28. See pages 199–200 of Martin (2002), as well as Sessa (2008) and references therein. See also Grim (2009) and Isaacson (2011). A Macintosh computer is part of the collection of the New York Museum of Modern Art. See the “Apple, Inc.” page at https://www.moma.org/collection/works/142218.
29. See Harman et al. (1966).
30. Attributed to Ovid as cited in Sessa (2008), but perhaps apocryphal. A related statement can be traced to Ovid’s contemporary Horace (Epistles, Book I, Epistle XIX): “No verses which are written by water-drinkers can please, or be long-lived.”
31. See Jarosz et al. (2012). Other studies are discussed on pages 116–117 of Bateson and Martin (2013).
32. See Rees (2010).
33. See Holberton (2005).
34. See Bailey (2010). For other examples of amalgamated art see Kaufmann (2004), Burke (2000), and Bailey (2001).
35. See Scott (2003) for an account of gothic architecture. See Verde (2012) for a historical account of the pointed arch.
36. See pages 160–161 of Csikszentmihalyi (1996).
37. Ibid., 194–295.
38. See page 127 of Simonton (1988).
39. See Hein (1966).
40. Koestler boldly declared that “all decisive advances in the history of scientific thought can be described in terms of mental cross-fertilization between different disciplines.” See page 230 of Koestler (1964).
41. See pages 163–165 and 173 of Simonton (1994).
42. See Isaacson (2011) and Appelo (2011).
43. See page 84 of Curtin (1980).
44. See Wilson (1992).
45. See Root-Bernstein et al. (2008).
46. See Simonton (1994) and Csikszentmihalyi (1996).
47. This statement is sometimes also attributed to the novelist Grant Allen.
48. Koestler called this process bisociation. See Koestler (1964).
49. See Chapter 6, page 121 of Koestler (1964).
50. See Chapter 8 of Root-Bernstein and Root-Bernstein (1999), and Schiappa and Van Hee (2012).
51. See page 19 of Arthur (2009).
52. See page 34 of Padel (2008).
53. Several connotations of metaphors used by Aristotle would no longer be associated with metaphors today. See Levin (1982).
54. See pages 145–146 of Root-Bernstein and Root-Bernstein (1999).
55. See page 6 of Pinker (2007).
56. See Tourangeau and Rips (1991).
57. See page 35 of Padel (2008).
58. See page 93 of Csikszentmihalyi (1996).
59. See Simonton (1999).
60. See Guilford (1959), as well as Guilford (1967).
61. Such tests had been used before Guilford’s times, but not to assess creativity. See Kent and Rosanoff (1910).
62. These are not the only two dimensions on which responses to a word association test can be scored. Others include flexibility—the ability to create responses that belong in different conceptual categories, such as the use of matches to construct an object or to set fire to it. See page 85 of Simonton (1999) or Kim (2006).
63. See Mednick (1962).
64. Even though the test asks for a single solution to each triplet, it speaks to an important part of creative thinking, the ability to link remote concepts. Its usefulness has been corroborated in validation studies. See page 81 of Simonton (1999) as well as Mednick (1962).
65. See Zeng et al. (2011) or Chapter 6 of Guilford (1967).
66. See Torrance (1966) and Kim (2006). An important class of tests and studies that I do not discuss here revolves around the ability to find problems instead of merely solving problems. See, for example, Csikszentmihalyi and Getzels (1971).
67. See page 4 of Kim (2006).
68. The two limitations of some creativity tests discussed here are not the only ones. Others include that creativity cannot be reduced to a single quantity—it has multiple dimensions—even though some tests aim to compute a single, scalar score. The most general limitation, however, is that the construct creativity itself as a dispositional trait is hard to define. In the words of E. Paul Torrance, the creator of the perhaps most widely used creativity test: “Creativity defies precise definition. This conclusion does not bother me at all. In fact, I am quite happy with it.… However, if we are to study it scientifically, we must have some approximate definition.” See Torrance (1988). More generally, test theory knows two fundamental criteria by which to address the question of whether a psychological test “works.” The first is to determine reliability, a test’s ability to measure a complex quantity such as intelligence or creativity—the psychological term is a construct—with similar results across time (test-retest reliability), among different judges (inter-rater reliability), or in other varying contexts. The second criterion is a test’s validity, and specifically its construct validity, the degree to which a test measures what it aims to measure. Efforts to estimate construct validity often compare the outcome of a test with that of other, independent assessments of creativity, such as creative products. For literature on the reliability and validity of prominent creativity tests see Zeng et al. (2011), Kim (2006), Runco (1992), Torrance (1988), Upmanyu et al. (1996), Mednick (1962), and Gough (1976).
69. See Amabile (1982).
70. The observation that all creativity is ultimately assessed by people is embodied in the widely used Consensual Assessment Technique for creativity, which relies on expert ratings. See Amabile (1982).
71. See Bronson and Merryman (2010).
72. See Torrance (1988) and Plucker (1999).
73. This is why some researchers prefer the term ideation test over creativity test.
74. The practice of measuring or estimating such distances is a whole lot more sophisticated than I let on, and I do not dwell on this practice, because it is too technical and no universal estimator of distance is agreed upon. Suffice it to say that realistic distance measures themselves are more complex than the simple one I mention in the text, and semantic spaces are not low-dimensional like our three-dimensional continuous space. See Landauer and Dumais (1997). Conventional distance measures in low-dimensional spaces, when applied to concepts, often violate mathematical axioms that distance measures need to fulfill, such that the distance between two objects A and B, d(A,B), is symmetric (d(A,B)=d(B,A)), and that it fulfills the so-called triangle inequality d(A,B)≤d(A,C)+d(C,B). Also, the relevant spaces need not be continuous but may be discrete, like the spaces of genotypes we encountered earlier. For example, many workers have studied networks of word meanings as graphs—objects that consist of nodes (concepts) that can be connected by edges if their meaning is closely related. Such a graph can be traveled along paths given by its edges. For these reasons, I am tacitly using the notion of a landscape in the most general sense, namely that of a set of objects and a mathematical function from this set onto the real and positive numbers that indicates the appropriateness of an object (such as a combination of concepts) for a particular purpose. As I mentioned in Chapter 7, we still understand little about how our minds represent such collections of objects and how they explore them. See Jones et al. (2011), as well as Griffiths et al. (2007), Landauer and Dumais (1997), Bengio et al. (2003), and Gärdenfors (2000).
75. See Sobel and Rothenberg (1980), as well as Rothenberg (1986).
76. See Rothenberg (1976), Rothenberg (1980), and Rothenberg (2015).
77. See Rothenberg (1995).
78. See Norton (2012) and Chapter 10 of Rothenberg (2015).
79. See Ansburg and Hill (2003).
80. Attributed to Szent-Györgyi on page 14 of IEEE Professional Communication Society (1985). Another pertinent study involved eighty Harvard undergraduate students and showed that the minds of some students ignored previous knowledge to a greater extent. The very same students also scored higher on creativity tests. More than that, they also produced more creative products, such as award-winning works of art. See Carson et al. (2003). The study quantified differences among individuals in latent inhibition, a
term from classical conditioning that means that new associations are more difficult to learn for a familiar stimulus than for a new stimulus. Latent inhibition has been detected in many animals, including rats, dogs, and goldfish. See Lubow (1973). Some people show low latent inhibition, and their minds are less capable of ignoring information that the minds of others would filter out because it is likely to be irrelevant for a problem at hand. Latent inhibition is closely related to the phenomenon of negative priming. See Eysenck (1993).
Chapter 9: From Children to Civilizations
1. See “Test-taking” (2013), as well as Lee (2013) and Koo (2014). The exam can be retaken, but those who fail in the first round are still stigmatized because they cannot just use the second, better grade.
2. See Walworth (2015).
3. See Larmer (2014), Zhao (2014), Walworth (2015), as well as Bruni (2015).
4. See the report “PISA 2012 results in focus” available at http://www.oecd.org/pisa/keyfindings/pisa-2012-results.htm.
5. A prominent example is Michael Gove, the British Secretary of State for Education from 2010 to 2014. See Gove (2010).
6. See the IBM survey “Capitalizing on Complexity” available at https://www-01.ibm.com/common/ssi/cgi-bin/ssialias?htmlfid= GBE03297USEN. See also Pappano (2014).
7. Cited on page 305 of Runco (2014).
8. See Table 5 of Bassok and Rorem (2014).
9. See Chapter 2 of Zhao (2014).
10. See pages 40–41 of Zhao (2014).
11. See page 139 of Zhao (2014), Koo (2014), as well as Zhao and Gearin (2016).
12. See Kim (2011), as well as Bronson and Merryman (2010).
13. See Niu and Sternberg (2001, 2003).
14. See Marcon (2002).
15. See Kohn (2015), as well as Rich (2015) and Marcon (2002).
16. See Ruef (2005) and the website The Private Eye at http://www.the-private-eye.com/.
17. See Arieff (2015) and her website Project H at http://www.projecthdesign.org/.
18. See Garaigordobil (2006). The children in this study were ten and eleven years old, but the benefits of play are evident even earlier. Elementary school boys, for example, who regularly engage in rough-housing in the schoolyard are better social problem solvers, and toddlers given the opportunity to play with toy blocks (instead of watching TV) develop better language skills. See Pellegrini (1988) and Christakis et al. (2007).
19. See Scott et al. (2004), as well as Runco (2001) and Niu and Sternberg (2003).
20. See, for example, Kamenetz (2015) and Grant (2014). And just as there are alternatives to assessing students, there are alternatives to assessing teachers. See Nocera (2015).
21. See Hiss and Franks (2014), and also National Association for College Admission Counseling (2008).
22. See Gaugler et al. (1987), as well as Grant (2014).
23. I focus here on college admission testing, but just as insidious is the problem of high-stakes testing throughout the school year, for example to determine teacher effectiveness, because frequent tests continually distract from broader teaching goals.
24. It is no coincidence that the Finnish education system, one of the world’s best-ranked school systems that does not run on high-stakes standardized tests, cultivates the autonomy of teachers and schools. Other contributors to its success include a high social standing of teachers and high-quality teacher training. See Sahlberg (2015).
25. See Amabile (1985). See also Hennessey and Amabile (1998).
26. The opposite is also true: simply thinking about intrinsic reasons for pursuing an activity may be sufficient to enhance creativity in that activity. More generally, only some kinds of extrinsic motivation are detrimental for creativity, especially those that make a person feel controlled or as though they have lost autonomy in performing a task. See Collins and Amabile (1999).
27. See page 328 of Csikszentmihalyi (1996).
28. See page 335 of Csikszentmihalyi (1996).
29. Cited on page 158 of Simonton (1994).
30. See Kim (2006), Westby and Dawson (1995), Torrance (1972), and page 173 of Runco (2014).Some older studies suggest that even their own parents view creative children unfavorably. See Raina (1975).
31. See Pomerantz et al. (2014).
32. Name changed.
33. I also remember being shocked at first by the limited biological knowledge of the American biology students I helped teach, as compared to European students. This knowledge gap reflects a difference between Central European and American high schools that is often mentioned as a symptom of America’s looming decline. Remarkably, it was already noted by a 1916 French visitor to the United States, long before the United States would become a global leader in science and technology. See Rosenberg and Nelson (1994). Maximizing the amount of information crammed into a young person’s head is clearly not the most important thing an education achieves.
34. See pages 170–171 of Ramon y Cajal (1951).
35. See Wuchty et al. (2007). It is important to not equate influence with quality, because some of the best research remains obscure for a long time. The best-known example is the work of Gregor Mendel in the nineteenth century, which remained without influence for half a century, but eventually helped trigger the genetics revolution of the twentieth century. Additionally, not all citations of a scientist’s work reflect intellectual debt. Some controversial works, for example, attract negative citations that disparage the work. This is why it is not advisable to evaluate the work of individual scientists based on citations alone, even though citation patterns can be helpful to identify broad historical trends. See also Adler et al. (2009).
36. Examples of two such highly influential publications include Newman et al. (2001), as well as West et al. (1997).
37. See page 238 in Bush (1945), which is a reprint of the original document by the US government printing office.
38. See the National Science Foundation website, the “The Nobel Prizes” page, at https://www.nsf.gov/news/special_reports/nobelprizes/.
39. See Figures 1, 5, and 13 of National Institutes of Health (2012), as well as Alberts et al. (2014).
40. The funding rates of full proposals lie in the neighborhood of 20 percent, but submission of a full proposal is preceded by a mandatory pre-proposal in several programs of the NSF Directorate for Biological Sciences, with acceptance rates that are similarly low, resulting in very low overall funding rates. See National Science Foundation (2014) and Appendix 2 therein.
41. See Adler et al. (2009).
42. See Lee et al. (2013).
43. See Alberts et al. (2014).
44. These would be the select few researchers who already have made the cut imposed by academic faculty search committees, which examine, compare, and discuss the records of many applicants in detail, an arduous process for which there is no shortcut. It is also worth pointing out that young US researchers do usually receive some noncompetitive “start-up” funding from their universities, but that funding is intended to equip and start a research laboratory. Because it runs out after a few years, it is no long-term solution to avoiding hypercompetition.
45. To be sure, some US institutions like the Howard Hughes Medical Institute (HHMI) effectively use a similar strategy by funding individuals rather than projects. Consistent with the Darwinian perspective on creativity, the strategy leads to more flops, but also to bigger breakthroughs, as a comparison between HHMI- and NIH-funded investigators shows. See Azoulay et al. (2011). Being available only to a small and well-established elite in limited fields like biomedicine, such funding is a drop in the proverbial bucket.
46. For comparative research statistics, see State Secretariat for Education and Research (2011). A caveat is that impact statistics fluctuate over the years, but Swiss science remains strong even when such fluctuations are taken into account. Other reasons include good public schools and high investment in R&D. (Switzerland invests 3.4 percent of its gross domestic product into R&D, more than the United States at 2.7 percen
t, according to 2015 OECD statistics available at https://data.oecd.org/rd/gross-domestic-spending-on-r-d.htm.) Also important are low corruption and little nepotism in academic hiring, which plagues academia in some developed countries.
47. See Zappe (2013), as well as Rosenberg and Nelson (1994) and Porter (2015).
48. Additional benefits provided by universities include not just workforce training, but also a knowledge base that is necessary to take advantage of the latest science. See Pavitt (2001), as well as Callon (1994), Salter and Martin (2001), and Rosenberg and Nelson (1994).
49. For the long breath needed to commercialize fundamental discoveries, see Rosenberg and Nelson (1994), as well as Pavitt (2001) and Zappe (2013).
50. See Gertner (2012a) and Gertner (2012b).
51. Albeit very visible, they are perhaps exceptions to a rule of declining research by large corporations. See Arora et al. (2015).
52. See Amabile et al. (2002).
53. And when creative people get busy, they usually work on multiple projects, another way of linking different domains of knowledge. See Schwartz (2013) and page 113 of Sawyer (2013).
54. See Schwartz (2013).
55. See Amabile (1998).
56. See Kelley (2001) on IDEO. Psychological research also holds some surprising lessons about how diverse teams operate best. One of them is about the time-honored device of brainstorming, which is not necessarily the best way to collect diverse ideas. The simple reason is that it is very difficult to completely turn off the evaluation of ideas in a group setting, perhaps because such evaluation can take exceedingly subtle forms. It may sometimes be better if individual group members think about a problem at hand and then compare notes and discuss their candidate solutions. See pages 158–159 and 188–189 of Runco (2014).
57. See Amabile (1998).
58. See Amabile et al. (2002).
59. See Frese and Keith (2015).
60. See Slack (2002), as well as Osepchuk (1984).
61. The reasons why this class of innovators could emerge are complex, but they include a participatory form of government and strong property rights. See Acemoglu and Robinson (2012), as well as Rosen (2010).
Life Finds a Way Page 29
