Direct and Indirect Implications of Pathogen Prevalence for Scientific and Technological Innovation

Disease Threat and Cultural Variability

Infectious disease has represented one of the largest threats to human survival throughout history (e.g., Inhorn & Brown, 1990; Wolfe, Dunavan, & Diamond, 2007). Humans thus have an array of mechanisms designed to mitigate this threat. One set of mechanisms comprises the physiological immune system. Although this immune system is generally effective, its activation is metabolically costly and can be temporarily debilitating. To minimize costly activation of the immune system, humans (and other animals) are also equipped with psychological defense mechanisms that inhibit disease transmission. These defenses include a set of affective and cognitive mechanisms that respond to disease cues in the local environment by promoting behavioral responses designed to inhibit contact with these potentially infectious objects (for review, see Schaller, 2011).

These psychological defenses have implications for an array of attitudes, values, and interpersonal behaviors. For example, a consistent finding to emerge from this literature is that xenophobic and prejudiced responses to foreign individuals may be due in part to disease-avoidant processes. These findings are theoretically underpinned by the fact that physiological immune systems are highly localized, and thus contact with out-group members increases the likelihood of exposure to novel pathogens. Indeed, xenophobic responses are exaggerated for people who are chronically worried about disease transmission (Navarrete & Fessler, 2006), for people whose immune system is temporarily compromised (Navarrete, Fessler, & Eng, 2007), and for people who are experimentally made aware of disease threat (Faulkner, Schaller, Park, & Duncan, 2004).

Conformist attitudes and behaviors are similarly affected by the perceived threat of disease. Conformity has disease-mitigating benefits: A large proportion of cultural norms (especially those pertaining to food preparation, hygiene, and sex) serve to buffer against disease transmission (e.g., Fabrega, 1997). However, conformity to established norms also carries costs, such as inhibiting innovative problem solving (e.g., Mokyr, 1991). The logical upshot is that conformist attitudes and behaviors should be positively predicted by the perceived threat of infectious disease. Indeed, individuals who are chronically more worried about disease transmission report more conformist attitudes, as do individuals from whom disease has been made temporarily salient (Murray & Schaller, 2012; Wu & Chang, 2012).

Just as individual attitudes vary in response to the immediate salience of infectious disease, so do group-level value systems vary in response to the real threat of disease within the local ecology. Numerous cross-cultural investigations have revealed effects of disease threat that are theoretically convergent with studies at the individual level of analysis. For example, higher regional disease threat positively predicts several measures assessing xenophobia and decreased frequency of out-group contact (Fincher & Thornhill, 2008a, 2008b; Schaller & Murray, 2010). Investigations reveal convergent effects of disease prevalence on conformity-related constructs too. Disease prevalence positively predicts several measures assessing behavioral and attitudinal conformity (such as average effect sizes on Asch-type experiments; Murray, Trudeau, & Schaller, 2011). It similarly predicts lower dispositional openness (across three independent cross-cultural personality investigations, Schaller & Murray, 2008) and cultural value systems associated with conformity and traditionalism (Fincher, Thornhill, Murray, & Schaller, 2008). These effects are evident in political attitudes as well: Higher disease prevalence predicts higher dispositional authoritarianism within a country and more institutional restrictions on political and individual freedoms (Murray, Schaller, & Suedfeld, 2013; Thornhill, Fincher, & Aran, 2009).

Disease Threat and Innovation

Although the above findings highlight the fact that regional variation in disease prevalence may partly account for cultural differences in numerous constructs pertaining to conformity and xenophobia, what has received less attention are the consequences of this psychological variation. How might this disease-caused psychological variation manifest in downstream cultural differences?

One such set of outcomes that may be affected by disease threat are variables pertaining to technological and scientific innovation. Key psychological variables conducive to innovation and invention include low traditionalism, low conformity, and low wariness of outsiders (Herbig & Dunphy, 1998)—variables that are all negatively predicted by disease prevalence. This leads to the logical hypothesis that higher threat of infectious disease within a local ecology should negatively predict inventive and innovative outputs.

These variables—traditionalism, conformity, and xenophobia—are three of the major dimensions upon which the distinction between collectivist and individualist societies are based. Whereas collectivistic societies tend to be characterized by values emphasizing traditionalism, conformity to group norms, and wariness of outsiders, more individualistic societies generally place lower value on tradition and conformity and have a looser boundary that separates the coalitional in-group from out-group (Cukur, De Gusman, & Carlo, 2004; Gelfand, Bhawuk, Nishii, & Bechtold, 2004; Oishi, Schimmack, Diener, & Suh, 1998; Sagiv & Schwartz, 1995). Several scholars have noted the strong positive relationship between individualism (vs. collectivism) and innovation across cultures, and most speculate that the values encapsulated by individualism are a cause, rather than a consequence, of higher innovation and invention (Barnett, 1953; Beteille, 1977; Shane, 1992; but for an alternative interpretation, see Hofstede, 2001). Therefore, the negative relationship between disease prevalence and innovation may be mediated by variation in collectivism and individualism.

Of course, disease prevalence shares relationships with other variables that influence innovation. Bidirectional relationships exist between disease prevalence and several demographic variables that influence innovation: Greater wealth and education decrease disease burden through greater access to, and proficiency of, medical systems; conversely, higher disease prevalence decreases average wealth, healthy life expectancy, and value placed on education within a country (Bloom & Canning, 2000; Gallup & Sachs, 2001; Sachs & Malaney, 2002)—all demographic features that negatively predict innovation across cultures. Therefore, examining whether disease prevalence predicts innovation independently of the effects of wealth, education, and healthy life expectancy provides an especially rigorous test of whether the psychological effects of disease prevalence uniquely account, at least in part, for cultural variation in innovation.

Although disease threat may have relatively immediate effects on psychological variables such as attitudinal conformity (Murray & Schaller, 2012; Wu & Chang, 2012) and xenophobia (Faulkner et al., 2004), its effects on innovation may play out over a relatively longer period of time. Innovative output requires not just a psychological disposition conducive to innovation; it requires cultural values and institutions that encourage innovation as well, and the development of these values and institutions does not happen instantaneously. The logical upshot is that changes in innovative output due to changes in disease prevalence may only become apparent over the course of generations, rather than weeks or months—an issue that is considered further in the “Discussion” section.

Overview of the Present Study

Despite the relatively large literature exploring the antecedents of cultural variation in innovation, no study to date has rigorously explored the influence of disease prevalence—be this influence direct or indirect—on innovative output. Innovation and invention can be defined and measured in several different ways. Therefore, five separate measures designed to assess a country’s innovative output were used for the current study. These measures include innovation in academics and science, technology, and business. The first analyses investigate uncorrected relationships between innovation, disease prevalence, and other purported causes of cross-cultural variation in innovation. For comparative purposes, these analyses include both a historical and a contemporary measure of disease prevalence. Further analyses investigate the unique predictive effects of disease prevalence on innovation when controlling for other key causal variables. Final analyses test whether variation in collectivism (and, finally, conformity) mediates the effects of disease prevalence on innovation. Overall, these analyses test the two complementary hypotheses that (a) higher threat of disease negatively affects innovation, independently of disease threat’s implications for wealth, health, and education, and (b) these impacts of disease prevalence on innovation are mediated by variation in values and behaviors associated with individualism/collectivism.

Method

For all central analyses, geopolitical regions served as the units of analysis. The majority of these regions were countries, but the sample also included several culturally distinct regions or territories within a country (e.g., Hong Kong). Although geopolitical borders are not perfectly analogous to cultural borders, much evidence indicates that geopolitical entities serve as useful proxies for societal cultures (e.g., Schwartz, 2004).

Five different measures of innovation were used for the current investigation.

Results

The conceptual hypothesis predicts a negative relationship between pathogen prevalence and innovation. Zero-order correlations between historical pathogen prevalence and each of the indicators of innovation are summarized in Table 1. As predicted, pathogen prevalence was significantly negatively correlated with the total innovation score, r = −.69, p < .001 (shown in Figure 1). Pathogen prevalence was also significantly negatively correlated with each of the five markers of innovation, rs ≤ −.49, ps < .001. Human development was positively correlated with each measure of innovation, rs ≥ .62, ps < .001, as was proportion of population living in cities, rs ≥ .37, ps < .001.

OPEN IN VIEWER

Table 1.

Zero-Order Correlations Between Historical Pathogen Prevalence, Human Development, Proportion of Population Living in Cities, and Innovation Scores.

	Historical pathogen prevalence	Contemporary pathogen prevalence	Human development	Proportion in cities
Historical pathogen prevalence	—
Contemporary pathogen prevalence	.83	—
Human development	.73	−.75	—
Proportion living in cities	−.45	−.46	.71	—
Innovation score (Total)	−.69	−.61	.77	.56
Nobel laureates	−.75	−.73	.63	.55
Global Innovation Index	−.70	−.68	.84	.67
Technological achievement	−.73	−.71	.90	.66
Innovative capacity	−.49	−.44	.62	.47
Patent applications	−.61	−.55	.76	.37

Note. All ps < .001.

OPEN IN VIEWER

Figure 1.

The correlation between historical pathogen prevalence and innovation (r = −.69, p < .001).

For sake of comparison, zero-order correlations between a measure of contemporary pathogen prevalence and innovation are also reported in Table 1. This measure of contemporary pathogen prevalence was obtained from the Global Infectious Disease Epidemiological Online Network (GIDEON), which reports detailed epidemiological information around the globe (for further information about this measure, see Fincher et al., 2008). This measure of contemporary pathogen prevalence was significantly associated with all of the measures of innovation, rs ≤ −.44, ps < .001. However, in all cases the strength of this relationship was smaller than each innovation measure’s relationship with historical pathogen prevalence. More importantly, when the unique predictive effects of both measures were tested by entering them into a multiple regression predicting total innovation scores, historical pathogen prevalence emerged as a far stronger predictor (β = −.54, p < .001) than did contemporary pathogen prevalence (β = −.17, p = .09). Given the temporal precedence of the historical measure, this pattern of results is consistent with the current disease threat hypothesis, and inconsistent with a reverse-causal explanation.

Complementary tests examined the relationship between historical pathogen prevalence and innovation using six world regions as the units of analysis. These tests produced results similar to those obtained at the country level. Even with this small sample size, pathogen prevalence was significantly associated with the total innovation score, r = −.95, p = .004. Replicating results obtained at the country level, pathogen prevalence was similarly highly associated with each of the five innovation measures individually, rs > −.74, ps < .09. Given the very small sample size of these analyses, however, the magnitude of these correlations should be interpreted with caution.

A multiple regression analysis investigated the simultaneous unique predictive effects of historical pathogen prevalence, human development, and proportion of population living in cities on innovation. Although these three predictors overlapped substantially, results revealed that pathogen prevalence uniquely predicted total innovation scores, β = −.29, p < .001, as did human development, β = .57, p < .001. Proportion of a country’s population living in cities did not uniquely predict innovation, p > .50. Further regression analyses investigated the unique effects of these three variables in predicting each of the five constituent innovation scores individually. Results of these analyses are reported in Table 2. As can be seen from the table, pathogen prevalence uniquely predicted four of the five innovation variables (all but innovative capacity), ps ≤ .002. Human development significantly uniquely predicted all five innovation variables, ps < .05. Proportion living in cities significantly uniquely predicted one of the five measures (patent applications per capita, p = .01); however, this predictive effect was opposite to that which would be predicted by previous scholarship, with a higher proportion living in cities negatively predicting number of patent applications when controlling for pathogen prevalence and human development.

OPEN IN VIEWER

Table 2.

Results of Multiple Regression Analyses Identifying Unique Predictive Effects of Historical Pathogen Prevalence, Human Development, and Proportion of Population Living in Cities on Indicators of Innovation.

Innovation indicator	Historical pathogen prevalence		Human Development Index		Proportion in cities
	β	p	β	p	β	p
Total innovation score	−.29	<.001	.57	<.001	−.03	.68
Nobel laureates (log transformed)	−.49	<.001	.35	.02	.01	.96
Global Innovation Index	−.23	.002	.61	<.001	.05	.51
Technological achievement	−.21	.008	.73	<.001	.01	.89
Innovative capacity	−.10	.37	.53	<.001	.02	.88
Patent applications (log transformed)	−.26	.002	.75	<.001	−.21	.01

Note. Two-tailed p values.

Next, mediation analyses investigated whether variation in individualism/collectivism mediated the relationship between pathogen prevalence and innovation scores. Path coefficients were determined by regression analyses, and indirect effects and their 95% confidence intervals (CIs) were determined via the SPSS bootstrapping procedure recommended by Preacher and Hayes (2004), based upon 10,000 bootstrapped samples. Results of these mediation analyses are summarized in Table 3. Across all six analyses, none of the CIs for these mediated effects contained zero, indicating that collectivism significantly mediated the effects of pathogen prevalence on innovation in each model. In five of these six analyses, the direct effect remained significant (p < .05; only in the model predicting innovative capacity was the direct effect of pathogen prevalence nonsignificant), suggesting that the impact of disease prevalence on innovation is not fully accounted for by the indirect effect of collectivism.

OPEN IN VIEWER

Table 3.

Standardized Path Coefficients, Bootstrapped Mediated Effects, and Confidence Intervals (CIs) for the Indirect Effects of Collectivism on Innovation.

Dependent variable	Standardized path coefficients				Mediated effect and CIs
	α	β	τ	τ′	BMED	95% LL	95% UL
Total innovation score (n = 95)	.66**	−.52**	−.70**	−.36**	−.34	−.47	−.24
Nobel laureates (n = 61)	.68**	−.40**	−.80**	−.53**	−.27	−.43	−.15
Global Innovation Index (n = 91)	.68**	−.50**	−.68**	−.34**	−.34	−.48	−.22
Technological achievement (n = 57)	.72**	−.53**	−.70**	−.32*	−.38	−.59	−.18
Innovative capacity (n = 91)	.67**	−.63**	−.48**	−.06	−.42	−.59	−.29
Patent applications (n = 70)	.71**	−.46**	−.60**	−.27 †	−.33	−.55	−.17

Note. α = path from pathogen prevalence to collectivism; β = path from collectivism to DV; τ = unmediated path from pathogen prevalence to DV (within set of countries used for given model); τ′ = direct effect of pathogen prevalence on DV after accounting for mediated effect; BMED = bootstrapped mediated effect; LL = lower limit; UL = upper limit; n = number of countries available for each model.

†

p < .05. *p < .01. **p < .001.

Finally, given that collectivism is a broad, multifaceted construct, a final model tested whether conformity specifically (rather than just collectivism more generally) might account for a significant proportion of the relationship between disease prevalence and innovation. To examine this more specific hypothesis, a double mediational model tested the simultaneous, unique indirect effects of both collectivism and conformity on total innovation scores. This analysis was based upon a total of 75 countries for which all data were available. As above, path coefficients were determined by regression analyses, and indirect effects and their 95% CIs were determined based upon 10,000 bootstrapped samples. Results from this analysis revealed that conformity significantly mediated the relationship between disease prevalence and innovation (bootstrapped mediated effect = −.14, 95% CIs = [−.27, −.05]), with conformity accounting for 21% of the total effect. Similarly, collectivism significantly mediated the relationship between disease prevalence and innovation (bootstrapped mediated effect = −.40, 95% CI’s = [−.55, −.28]), with collectivism accounting for 60% of the total effect. In this double mediational model, the direct effect of pathogen prevalence was no longer significant, β = −.12, p = .25. These results provide further evidence that pathogen prevalence’s effects on conformity—in addition to its effects on other facets pertaining to collectivism—have unique implications for innovation.

Discussion

The results can be summarized as follows: Regional variation in historical pathogen prevalence is associated with cross-cultural differences in scientific and technological innovation. This relationship is independent of other purported causes of innovation, such as healthy life expectancy, average wealth, education, and population structure. That pathogen prevalence remained significant when controlling for these markers of human development provides especially strong evidence of pathogen prevalence’s relationship with innovation, given that these human development variables are highly associated with pathogen prevalence as well (and, not surprisingly, human development was also strongly associated with innovative outputs). This pattern of results emerged for a composite measure of innovation comprising diverse measures of innovation and also emerged for all but one of these innovation variables when analyzing them individually.

These results make at least two novel contributions. First, they demonstrate that the impact of disease threat on attitudes and personality may have downstream consequences for domains well beyond those pertaining to individual psychosocial functioning; disease threat may have implications for a culture’s propensity toward creative outputs, Nobel Prize–worthy scientific research, and technological innovation. Second, the current set of results suggests that disease threat may influence innovative outputs via two distinct pathways: through its effects on demographic variables such as economic development, education, and healthy life expectancy as well as through influencing attitudes toward new ideas, conformity, and out-groups.

Mediational results suggest that one psychological mechanism by which disease threat affects innovation is through cultural value systems—specifically, through values that are associated with individualism or collectivism. These results are consistent with previously documented relationships between higher individualist values within a culture (which encourage nonconformity and nontraditionalism) and innovation. Further mediational analyses demonstrated that conformity specifically—concurrently with collectivist value systems more generally—partly accounts for the relationship between disease prevalence and innovation. The current results highlight how disease threat affects the costs of cultural values and behavioral dispositions and that these effects may have downstream consequences for other cultural outcomes. These results complement recent research showing that pathogen prevalence influences individual authoritarian attitudes, and that these attitudes have downstream consequences for authoritarian political systems (Murray et al., 2013). The current study demonstrates how varying costs and benefits across cultures may have a different type of downstream consequence: Although values emphasizing traditionalist and conformist dispositions may mitigate the threat of infectious disease (which may be especially important in high-disease ecologies), these characteristics also incur the cost of fewer opportunities for creativity and innovation.

A few limitations of this study deserve note. First, while the models and conceptual hypotheses are clearly causal, these data are necessarily correlational. Cross-cultural investigations of this type cannot be experimental in nature—individuals cannot be randomly assigned to cultural groups. A reverse-causal explanation for these relationships remains possible, wherein higher levels of innovation within a country lead to lower levels of disease. For example, higher historical within-country levels of innovation may have lead to regional improvements in sanitation, healthcare, pharmacology, and nutrition, all of which lead to lower levels of contemporary disease prevalence. Alternatively, levels of both innovation and disease prevalence may be the consequence of some other third variable. Economic development, for example, most likely leads to higher expenditures on technological research and development concurrently with higher expenditures on areas that influence public health. These explanations need not operate at the exclusion of one another, and some degree of bidirectionality likely exists. That historical (rather than contemporary) pathogen prevalence emerged as the stronger predictor of innovation is consistent with the hypothesis that disease is a cause, rather than a consequence, of innovation. However, these results alone do not warrant firm causal conclusions about the relationship between pathogen prevalence and innovation.

A second consideration with these results is that they are derived from data that use nation states as independent units of analysis. Although countries can serve as useful proxies for societal cultures in contemporary cross-cultural investigations (see Schwartz, 2004), neighboring countries can hardly be considered functionally independent. This statistical nonindependence can spuriously inflate the magnitude of relationships between variables in cross-cultural analyses; therefore, the magnitude of the relationships between innovation and disease prevalence (as well as between innovation and its other predictors) may be attenuated due to this nonindependence. However, a relationship between disease prevalence and innovation also emerged at the more-independent world region level of analysis—a pattern of results that suggests this conceptual relationship is not simply statistical artifact. Furthermore, several previously documented relationships between disease prevalence and psychological outcomes using countries as units of analysis—including the relationship between disease and collectivism—have now been replicated in investigations using traditional, small-scale societies (Cashdan & Steele, 2013; Murray et al., 2013). Such convergent results provide further reassurance that these country-level results are not spurious statistical artifacts.

Although this set of results suggests that disease threat has a causal impact on innovation, what is not clear from these data is the mechanism by which disease influences variation in psychological variables conducive to innovation. At least four such mechanisms are possible. One possible mechanism is purely facultative, wherein individuals respond to disease threat within their immediate environments in innovation-inhibiting ways. Alternatively, this variation may be more genetically based, whereby historical variation in pathogen prevalence led to differential selection for genes associated with innovative dispositions. A different type of genetic mechanism—an epigenetic mechanism—is also a possibility, wherein genes associated with traits such as collectivism and conformity are less likely to be developmentally expressed in regions of lower disease. Finally, this variation may have arisen wholly through cultural transmission, wherein the most locally adaptive sets of traits, norms, and cultural values (many of which affect innovative output) are selectively communicated across generations (see Schaller & Murray, 2011, for a detailed discussion of the evidence for each of these mechanisms in producing cross-cultural differences).

Exactly when changes in disease prevalence may result in changes in innovation depends upon which of these mechanisms is primarily at play. A facultative mechanism would entail a time lag of months or years; a genetic mechanism would require a lag of at least several generations; an epigenetic mechanism would require just a one-generation time lag; a cultural transmission mechanism could see changes accrue across one to several generations. These mechanisms may operate in parallel: Epigenetic processes, for example, may give rise to a generation of individuals who are psychologically inclined toward innovation. That generation may, in turn, establish institutions conducive to innovation, and be more likely to culturally transmit values that encourage innovation within future generations. In the current set of results, a measure of historical disease prevalence (from about four generations ago) emerged as a better predictor of contemporary innovation than did a measure of contemporary disease prevalence—a pattern of results consistent with other research exploring the potential cultural consequences of disease threat (e.g., Fincher et al., 2008; Murray et al., 2011). Although still largely speculative, this pattern suggests that the time lag between changes in disease prevalence and appreciable changes in innovation may be most appropriately measured in generations, rather than in weeks or months.

This generational time frame would also be consistent with other culture-level research exploring the relationship between external variables and innovation-relevant outputs. For example, some analyses suggest that events such as revolts and revivals of culture and language most strongly predict creative output in the next (rather than the current) generation (e.g., Simonton, 1975). Similarly, in a transhistorical analysis of Japan, Simonton (1997) found that the influx of foreign ideas and peoples best predicted levels of “creative achievement” after a time lag of two generations. This time frame is also loosely consistent with the multi-generational changes in cultural values associated with immigrant populations that move from countries of high disease prevalence to countries of low disease prevalence (e.g., Hardyck, Petrinovich, & Goldman, 1976). Although global cross-cultural disease prevalence measures from these one- or two-generational time frames are not currently available, future research may reveal that this is the most consistent time lag between changes in disease prevalence and innovation as well.

The current results suggest that disease prevalence indirectly affects innovation; however, disease is not the only ecological factor influencing variables that affect innovation, nor is it likely that disease operates independently of these other ecological influences to produce cultural outcomes. Several other compelling explanations for the origins of many cultural differences have recently been put forth. Van de Vliert (2006), for example, provided data showing that the interaction between climate and wealth affect several variables pertaining to human freedoms. Similarly, Varnum (2013) showed that both frontier settlement patterns and pathogen prevalence predict patterns of nonconformist voting. Alternatively, Hackman and Hruschka (2013) suggested that variation in life-history strategies may account for regional variation in value systems. These different explanations need not operate at the exclusion of one another, and they may each operate at different explanatory levels of analysis. Single-factor explanations of cultural differences are at best short-sighted, and at worst a hindrance to scientific progress; the next step in research investigating the origins of cross-cultural differences will be to test increasingly complex causal models as more exhaustive and reliable data become available. These models will not just include more predictor variables but will likely contain multistep pathways as well. For example, whereas some recent work suggests that higher pathogen prevalence negatively predicts cognitive development cross-regionally (Eppig, Fincher, & Thornhill, 2010, 2011), other work directly implicates pathogen prevalence in inhibiting the development of institutions that foster cognitive development (Gallup & Sachs, 2001). However, the causal interplay between these individual and institutional variables remains poorly understood. Such future models, then, may also help to shed light on the empirical question of how individual-level attitudes and predispositions causally affect institutional and societal values and vice versa—a question that has been the subject of interest and speculation across the behavioral sciences (e.g., Bowler & Donovan, 2002; Jost, Glaser, Kruglanski, & Sulloway, 2003; Hofstede, 2001; Lavine, Hodge, & Freitas, 2005). However, the size of models that can be evaluated with current datasets is very finite; adequately testing such complex models will require researchers to find increasingly creative datasets or to use increasingly creative methods in evaluating current datasets.

These results add to the growing body of research demonstrating that the threat posed by infectious disease may be responsible, at least in part, for cross-cultural variation in attitudes, cultural value systems, intergroup relations, and contemporary political systems. Together with the current results, this emerging literature highlights the fact that widespread disease eradication and vaccination programs may have implications that extend well beyond the impacts on physical health and well-being. The current results suggest that such health programs may have implications for conformist and xenophobic attitudes, which in turn may have downstream implications for technological and scientific innovation. These social and economic benefits of lowering the prevalence of infectious disease may eventually deserve consideration when weighing health policy decisions as these ancillary benefits become clearer.