee 1963; Salganik, Dodds, and Watts 2006), economics (Rosen 1983; Frank and Cook 1996), and business (Elberse 2008)—and for varied products, including those trafficking scientific ideas (Merton 1968)—demonstrate that the most popular ideas and products attract disproportionately more attention as (i) fields grow, (ii) people gain more exposure to others’ choices, and (iii) the marginal cost of reproducing and distributing ideas and products is low. As knowledge and culture are encoded and become available via Internet, each of these properties is enhanced.
There are three primary processes through which popular ideas become more popular as people become more interconnected. First, people are social and often find value embracing beliefs held by others. Montesquieu argued that commerce enhances civility between peoples (Hirschman 1977; Hirschman 1982; Montesquieu 1748/1949). Durkheim generalized Montesquieu’s conception into a theory of societal interdependence (Durkheim 1947; Durkheim 1960). When people share protocols, trade, and communicate, they often converge around common tastes and become more similar.
A second, complementary motivation for choices to converge is that people often act in a manner consistent with a Bayesian reasoner. When lacking firsthand experience, they use others’ conclusions to calculate their own prior expectations. As they become exposed to foreign information through the Internet—and also to others’ choices of what is most reasonable, important, or beautiful—people will tend to factor these choices into their own calculus and converge to consensus.
Finally, in those native domains of experience where people trust their independent judgments (Rosen 1981), if the costs of duplicating representations of knowledge and culture are cheap and standards are shared, then people have no reason not to choose the best: Why would someone read a stilted editorial about an event they preferred to watch themselves if a video of it is streaming online? Why would someone listen to their third favorite recording of Carmen if their first is readily available (Elberse 2008)? In cases where people reach beyond their domain of experience, however, they lack confident judgment and tend to increase the weight they give to the choices of others.
Collectively these patterns suggest the Internet’s primary influence on knowledge and taste: it broadens individual reach while reducing global range (Evans 2011). This suggests why Google’s search engine, which was first to use popularity to rank queried pages, has been so phenomenally successful: it helps people follow the crowd. Google’s PageRank exploits a Markov process to simulate a random surf through hyperlinks and then ranks sites according to their expected traffic. To calculate this efficiently and add a measure of realism, it augments the pattern of existing hyperlinks with a “teleportation” matrix enabling the random surfer to move from any web page to any other with a low, constant probability (Langville and Meyer 2006).2 In this way, Google allows searchers to “teleport” into an alien domain and immediately follow the crowd as if they were natives with local insight and taste. This narrowing impact has clear implications for the way in which the web scaffolds future knowledge: that which is unacknowledged is harder to discover and will, in turn, be less likely built upon in subsequent periods.
How the World Wide Web Scaffolds Science and Scholarship
The influence of the web has been recently examined on academic knowledge production and consumption. Because the institutions of science and scholarship have sufficient rigidity that they retain the same units of communication—articles—and the same units of reference—citation lists—before and after going online, they have enabled large-scale empirical study.
Like other forms of knowledge, academic science and scholarship have increasingly moved online in the past decade. Recent investigations into patterns of library usage demonstrate that users prefer online material to print and that accesses of print material have declined as electronic use rises (De Groote and Dorsch 2003; Black 2005; De Groote, Hitchcock, and McGowan 2007). Studies also show that with the shift online, search and reading practices have changed. Experts still browse a few core journals in print or online to build their awareness of current research (Tenopir et al. 2003). After relevant articles are discovered, these are often printed and perused in depth on paper (Friedlander 2002). With the web, however, researchers are also much more likely to search by topic. The percentage of papers read as a result of browsing has dropped and been replaced by the results of online search, especially for the most productive scientists and scholars (Boyce et al. 2004). Subject experts also use hyperlinks in online articles to view referenced or related articles (Tenopir et al. 2003).3
The amount of time researchers read has increased only slightly in the past thirty years, but the amount read has almost doubled since the mid-1990s. Internet-assisted “reading” has become a form of focused click-through scanning and within-document searching (Renear and Palmer 2009). Scientists and scholars describe the Internet approach to reviewing literature as much more efficient and claim it allows them to selectively cover a wider swath of research material (Tenopir 2008).
Researchers historically reached online articles by searching major online scholarly databases (e.g., ScienceDirect, ProQuest, EBSCO, JSTOR, etc.). When a publisher’s archive is searched, titles, abstracts, and often the full text can be ranked by relevance, by date. As more of these archives have been indexed on the Web in recent years, more literature searching is done through Google and other search engine interfaces where full-text articles are searched via informal text and ranked by relevance and popularity.
In a series of surveys querying international scientists and humanists about their use of scholarly materials, respondents described having a broader reach as a result of the Internet (McClanahan et al. 2010; Tenopir 2009). Interviews associated with these surveys demonstrate the degree to which researchers across the globe experience Internet search and hyperlinking as a transcendence of linguistic, temporal, and disciplinary boundaries. For example, a Japanese researcher explained: “Having access to a much, much wider range of English-language resources than before is … a huge change for the better, as an English-speaking researcher in Japan. I have always crossed disciplinary boundaries in my own research, but this is much easier to do with access to electronic resources” (Tenopir 2006). In one of these usage studies covering U.S. and Australian readers, only 52% of within-discipline reading was done online, but 63.6% of interdisciplinary reading was done there (Tenopir 2006). By searching, researchers more likely read and cite things beyond their discipline. As they enter unfamiliar territory, however, they must rely on the judgments of others by picking work from the most reputable journals and most cited papers. This is not difficult. By searching and hyperlinking through a new domain, researchers naturally come across the most recognized articles.
For example, if a psychologist reads and cites psychology, it will be from the full diversity of outlets with which he or she has experience. If that researcher reads biology, however, it will likely be from Nature or Cell; if economics, from the American Economic Review. These central journals, when they come online, become global Internet hubs. If all researchers become more interdisciplinary, but their every outside-discipline selection refers to the global center of that other discipline, as individuals broaden, the span of all work cited will decrease. These patterns are illustrated in figure 5.1.
Previous research has noted the extreme inequality of scientific citations (Price 1965; Merton 1968), independent of that same pattern of inequality among Internet hyperlinks (Barabasi and Albert 1999). As researchers become more aware of each other’s choices and can literally “click-through” them in hyperlinked reference lists, the inequality of attention already facilitated by the social structure of science will increase.
The influence of the World Wide Web on science and scholarship has been examined before and after it became available via the web. A large sample of citation data was used to assess the influence of journals’ Internet availability on (i) the individual breadth and (ii) global selectivity by which articles are cited in subsequent research. Findings demonstrated that journals are cited approximately 13% more broadly beyond their field as they come available online, which suggested Internet availability broadens researchers’ disciplinary horizons. Material within those journals was cited more selectively—that attention was concentrated on 5% fewer articles—which suggests Internet availability narrows the global stock of knowledge acknowledged and built upon in future generations (Evans 2008, 2011).
Science and scholarship facilitated a before– and after–World Wide Web comparison, but the institutional rigidity of science and scholarship likely underestimates the impact of the Internet on other domains of knowledge and culture. Songs or T-shirt and shoe lacing styles, for example, have undoubtedly extended their reach more than 13% and increased their global concentration by more than 5% since entering the World Wide Web.
The way in which the web scaffolds the knowledge and culture that it serves has broad consequences. By connecting so many people so cheaply, the Internet facilitates the rapid spread of knowledge and culture fads. Near-random differences in the quality of an idea or the appeal of a taste are more likely to become amplified in this highly interconnected system. But in the wake of high velocity Internet fads, what gets left behind?
Buskens and van de Rijt recently posed a thought experiment in the wake of Ron Burt’s demonstrations that individuals who bridge structurally distant others garner information and control advantages (Burt 2004): What would happen if everyone tried to broker structural holes? They found the advantages of such positions would effectively disappear (Buskens and van de Rijt 2008). Differences between individual knowledge will diminish if many shift their orientation from numberless local contexts to a handful of global Internet hubs. This rush to global hubs may be self-reinforcing. In his research on mass behavior, McPhee found that “the larger the proportion of people [unfamiliar] with a given alternative, … the less likely are those who are familiar with it to like it” (McPhee 1963, 134). In this way, local, idiosyncratic knowledge will be more likely to disappear into obscurity than before the World Wide Web. Novelty must still come from somewhere, and the web makes it easier for something to “go viral” once discovered. If less time is spent reviewing idiosyncratic knowledge overall, however, there will be fewer opportunities for any particular item to be “discovered” even as items discovered will go on to become bigger hits.
In science and scholarship, local knowledge equates to ideas or findings published in a subfield or specialty journal not widely appreciated beyond its community soon after debut. As scientists and scholars spend more time outside their training fields and less time reading each article they come across, new ideas in specialized venues that are not easily searchable or that take time to digest will likely be more neglected. In contrast, terse hypotheses, singular measures, and one-phrase findings are likely to take off with the scaffolding of the web.
Recently, much has been made of the “long-tail” argument that the Internet facilitates a wider distribution of tastes than ever before (Anderson 2006). Prior social science reflections on the flexibility of new technology (Piore and Sabel 1984) and the diversity of connections enabled by “network society” (Castells 2000) came to the same conclusion: the death of the mass market. This privileges the Internet’s technical over its social reality. More products can be offered to more people in a virtual “store” like Amazon.com than in a roadside Borders Books and Music. Online, however, people can more clearly see what others have purchased, what others have appreciated, and what others believe true. Although there is a longer tail to the distribution of widgets purchased, tastes adopted, and claims believed on the Internet, more of the probability mass is packed into the distribution’s center due to the social forces of awareness and convergence (Elberse 2008).4 The World Wide Web allows individuals to reach further beyond their current domain, but as with Google’s search engine, the further they reach, the more likely they are to follow the crowd, even if they don’t see it as their own. In this way, the web scaffolding tilts society’s system of knowledge and culture, creating deeper canals and basins of attraction.
How Theory Scaffolds Science
Shared scientific theory allows more scientists to communicate and coordinate with one another and to efficiently organize around scientific problems. In this way, they enable measurable scientific advance. To the extent, however, that scientific theory also facilitates the spread of auxiliary axioms, values, frameworks, and methods, it advances science by narrowing the scope of problems considered. This second point is the central insight from Kuhn’s The Structure of Scientific Revolutions (1962), where broad theoretical and technical paradigms integrate fields of science. Combined with the first point, it suggests that more communicated agreement about how to do science leads to faster accumulation of scientific insight.5 In this sense, the scaffolding of shared theory can nudge science to know more about less in a process very similar to how World Wide Web scaffolding influences the world to attend more to less.
This influence of shared scientific theory is illuminated when academic scientists are contrasted with those located in industry, where theoretical commitments can be diluted by company interests and directives. Even though scientists in academy and industry may perform similar analyses, the institutional ends of the academy contrast with those of industry. The academic project to generate explanatory theories contrasts with the industrial one to develop working technologies and products. Industrial science is less committed to the development of theory.
Academic science reserves its greatest rewards for discoveries and the development of theories that explain the world. Industrial research, by contrast, receives compensation through markets for technologies that provide consumers with control over the world (Fleming and Sorenson 2004). There are obvious exceptions: the chemist who receives the Nobel Prize for a novel synthesis technique or the well-compensated biotech-based scientist whose acclaimed discoveries attract capital for her firm. These exceptions underscore the rule: that is, the chemistry technique is academically honored because it underlies basic discoveries, and the biotech researcher attracts capital insofar as her work is a credible indicator of market-relevant research at the firm. The internal organization of universities and companies reflects these reward structures. Academic departments are typically organized around modes of inquiry and explanation (e.g., ecology and evolution, physics), while divisions in industry cleave to product lines and their markets (e.g., industrial chemicals, portable electronics).
The process of building explanation differs from that of constructing technological control. In Latour’s account of academic rhetoric, he illustrates how academic scientists typically develop theory by constructing the most inclusive accounts possible. They do this because more universal theses interest more scientists and, if convincing, garner more scientific credit (Latour 1987, 51). For inventors of technology, the bar is both higher and lower. They must control but not necessarily understand, and often only in one particular setting. When Edison trolled through materials for his light bulb filament, he was searching for not a universal property but a particular one (Josephson 1959). He found it in high-resistance carbon without understanding why, and today we use tungsten, but it would not have mattered if the substance had been unique. This is because a technological component built on a unique material becomes widely connected within technological networks not by virtue of its representation of an external world but by its control over an internal one. Effective control of a particular substance is sufficient for technological advance but insufficient for science.
As a result, when academic and industrial researchers collaborate on an empirical project, they are more likely to produce theoretically unanticipated experiments than when academics from the same discipline collaborate with each other. This occurs for two primary reasons and two auxiliary ones. First, industry is neither primarily rewarded by nor committed to the development of theory.6 Companies bring to academic collaborations an interest in marketable technologies that may be discovered or validated by experiment but that might not have been academically anticipated or prioritized without that interest. For example, a biotechnology firm trying to capture a share of the $15 billion agrochemicals market may have an interest in genetically placing resistance to insects and fungi directly within a new generation of crop plants. This interest would suggest a sequence of experiments involving plant genetics and plant–environment interactions that might never have been conceived from prior theory.
The second key reason why industrial and academic collaboration can produce novel experiments is that because industry is not primarily committed to the development of theory, its researchers may profit from findings with technological implication even if they do not harmonize with existing theory. As a result, industry scientists place a greater weight on experimental findings than academics because success in one setting is not conditioned on success or failure in others. The annals of industry abound with groping experiments, lightly guided but not inhibited by theory. These range from Edison’s globe-trotting quest for an efficient lightbulb filament to DuPont’s search for the most functional polymer fiber and to genomics companies that assay millions of biocompounds for medical use. The vast majority of these low-probability searches end in failure, but a single success can be immediately useful for the development of technology. Academic scientists, for whom the most general findings receive the most acclaim, are less likely to benefit from a theoretically idiosyncratic discovery until a more inclusive theory is devised (Kuhn 1962). In the idiom of statistics, industrialists are frequentists, pleased to profit from the unexpected. Academics are Bayesians, slower to believe.
Auxiliary reasons for industry to push academics from theory include industry’s interest in appropriating the value that flows from developed technologies. The more publishing scientists contribute to the ideas underlying a new technology, the less likely exclusive control over that technology can be obtained, through legal patent or secrecy. It is in firms’ strategic interest to invest in areas with less competition and research precedent (Evans 2010a, 2010b). Businesses are also more likely to collaborate with universities when they are exploring new product possibilities than when they are exploiting existing product know-how in product development. This brings early, exploratory research from firms into contact with mature, theory-invested science in universities.7
Industry’s tendency to push collaborating academics away from their projects of developing theory underscores the internal strength of the theoretical project. A shared theoretical framework maximizes the coordination of scientists and facilitates the ability of one scientist to access the insights of others who are distant in the landscape of research. At the same time, a shared theoretical framework necessarily reduces the scope of things considered and enables the kinds of information cascades and consensus processes as does the World Wide Web. This is not primarily because a theory is itself a simplified view of the world but because it is shared. By limiting the assumptions, entities, and forces considered by a community of scientists, shared theory scaffolds the accumulation of findings and enables advance.
In the parlance of Bruno Latour, science is a complex network in which scientists combine with ideas, entities, instruments, places, money, and samples via “associations they knit, weave and knot” (1987, 94). In this view, academic science draws scientists together while speculative science, sometimes sponsored by industry, pushes them further apart. Academic networks often form dense knots—tightly fashioned subcommunities collaborating and competing around a theoretical–empirical conundrum. In this way the increased quantity of communication enabled by theoretical agreement allows researchers to reach further within the scientific system while narrowing the boundaries of that system.
The way in which theory acts as a protocol to increase the quantity of communication in science and raise the amount of convergence suggests a nonobvious hypothesis. It would lead us to expect that more complex empirical problems will lead not to more divergent but more convergent hypotheses as researchers rely on scientific agreements to make apparent, hard-won progress.
How the Scientific Theory Scaffolds Plant Biotechnology
I now describe the way in which theory increases communication within science and constrains its diameter in an expanding corner of molecular biology: research performed using the model plant Arabidopsis thaliana. Arabidopsis, a wild, purple-flowered relative of the mustard plant, was the first higher plant with a sequenced genome and has become the dominant genetic model organism in plant biology and agricultural biotechnology (Walbot 2000), just as the mouse and Drosophila (fruit fly) serve as genetic model animal organisms. Arabidopsis research has been used to probe fundamental biological questions, but also to pilot development of crops with social and profitable implications (e.g., drought-resistant corn, vitamin-A enriched rice).
Monsanto, the chemical and agricultural giant, was first to insert foreign genes into Arabidopsis (Lloyd et al. 1986). Zoëcon Research Institute, an insect-control division of Sandoz Crop Protection Corporation, and then DuPont expanded on the Monsanto demonstration to produce a vast number of genetically transformed plants in 1987. These companies were interested in potential products and pioneered exploratory techniques without theoretical understanding.
In contrast, academics from the same period sought to understand how these techniques worked and to justify Arabidopsis and plant molecular genetics as theoretically relevant to biology as a whole (Somerville and Koornneef 2002, 886). Chris Somerville and Bill Ogren established that mutations in any Arabidopsis gene could be isolated by screening a reasonably sized sample of seeds treated with chemical mutagens but proved its theoretical importance by using the method to expose the molecular mechanism of photorespiration (Somerville and Ogren 1979a, 1979b). Martin Koorneef followed this with a promising linkage map that showed, from the distances between genetic loci, that a number of scientifically interesting Arabidopsis mutants could be easily generated (Koornneef et al. 1983). Next, in 1984, the lab of Elliot Meyerowitz at Caltech demonstrated that the number of Arabidopsis genes was very small (Leutwiler, Houghevans, and Meyerowitz 1984). At about 70 megabases, the Arabidopsis genome is more than thirty times smaller than corn and a hundred times smaller than wheat, which made it relatively easy to clone genes and an appropriate model in which to explore more general plant processes. Somerville and Ogren’s academic contributions made a substantive advance, Koorneef’s generalized its method, and Meyerowitz’s put forward a powerful justification for Arabidopsis as a general model in which to efficiently study the processes of many biological systems. While Monsanto and Zoëcon took the speculative steps of actually transforming Arabidopsis, without knowing how, academics focused on understanding and theoretical relevance. This understanding scaffolded biological knowledge by rightly or wrongly generalizing discoveries made in Arabidopsis to all other plants or, in some cases, all eukaryotes.
The manner in which academic groups organize their respective Arabidopsis investigations suggests another scaffolding role of theory. The process of creating new theory in the molecular biology of photosynthesis involves the persistent recombination of ideas but it also usually involves new methods (e.g., cDNA microarrays), new biological materials (e.g., new mutant plants with parts of the photosynthetic apparatus inoperative), and new findings. With so many variables, academic scientists repeatedly turn to theory to help them think carefully about the limited sample of combinations they will attempt in their incremental search for insight into photosynthesis.
In contrast, companies like giants DuPont and Novartis, and small biotechs Ceres, Mendel, and Paradigm Genetics have taken a single scientific frame or research tool and then sampled across the entire population of combinatorial possibilities at substantial expense. DuPont incrementally knocked out thousands of Arabidopsis genes and identified economically promising mutants through a large number of molecular and environmental assays. Mendel, a biotech company started by several of the academic leaders in the field, did the same, but with a bit more theory to focus its search on transcription factors, the genetic control-switches that turn other genes on and off. Firms consequently selected academic partners whose work overlapped their own to scaffold their current knowledge with theory.
The written and oral history of industry involvement with Arabidopsis research underscores the different weight that theory placed on the shoulders of university- and company-based scientists. It also suggests how theory acted as a scaffold to infuse investigations with broader significance while limiting the range of experiments to be considered.
A large-scale analysis of the collaboration between academic and industry plant scientists that drew on tens of thousands of Arabidopsis articles8 demonstrated that industry collaboration and funding influenced academic scientists to become more exploratory in their experiments (Evans 2010a). After engaging in personal collaboration with industry, academics were more likely to discover and originally combine genes in novel ways and also creatively mingle methods and biological phenomena.
Industry funding also exerted a cumulative influence on the complete network of scientific genes, terms, and methods. This network was constructed by copresence of these elements in a published experiment. Industry sponsorship pushed genes, terms, and methods to the periphery of the network, away from the center of scientific attention and broad relevance. With some exceptions, government sponsorship did the opposite, more tightly knitting genes and methods into the existing fabric. This highlights the scaffolding influence of theoretical commitments. Government funding, allocated by peer review, made academics more beholden to each other, influencing them to know more about less. By knitting new experiments tightly around the edges of existing, theoretically informed hubs, government-sponsored academic research approached the unknown from the known—from the scaffolding base of existing theory. Industry-sponsored work neglected this theoretical scaffolding in pursuit of leads for marketable products. This highlights the limits of communicative scaffolding. By supporting one structure—in this case generalizable biological insight—it may fail to support others.
Social, Geographic, and Technical Protocols
Broadly shared communication protocols beyond WWW hyperlinks and theoretical commitments in science influence the volume of system-level communication, which in turn both broaden individual reach and narrow the global range of knowledge and culture.
Human societies vary in their cognitive institutions of knowledge and culture—initially from geographic dispersion and, more recently, status and specialization. Individuals’ and organizations’ own knowledge stocks broaden, however, when they share communication protocols and can exchange information with more and more distant others. The global distribution of knowledge contracts, however, as the “independent” production of different societies’ knowledge and culture mix and come to depend on one another.
Immigration and colonization have historically had this effect. The most consistent trend in analyses of modernization is convergence and the eradication of local difference. The central narrative in studies of liberal colonization is assimilation and the loss of indigenous culture (Lange, Mahoney, and Vom Hau 2006). After the U.S. colonized the Manu’a Islands, Samoans began to learn English, took up football and baseball, and immigrated to the United States in pursuit of advanced education and often permanent residence. They came to value what most other Americans valued. As they did so, awareness of Samoan language, indigenous knowledge, and culture naturally declined. Even in symmetrical contexts, where one individual or group has no more wealth or power than another, there remains additional value in watching a sports program, political debate, or DVD that everyone else watches. It enables conversation, taste validation, and solidarity.
Alternately, in oppressive or “extractive” colonization, where repression and social conflict between groups drives cultural conflict, knowledge and tastes still may become more equivalent than they would if they were not in argument with one another. For example, as twentieth-century conflict grew in Algeria between North African natives and the French colonists, the Algerians organized around a unified identity of shared language, Arabic, and religion, Islam, an alternative system of schooling, and even an informal “parallel” legislative body. In this way, as Algeria polarized, by making their native system in opposition to the French one, Algerians also made their system structurally similar to French institutions and national identity (Willis 1997; Ruedy 2005).
Early-twentieth-century urban researchers noted a comparable effect of geography in facilitating communication and scaffolding knowledge. With the rise of the metropolis and the breakdown of distance, Simmel noted “the atrophy of individual culture through the hypertrophy of objective [mass] culture” (Simmel and Wolff 1950).9 Increased human density and mixing effectively increased human reach and resulted in cultural speciation (Park, McKenzie, and Burgess 1925). New, urban subcultures emerged, grew, and accumulated gravity—drawing individuals with some features of commonality into communication. In this way, city communication increased cultural clustering and so likely decreased variation in formerly independent behaviors and identities.
The reduction of class distinctions associated with the rise of democracy has increased the quantity of communication with similar effects. Consider the social interchanges facilitated by the breakdown of class barriers in early U.S. democracy (Tocqueville 1840/1956). While this invariably extended the influence of most individuals, it also enabled each individual more exposure to the crowd. De Toqueville’s analysis of American democracy characterized this as the greatest cost associated with the breakdown of class barriers and the democratic interchange:
At periods of equality, men have … almost unbounded confidence in the judgment of the public; for it would not seem probable, as they are all endowed with equal means of judging, but that the greater truth should go with the greater number…. Under the dominion of certain laws, democracy would extinguish that liberty of the mind to which a democratic social condition is favorable; so that, after having broken all the bondage once imposed on it by ranks or by men, the human mind would be closely fettered to the general will of the greatest number.... a new physiognomy of servitude (Tocqueville 1840/1956, 148–149).
This argument suggests a Bayesian calculus. As more voices in public life share the same position, individuals give the collective position greater weight in making their own calculations. This results in an information cascade (Anderson and Holt 1997; Anderson 2001) that can, at its limit, end in mob rule.
A similar phenomenon is described by John Meyer and colleagues as the rise of “world society” (Meyer in Krücken and Drori 2009). This is a world in which globally integrative organizations, widespread communication, and transportation drive a few modal cultural forms to dominate the global sphere and drive convergence in everything from schools (Meyer 1978) to social movements (Hafner-Burton and Tsutsui 2005) to nation states (Meyer, Boli et al. 1997) to a stance toward the environment. The quantity of global communication results in an increase in individual reach across the world system, coincident with a shrinking of the diversity in that system.
Standards
The quantity of communication increases with the spread of standards of virtually any type, including telegraphy and telephony, measurement (Levinson 2006), financial accounting (Jang 2005), and disease classification (Bowker and Star 1999). Because standards often act as communication protocols, synchronization in any one broadens individual and organizational knowledge horizons by reducing the cost and decreasing the ambiguity of interaction with others. This allows individuals to more selectively sample those with whom they choose to communicate.
Standardized communication protocols do more than merely broaden individual and organizational horizons: they globally expand the field or “market” of interaction. When English emerged as the dominant scientific language following World War II, the pool of knowledge producers and consumers who could read one another’s work increased. English prose and poetry has similarly broadened its international circulation. Wide adoption of the metric system in nineteenth-century Europe did the same for interstate commerce, as did global adoption of standardized containers for shipping (Levinson 2006) and the International Classification of Disease for monitoring illness and outbreak (Bowker and Star 1999).
Once knowledge becomes distributed broadly, opportunities for analysis and recombination become possible. Standardized commercial and medical data, for example, gave rise to the economic analysis of trade flows and the fields of epidemiology, public, and population health. These fields, in turn, produced economic and health policies tuned to further standardize trade and health behaviors. The widespread use of diatonic scales in popular Western music once made it easy for arrangers to construct medleys just as it now enables disc jockeys and film scorers to trivially mash up recordings into novel tracks.
The recent standardization of databases for biological information, like gene sequences and protein structures, highlight another consequence of standardization. Databases like GeneBank and Swiss-Prot have enabled the rise of bioinformatics, a field in which scientists draw upon large samples of standardized data to make statistical inference about biological phenomena at larger scales. Not all of the uneven knowledge held by particular protein laboratories, however, can be captured in Swiss-Prot’s standard database fields or described with its restrictive syntax. Standardization of virtually any type extends individual reach and narrows global range, isolating unstandardized knowledge that is unusually deep, inconsistently held, or local.
In this way, virtually every protocol or standard that increases the quantity of system-wide communication integrates the knowledge and culture of that system in such a way that it increases individual reach and reduces global span. This scaffolds knowledge and culture by allowing members to flexibly combine and reason over more system units, ultimately knowing more. In this way, the quantity of communication scaffolds and shapes the development of news information, fashion, rumor, and religion. In fact, the effect is so consistent that it is doubtful whether it is only applicable to human communication or also relevant, at some level, to herding and swarming behavior in a variety species, from rodents to quorum sensing bacteria.
Quality of Communication
Beyond the quantity of communication, the quality or structure of communication also influences the knowledge and culture it scaffolds. Dense communication inside nested scholarly subfields facilitates efficient specialization within the context of an existing disciplinary framework. Inconsistently connected scientists can enable interdisciplinarity and the diffusion of ideas from one domain to another. But social structures not only constrain and enable the flow of information. They also act as patterns that can be evaluated in reference to patterns within the communicated ideas themselves. In this way, communicative structures can make some scientific ideas seem more plausible, and other ideas less so. The relationship between social structures and the knowledge they scaffold was developed by Durkheim in The Elementary Forms of the Religious Life (Durkheim and Swain 1915). In his analysis of Australian aborigines and the American Pueblo Indians, Durkheim suggests that the structure of human religion reflects a worship of the political structure of society. Guy Swanson extended this argument empirically in his analysis of tribal pantheons and the decision of European polities to remain Catholic or shift to Protestantism during the Reformation. Swanson shows that the centralization of political order in tribal life affected the degree to which tribes’ religious pantheons were centralized and whether they possessed a high god. He also shows that sovereignty in European governments determined the likelihood that a country would retain Catholic or accept a Protestant faith with its decentralized religious authority. In this way, Durkheim and Swanson suggest that the structure of social and communicative life both guide and, in the case of the Reformation, confirm hypotheses about the spiritual order of the world.
The importance of homology between communicative structure and scientific knowledge became important in the first half of the twentieth century. This occurred in the wake of claims from totalitarian regimes including Nazi Germany and the U.S.S.R. that they were based on and justified by modern science. This appeal involved, in the case of the U.S.S.R., a rejection of religion as a source of higher knowledge, and in both a deployment of scientific practices like Taylor’s scientific management and mass media experiments to organize their governance and conquest. Moreover, a command-and-control approach to science was successful and resulted in rapid advances, both in military and medical investigation (Proctor 1999). Totalitarian claims to scientific authority and compatibility inspired a generation of thinkers from democratic regimes to justify the compatibility of democracy and science.
One attempt involved Karl Popper’s two-volume work The Open Society and Its Enemies (Popper 1945), in which Popper criticized Plato’s fantasy of a philosopher king and argued that only liberal democratic regimes could bring about institutional improvements without bloodshed and so were more compatible with long-term progress.
Another approach involved Robert K. Merton’s characterization of science as a social system with its own distinctive culture, in which the free flow of scientific ideas and findings was central. Among the four institutional imperatives that Merton claimed govern scientific activity, three directly mandate the free flow of knowledge given and received: all discovered knowledge (the norm of disinterestedness) should be shared (the norm of “communalism”) freely and without distinction (the norm of universalism). The fourth norm, organized skepticism, allowed everyone a negative vote or veto on which scientific claims were meritorious.
In Merton’s system, the ultimate reward in science—credit—was obtained through recognition of priority (1957). Priority functions like an invisible hand in a free market of ideas: scientists share knowledge universally by publishing in their own self-interest (see also Dasgupta and David 1994; Polanyi 1962). In this way, Popper and Merton rendered science as both homologous with and best fostered by liberal democracy and free markets.
Beneath this debate over what type of regime could claim the authority of science was the deeper issue that different regimes had inspired different kinds of science. Hierarchical regimes had created hierarchical science institutions, which, for targeted research projects like cancer reduction or ballistics advance, had been extremely efficient. Democratic regimes had fostered diverse scientific investigation that were less efficient for achieving targeted, short-term goals but explored the promise of a wider range of long-term scientific possibilities.
I argue that this relationship between the structure of communication and the process of science can be extended to the content of particular scientific hypotheses, theories, and orientations. For example, because liberal democracies posit the importance of wide public participation, the existence of these regimes and the participation of scientists within them increase the likelihood that “diversity” will be valued, proposed, and confirmed in other contexts—from biology and anthropology to management (Edelman, Fuller, and Mara-Drita 2001). To someone who lives in a democracy and values it, the scientific answer of “diversity” will likely feel right, regardless of the question.
In a similar way, contemporary mathematical models often favor stochastic processes, which mirror the uncertainty and diversity of modern culture and society. Mathematical models in the early twentieth century and before were much more often deterministic, reflecting religious and political certainties. Perhaps even the outcomes of interdisciplinary science are so lauded today in part because they are seen to match with the liberation of constraining social and communicative structures. Paul Forman suggested a similar type of influence when he argued that cultural values of Weimar Germany inspired the character and lessons ascribed to quantum mechanics (Forman 1971, 1984). In this way, patterns of communication and social structure may unconsciously inspire broad hypotheses to science. More importantly, these patterns may also help to resolve which hypotheses are viewed with greatest confidence.
When multiple explanations in science appear equally likely, researchers often appeal to Occam’s razor, or the principle that the simplest explanation is most plausible. The minimum description length principle, defined by information theorist Jorma Rissanen, is an attempt to formalize Occam’s razor. The explanation that can be encoded with the least information is the most theoretically parsimonious and is to be viewed as most likely. Many researchers use an informal version of these criteria in their evaluation of alternative hypotheses. The assumptions associated with familiar or ubiquitous features of modern life, however, like democracy or uncertainty, will conceal more of their underlying assumptions and so involve a shorter description than patterns less familiar. Moreover, it is efficient and evolutionarily likely for frequent, familiar things to receive the shortest descriptive encodings in language (Zipf 1932, 1949; Lieberman et al. 2007). This gives familiar patterns a special advantage when they are accounted more simple and more plausible.
These same principles are likely amplified in the context of culture and taste where the mandate to be correct is replaced by one to be consistent with some standard. Even in the context of Enlightenment aesthetics, most notably Hume’s “Of the Standard of Taste” and the related work of Joseph Addison and Francis Hutcheson in which aesthetic judgments reflect moral ones and are not entirely subjective, the importance of “impression” and “imaginative association” suggest that internal consistency plays at least as strong a role as external validity. As a result, cultural institutions layer upon one another based largely on their perceived consistency, in a manner similar to Durkheim and Swanson’s account of religion.
Conclusion
I have argued that human communication scaffolds knowledge in two consequential ways. Shared communication protocols increase the quantity of communication in a system, which both increases an individual’s reach to knowledge and culture across that system but also decreases the diameter and diversity of knowledge and culture as a whole. This increases the developmental capacity of the system by enabling newcomers to traverse and know more, faster. Insofar as communication protocols increase the speed at which distributed information can be accessed, they also allow people to think with more elements of the system, reinvesting capacity spent remembering into reasoning. This comes at the cost of persistent variation, which could be better enabled through isolation.
Qualities of communication also scaffold knowledge and culture by making certain messages, which appear more compatible with those qualities, seem more plausible or resonant than others. In this way, the structure of communication not only enables information to be distributed, efficiently stored in and accessed through others, but it enables information to be stored in itself. Together, these represent powerful processes through which features of communication support, constrain or undercut the knowledge and culture that exist atop them.
Acknowledgments
I am grateful to the workshop organizers for creating a stimulating venue for scaffolding ideas about scaffolding and for their superb comments to drafts along the way.
Notes
References
