How Biologists Make a Difference
1. For a general discussion of the problems of visibility and observation in science, see Hacking 1983.
2. Arguments about body structure are not new. In the nineteenth century some well-known biologists poured lead shot into empty skulls and then held forth on which group of people (males or females, blacks or whites) had larger skulls. The idea was that the larger skulls held larger brains and that the larger the brain, the smarter the person. See Gould 1981 and Russett 1989. Although the claims that there are racial differences in brain structure are made less frequently, they do occasionally appear in scientific journals. See Fausto-Sterling 1993 and Horowitz 1993. The question of the reality and meaning of brain size differences has been the subject of debate for almost two centuries. The mode of analysis I develop in this chapter is easily applicable to claims of racial and ethnic differences in brain structure.
3. The natural world, of course, does have input into the conversation. Some natural ‘‘facts’’ are more visible, more easily agreed-upon than others. There is no scientific disagreement, for example, that the brains of cats and the brains of humans look different. But there are also no commissions to promote a national dialogue about cats. On the other hand, there is disagreement—both social and scientific—about the nature of animal intelligence and how human and animal minds may or may not differ. So if scientists attempted to locate a brain center for a humanlike cognitive process in the cat, disagreement would be inevitable because there is no consensus on the nature of animal cognition itself.
4. Often, when a research system is too complex to give satisfying answers, scientists abandon it and turn to ‘‘doable’’ problems. The most famous example in my own field involves Thomas Hunt Morgan, who made fruit flies into a model organism and who developed Mendelian genetics. Morgan started life as an embryologist, but found embryos so complex that he despaired of finding answers. Initially he was skeptical of both genetics and evolution, but when, almost by accident, he began finding consistent and interpretable results that others generalized beyond the fruit fly, his research path became clear. For more on this history, see Allen 1975 and 1978 and Kohler
1994. For more on the concept of‘‘doability,’’ see Fujimura 1987 andMitman and Fausto-Sterling 1992. Several neuroscientists who read and commented on the first draft of this chapter pointed out that a goodly number of people in the field think that research on callosal size should be dropped because the CC is so intractable. But the field of neuroscience is nothing if not diverse and subdivided into different workgroups with different understandings of what constitutes ‘‘the best’’ form of research. So for others, whose work I examine here, the beat goes on. In the case of the corpus callosum, the collective failure to move on is one sure sign that a lot more is at stake than the reputations of a few neuroscientists.
5. Gelman 1992; Gorman 1992.
6. Black i99 2,p. 162.
7. Foreman 1994.
8. Wade 1944.
9. Begley 1995, pp. 51—52. Elsewhere I offer a different take on the Newsweek article: Fausto-Sterling 1997.
10. The author does present the alternate ‘‘social’’ explanation, and in that sense does not take sides in the debate. Begley writes: ‘‘Is it farfetched to wonder whether parts of girls’ brains grow or shrink, while parts of boys’ expand or shrivel, because they were told not to worry their pretty heads about math, or because they started amassing Legos from birth?’’ (Begley
1995, p. 54).
11. (Unsigned 1992). This is an idea that more than a few sexologists take quite seriously. During the winter/spring of 1998, the Listserve of professional sexologists, ‘‘Loveweb’’ (a pseudonym), had an extended and heated debate about if and why gay men gravitate to certain professions. In this debate the question of differences in spatial abilities and brain structure figured prominently.
12. Witelson 1991b; McCormick et al. 1990.
13. Schiebinger i992,p. 114.
14. Schiebinger i992.
15. Questions about the localization of function within the brain and brain asymmetry changed throughout the century. In the first half of the nineteenth century, the belief that faculties of the mind were located in particular parts of the brain met a resistance that stemmed from an association of the idea of localization with social change movements and from a struggle between theology and the emerging field of experimental biology. The localizers fell into a political camp that advocated social reforms, such as doing away with the monarchy and the death penalty and broadening the right to vote. The
antilocalizers cheered the coronation of Charles X and advocated the death penalty for blasphemers (Harrington 1987). The French neurologist and anthropologist Paul Broca finally settled the matter by correlating the loss of language ability in brain-damaged patients with a particular region (Broca’s area) of the frontal lobe of the cerebral cortex and concluding that, at least for language, the brain hemispheres were asymmetrical. Broca’s conclusions threatened ‘‘deeply rooted aesthetic and philosophical beliefs. … If it were established that the brain was functionally lopsided, this would make a mockery of the classical equation between symmetry. . . and notions of health and human physical perfection. . . .It might even undermine all recent efforts to bring logic and lawfulness to the study of the cortex, raising the spectre of retrograde movement toward the implicitly theological view of the cerebral cortex as an organ beyond all scientific classification’’ (Harrington 1987, p. 53).
Broca and other French neurologists, then, faced the specter of being dragged backward in time, away from an era of middle-class democracy and into a discourse that linked symmetry, a lack of localized brain functions, religion, and monarchy. Broca compromised by proposing that there were no innate cerebral asymmetries; instead, the brain grew unevenly during childhood. Broca’s ideas about development during childhood rested, in turn, on a set of beliefs about racial brain differences that were also thought to emerge during childhood. See Gould 1981; Harrington 1987; and Russett 1989. Thus, asymmetry not only separated humans from animals; among humans, it divided the ‘‘advanced from the primitive races’’ (Harrington 1987, p. 66). Broca effected a major change. Whereas in the first half of the nineteenth century, perfectibility had been linked to symmetry, before long the ideas of perfectibility and asymmetry became linked. Soon it grew obvious that women (dubbed Homo parietalis, in contrast to white men, who became known as Homo frontalis; Fausto-Sterling 1992), small children, and the working classes all had more symmetrical brains. By the end of the century, the list of the imperfect had grown to include madmen and criminals (who as a group tended toward higher frequencies of left-handedness and ambidexterity, both of which correlate with lessened asymmetry). Broca advanced a new scientific view by separating it from an older set of political belief systems to which it had been linked, and attaching it to a new constellation. His one area of overlap (innate symmetry but developmental asymmetry) provided continuity and acceptability; once the new scientific belief system became strong enough it flourished, generating offspring of its own.
16. Donahue suggested that the difference could account for ‘‘women’s intuition’’ (Donahue 1985).
17. De Lacoste-Utamsing and Holloway i98 2,p. 1431.
18. Efron 1990; Fausto-Sterling 1992b.
19. Stanley 1993 ,pp. 128 (emphasis in the original), 136.
20. An entire issue of the journal Brain and Cognition (26 ) is devoted to a critique of a theory by Geschwind and Behan on which Bendbow relies for her claims about innate skill differences in males and females.
21. Benbow and Lubinski 1993. The debate about a biological basis for possible differences in mathematics ability, possibly lodged in the corpus callosum, continues. For a more recent exchange on this topic, see Benbow and Lubinski 1997 versus Hyde 1997.
22. Haraway 1997, p. 129. The technoscientific objects that Haraway mentions are ‘‘fetus, chip/computer, gene, race, ecosystem, brain.’’ She doesn’t discuss the corpus callosum, but she does pay a lot of attention to the intersections between race and gender. Indeed, the paths traveled by the sticky race and the sticky gender thread cross many times, and entangle themselves more than once when they meet up in the CC.
23. Other aspects of education and child development are also grabbed up by these sticky threads. One paper, for example, claims a correlation between dyslexia and an altered corpus callosum structure (Hynd et al. 1993). This sticky node includes a host of issues in the diagnosis and treatment of learning disabilities, which reach far beyond the scope of this book.
24. One recent link involves theories of mental illness (Blakeslee 1999).
23. But see Efron 1990.
26. This claim was made by Bean (1906), who also wrote in the September 1906 issue of Century Magazine that: ‘‘The Caucasian and the negro [sic] are fundamentally opposite extremes in evolution. Having demonstrated that the negro and the Caucasian are widely different in characteristics, due to a deficiency of gray matter and connecting fibers in the negro brain. . . we are forced to conclude that is is [sic] useless to try to elevate the negro by education or otherwise.’’ Quoted in Baker 1994, p. 210.
27. Allen et al. 1991.
28. Rauch and Jinkins 1994, p. 68.
29. Latour 1988; Latour 1983.
30. Kohler 1994.
31. For additional and varied discussion of how natural objects become laboratory tools, see the several articles in Clarke and Fujimura 1992.
32. Bean 1906.
33. Which look identical to tracings made by modern scientists. See, for example, Clarke et al. 1989 and Byne et al. 1988.
34. This is remarkable in a scientific world in which few publications are referred to ten years after their initial appearance.
33. I believe the two-dimensional CC is what might, in semiotic jargon, be called a free-floating signifier.
36. Bean 1906, p. 377. If you didn’t know the context, might you not think this was a description of gender, rather than racial difference?
37. p. 386.
38. In 1999 it is the splenium, now linked to cognitive functions, that is supposed to be larger in females.
39. Mall mentored an important woman anatomist, Florence Rena Sabine (1871—1933). For a brief biography, see Ogilvie 1986.
40. Mall i909,p. 9.
41. Ibid., p. 32. Thirteen of the papers I summarize in tables 3.3 to 3.3 refer to Bean and/or Mall. Five that report sex differences and four that find no difference quote only Bean. None quote Mall alone, although his paper stood for decades as the defining work. Three groups that find their own sex difference quote both Mall and Bean, while one that reports no difference cites the earlier controversy.
42. Seenote 26 and Baker 1994.
43. For additional discussion of how maps, atlases, and other representations of the brain came to stand for the invisible brain ‘‘and all the forms of invisible work and failure hidden’’ therein (p. 224), see Star 1992.
44. Rauch and Jinkins (1994) write: ‘‘Measurements of the entire corpus callosum in three dimensions would also be a complex undertaking, since the corpus callosum is shaped much like a bird with complicated wing formation. Further these wings co-mingle with the ascending white matter tracts. . . making the lateral portion of the corpus callosum essentially impossible to define with certainty’’ (p. 68).
Even this domesticated CC presents problems, because it never separates entirely from the rest of the brain. Some of the research groups are careful to point this out: ‘‘The boundary of the CC is unequivocal dorsally but not ventrally. As in monkeys the splenium and adjacent part of the body cannot be macroscopically demarcated from the dorsal hippocampal commisure, which was therefore included to an unknown extent in our CC correction. . . the limit between the CC and the septum pellucidum was at times difficult to determine by inspection only’’ (Clarke et al. 1989, p. 217). This level of difficulty, however, experimenters feel they can live with, since the main body of the domestic CC is clear enough.
43. One scientific problem involves interpreting the huge variability found among men and among women. Elster et al. (1990) write: ‘‘As seen from our own data and that of others, callosal measurements vary nearly as much within sex as they do between sexes’’ (p. 323). See also Byne et al. 1988. A second question concerns the best method of looking at the corpus callosum. In the current dispute, investigators have used variations on two major methods. The first involves postmortem measurements on brains preserved from patients who have died from illnesses not affecting the brain. The revealed, two-dimensional surface of this CC cross-section then becomes the object of a variety of measurements. The alternate method is to use live volunteers who have agreed to have their heads examined by a magnetic resonance imager (MRI). This machine uses the body’s natural chemical activity to visualize the brain. The machine creates images on a TV screen of optical ‘‘slices’’ of the brain. Just as one might slice a loaf of bread, the machine begins at the outer surface, pictures the first thin slice, then proceeds toward the center, offering up visual slices. The visible outlines of the corpus callosum become the two-dimensional structure that the scientist then measures. The authors of a recent paper write:
studies using autopsy or cadaver material also tend to have low sample sizes. While there are advantages in using postmortem material, such as direct measurement and the ability to measure brain weight, the paucity of specimens makes for questionable statistical conclusions. Other problems associated with the use of embalmed postmortem material are the changes resulting from formalin fixation. . . . Studies using magnetic resonance images have benefited from larger sample sizes. MRI studies using a slice thickness of 7—10 mm have been criticized, as the partial volume effect may lead to inaccurate results. [Constant and Ruther 1996, p. 99]
A third technical problem concerns the concept of ‘‘allometry.’’ See, for example, Fairbairn 1997. For allometry debates applied to the problems of CC comparison, see Going and Dixson (1990), p. 166, who write:
It is well known that the brains of men are larger and heavier than those of women. This presents a difficulty for studies of sexual dimorphism, in that real differences between the brains of men and women may be obscured, or spurious differences created, by this difference in size. The question arises whether it is proper to attempt correction for brain weight. Correction reflects the theoretical model of relationships between brain weight and the quantities under consideration, and the model may not be correct. Corrected data must therefore be interpreted with caution, even scepticism.
Contrast this point of view with Holloway, who finds relative differences to be of great interest (Holloway 1998; see also Peters 1988).
46. The modern dispute about CC gender differences began with measures of corpus callosums from brains obtained at autopsy (PM) (de Lacoste – Utamsing and Holloway 1982). As subsequent reports differed both from the original and from each other, a debate about method also emerged. Postmortem studies had smaller sample sizes, for example. For fifteen studies using MRI’s the average sample size was 86.3 (range 10—122), while for fifteen studies using postmortems the sample size averaged 44.2 (range 14—70). The studies surveyed are listed in note yo.
47. Various forms of brain scans are gaining public recognition as a supposedly objective way to read the brain. Of course, MRI’s and the especially popular PET scans are constructed images. For more on brain scans, see Dumit 1997, i999aand 1999b.
48. Witelson and Goldsmith 1991; Witelson 1989.
49. Clark et al. 1989, p. 217; Byne et al. 1988. Witelson points out that ‘‘study of the concordance between direct postmortem and MR measurement of callosal size remains to be done’’ (Witelson i989,p. 821).
Using different technoscientific objects can lead to different results. I tallied up whether or not a research group had found sex or handedness differences in whole or part of the corpus callosum (either absolute or relative area differences). When MRI was the choice of method, seven research groups found a sex difference, while fourteen found no difference. In contrast, eight publications using postmortems reported sex differences, while seven did not. Is there something about using PM’s (smaller sample size, nature of the object produced?) that makes it more likely for one to find a sex difference? (I used the studies listed in the following note.)
50. The papers are: Witelson 1985, 1989, and 1991a; Witelson and Goldsmith 1991; Demeter et al. 1988; Hines etal. 1992;Cowell etal. 1993; Hol – lowayetal. 1993; de Lacoste-Utamsing and Holloway 1982; de Lacoste et al. 1986; Oppenheim et al. 1987; O’Kusky et al. 1988; Weiss etal. 1989; Habib et al. 1991; Johnson et al. 1944; Bell and Variend 1985; Holloway and de Lacoste 1986; Kertesz et al. 1987; Byne etal. 1988; Clarke et al. 1989; Allen et al. 1991; Emory et al. 1991; Aboitiz, Scheibel et al. 1992b; Clarke and Zaidel 1994; Rauch and Jinkins 1994; Going and Dixson 1990; Steinmetz et al. 1992; Reinarz et al. 1988; Denenberg et al. 1991; Prokop et al. 1990; Elsteretal. 1990; Steinmetz et al. 1995; Constant and Ruther 1996.
51. Habib et al. 1991.
52. Witelson i989.
53. Lynch ^90 p. i7i.
54. ‘‘Starting,’’ Lynch writes, ‘‘with an initially recalcitrant specimen, scientists work methodically to expose, work with, and perfect the specimen’s surface appearances to be congruent with graphic representation and mathematical analysis’’ (Lynch 1990, p. 170).
55. For a discussion of other aspects of simplification in scientific work, see Star 1983. For more on the construction of research objects within social networks, see Balmer 1996 and Miettinen 1998.
56. If CC differences appear during childhood, they may, presumably, be affected by developmental experiences. In other words, differences in adult brain anatomy, may, in fact, have been produced by social differences in the first place. See, for example, Aboitiz et al. 1996 and Ferrario et al. 1996.
57. There is an ongoing dispute about how the CC changes with age and whether male and female CC’s age differently. The principles culled from this aspect of the argument don’t differ from those developed in this chapter, so I have chosen not to plumb the depths of the aging argument. See, for example,
Salat et al. 1996. How men and women age and the problems of old age are yet other social bits picked up by the sticky CC threads.
38. Holloway et al. 1993; Holloway 1998.
39. The explanation offered for this relationship between sex and handedness is that men’s brains are more lateralized than women’s (at least for certain cognitive functions). But in general, left-handers are less lateralized than right-handers. If one assumes that a larger CC area implies less lateralization, but that women, regardless of handedness, are already less lateralized, then adding handedness into the picture won’t matter for them, but it will make a measurable difference for men.
60. Cowell etal. 1993.
61. Bishop and Wahlsten 1997. See also a detailed discussion by Byne (1993), who reaches conclusions similar to mine and Bishop and Wahlsten’s.
Meta-analysis is, itself, a controversial process. Debate continues over how to evaluate conflicting results in the scientific literature. Some find the beancounting method seen in my tables 3.3 to 3.3 most appropriate, others metaanalysis (Mann 1994). For a technical account of the effects of meta-analysis on research standards in psychology, see Schmidt 1992; for more on metaanalysis, see Hunt 1997.
62. Driesen and Raz 1993. They also concluded that left-handers have larger CC’s than right-handers.
63. Fitch and Denenberg 1998. They argue that one cannot use relative values to compare different groups unless there is a proven correlation within each group. They use IQ to illustrate their point. ‘‘On average there is no sex difference between men and women on IQ tests. However, female brains are smaller than male brains, and weigh less.’’ If one made a ratio of IQ to brain weight, women would be significantly smarter ‘‘per unit brain’’ than men. ‘‘The reason we do not use such a statistic is that research has established that there is no within-group correlation between IQ and brain size’’ (‘‘within – group’’ means comparing women with smaller brains to women with larger brains). With regard to CC, they conclude: ‘‘the procedure of dividing brain size into CC area as a ‘correction factor’ is incorrect, and, because the female brain is typically smaller, can lead to false results suggesting a larger ‘relative’ CC in females’’ (p. 326).
Aboitiz (1998) argues that correction for brain size might be appropriate if one had a better idea of how function and size correlate. Holloway (1998) takes serious exception to the case against relative measurements: ‘‘Physical anthropologists. . . routinely use ratio data. . . we do so because an extremely interesting set of facts emerges: the relative size of the brain. . . does show sexually dimorphic differences, and they vary considerably within the mammalia’’ (p. 334). Wahlsten and Bishop (1998) also argue against the wanton use of ratios, although they believe such use can be legitimate under certain circumstances, ones not met in the CC studies.
64. Halpern (1998), p. 331. This asymmetric analysis of a scientific dispute suggests that one side (the feminists, in this case) has political investments that impair their ability to impartially evaluate a literature, while the other side can clearly hear the truth that nature speaks because they have no political investment. Halpern implies that one explanation of a failure to find sex differences is sloppy work, perhaps resulting from political commitments rather than a commitment to finding truths about the natural world. This argument against feminism takes the same form as Gould’s analysis of Morton’s work on racial differences in brain size (Gould 1981). Whichever side one is on (God’s or the bad guy’s) in these disputes, such asymmetric arguments paint one into a corner (see also Halpern 1997).
63. Driesen and Raz (1993) suggest that researchers could improve the situation by improved reporting on the nature of their sample and even more measurements and different statistical tests. Bishop and Wahlsten (1997) argue that ‘‘it would be unwise to engage in further research on this topic unless a large enough sample is used in a single study’’ (p. 393). They think a minimum sample size would have to include 300 of each group—for a total of 600 brains! This sample size could accommodate the enormous variation within members of the same gender.
66. I found the concept of hypertext useful in incorporating the history of statistics into an analysis of the CC wars. Hypertext are those words or pictures that an internet surfer can signal in order to be transported to a whole new screen of information or activities. Haraway’s description of hypertext is also helpful:
In hypertext readers are led through, and can construct for themselves and interactively with others, webs of connections held together by heterogeneous sorts of glues. Pathways through the web are not predetermined but show their tendentiousness, their purposes, their strengths, and their peculiarities. Engaging in the epistemological and political game of hypertext commits its users to search for relationships in a funguslike mangrove or aspen forest where before there seemed to be neat exclusions and genetically distinct, single-trunk trees. [Haraway 1997, p.
67. For examples of the literature on the social history of statistics connections between statistics, gender, race, and the social construction of scientific knowledge, see Porter 1986, 1992, 1993, and 1997; Porter and Mikulas 1994; Porter and Hall 1993; Hacking 1982, 1990, and 1991; Wise 1993; and Poovey 1993.
As I write, the news is full of a politically charged battle over how to collect numbers for the year 2000 census. See, for example, Wright 1999.
68. The history of statistics as a technology of social management is poorly known even among scientists who use statistical procedures to ensure mathematical objectivity. For the interested reader, therefore, I’ve included several endnotes on the origins of statistics. Once again, we find that scientific arguments, this time about numbers, are also social arguments.
Head measurements are a longtime favorite. At the turn of the century, criminologists measured as many parameters of the heads of criminals as they could think of (Lombroso and Ferrero 1895). Similarly, Quetelet presented dozens of tables about criminality, and Lombroso’s little volume is packed with numbers. One table compared prostitutes, peasants, educated women, thieves, poisoners, assassins, infanticides, and normal women by measuring the following aspects of the cranium and face: anteroposterior diameter, transverse diameter, horizontal circumference, longitudinal curve, transverse curve, index of cephalon, anterior semicircumference, minimal frontal diamter, diameter of cheekbones, diameter of jaws, and height of forehead (Lombroso and Ferrero 18959 pp. 60—61).
69. Between 1820 and 1850, Europe experienced a great numerical explosion. From 1820 to 1840, ‘‘the rate of increase in the printing of numbers appears to be exponential whereas the rate of increase in the printing of words was merely linear’’ (Hacking 1982, p. 282). The increasing number of published statistical reports covered a growing diversity of measured things. Consider, for example, A Treatise on Man and the Development of His Faculties, by the Belgian astronomer-turned-statistician M. A. Quetelet. Originally published in Paris in 1835, the Treatise contains hundreds of numerical tables. Quetelet enumerated—that is he counted and categorized—‘‘the development of the physical properties of man. . . development of stature weight, strength, &c.,
. . . development of the moral and intellectual qualities of man. . . [and] of the properties of average man, of the social system. . . and of the ultimate progress of our knowledge of the law of human development’’ (Quetelet 1842, table of contents). In the fourteen-page section on ‘‘The Development of the Propensity to Crime’’ alone, Quetelet included twenty-five statistical tables listing the numbers of people committing crimes in a particular year, their educational level compared to whether the crime was against property or people, the influence of climate and season on crime, the disposition of legal cases by city and town, crimes in different countries, sex differences in the types of crime, age of the criminal, motive for the crime, and much, much more. England, France, and Belgium all experienced a grand period of statistical gathering. Governments needed information about a changing populace. Was the birthrate high enough? What was the state of the working classes (and how likely were they to revolt)? How healthy were army recruits? The social and political questions of the time dictated the types of information sought and their tabular presentation. By the time of the French Revolution, statistics was not regarded as an arm of pure and applied mathematics, free from social import and content, but rather had come ‘‘to be conceived in
France and England as the empirical arm of political economy’’ (Porter 1986, p. 27).
70. Statistical tabulations required the creation of categories, a process the philosopher Ian Hacking calls subversive: ‘‘Enumeration demands kinds of things or people to count. Counting is hungry for categories. Many of the categories we now use to describe people are byproducts of the needs of enumeration’’ (Hacking 1982 ,p. 280; emphasis in original)—just as the application of measurement to the human body (morphometry) requires the creation of subdivisions such as the 2 – D CC, the splenium, the genu, or the isthmus. As the historian Joan Scott writes: ‘‘Statistical reports are neither totally neutral collections of fact nor simply ideological impositions. Rather they are ways of establishing the authority of certain visions of social order, of organizing perceptions of‘experience’’’ (Scott i988,p. 115). See also Poovey 1993.
In the first half of the nineteenth century, Quetelet formulated a way to characterize populations. For Quetelet, a group of individuals seemed chaotic, but as a population, they behaved according to measurable social laws. He believed so strongly in statistical laws that he devoted himself to creating a composite human: the average man whom he viewed as a moral ideal. He examined many facets of the average man: How had he been described by the literary world and in the fine arts? What physical and anatomical measures did anatomy and medicine offer? (Stigler 1986). Moreover, Quetelet standardized racial, sexual, andnational types, which he believed enabled scientists to compare intelligence across the races. Caucasians, he felt, came out ahead. See Quetelet 1842,p. 98.
Quetelet equated deviation from a statistical norm with abnormality in the social, medical, or moral sense. Crime and social chaos resulted from the great disparity between the very wealthy and the very poor, while middle – income people who lived moderate lives were bound to live longer than those on the extreme. ‘‘The progress of civilization, the gradual triumph of mind, was equivalent to a narrowing of the limits within which the ‘social body’ oscillated’’ (Porter 1986, p. 103). Deviation from the mean represented a mistake or error.
71. The sociologist Bruno Latour uses metaphor to transform the drablooking scientific text—filled with graphs, tables, and statistical testing— into a thrilling epic. Note that the hero here is the result—in this case—a finding of sex differences:
What is going to happen to the hero? Is it going to resist this new ordeal?
… Is the reader convinced? Not yet. Ah ha, here is a new test. . . Imagine the cheering crows and the boos. . . . The more we get into the niceties of the scientific literature, the more extraordinary it becomes. It is now real opera. Crowds of people are mobilized by the references; from offstage hundreds of accessories are brought in [e. g.. statistical tests and analyses]. Imaginary readers. . . are not asked only to believe the author but to spell out what sort of tortures, ordeals and trials the heroes should undergo before being recognized as such. The text unfolds the dramatic story of these trials. . . . At the end, the readers, ashamed of their former doubts, have to accept the author’s claim. These operas unfold thousands of times on the pages of Nature. [Latour 1987^. 53]
72. Statistics can be seen as a specialized technology of difference. Statistical analyses and the establishment of population means (which often became norms) became an essential part of the field of psychology in the twentieth century. Only then was a ‘‘normal’’ psychological subject established—built by heavy reliance on population aggregates. For a full treatment of the role of statistics in the narrowing of ‘‘epistemic access to the variety of psychological realities,’’ see Danziger 1990, p. 197. Danziger’s history is especially important in analyzing lateralization studies, which are often used to demonstrate the psychological relevance of CC studies.
73. During the second half of the nineteenth century, statisticians reinterpreted the bell curve as representing mere variability rather than a distribution of error around an average, ideal type, as Quetelet thought. Eventually, scientists renamed standard error, calling it standard deviation instead. Charles Darwin’s first cousin, Sir Francis Galton, did not extol the virtues of the median (see Porter 1986, p. 129). In contrast to earlier scientists, who focused on improving humankind through improving environmental conditions, Gal – ton wanted to use knowledge about the exceptional variant in order to use evolution (selective breeding) to improve upon the bodies making up a population. To this end, he invented a new field of study and a social movement: eugenics. In his book Hereditary Genius: An Inquiry into Its Laws and Consequences, he wrote a prescription for improving the health of English society: ‘‘I propose. . . that a man’s natural abilities are derived by inheritance. . . . Consequently, as it is easy… to obtain by careful selection a permanent breed of dog. . . gifted with peculiar powers. . . , so it would be quite practicable to produce a highly-gifted race of men by judicious marriages during several consecutive generations‘‘(Galton 1892, p. 1). Dismissing the possibility that variations in human ability resulted primarily from differences in training and opportunity, he wrote: ‘‘I have no patience with the hypothesis that babies are born pretty much alike, and that the sole agencies in creating differences between boy and boy, and man and man, are steady application and moral effort’’ (Galton 1892, p. 12). As evidence, he noted that despite the wider educational opportunities available in America (compared with the more rigid class system of Great Britain), England still produced more great writers, artists, and philosophers: ‘‘The higher kind of books. . . read in America are principally the work of Englishmen. … If the hindrances to the rise of genius were removed from English society as completely as they have been removed from that of America, we should not become materially richer in highly eminent men‘‘(Galton 1892. p. 36). Galton feared for the future of English civilization, but hoped that if he could figure out how to predict the inheritance of mental characteristics and devise a breeding program, higher civilizations could be saved. Galton and his students oversaw a gradual transition from Quetelet’s concept of probable error to that of a standard deviation—free from any implication of natural error and providing the raw material with which eugenic programs could work. Similarly, Quetelet’s law of error became a normal distribution. The same old bell curve, once seen to conceptualize nature’s difficulties in making perfect copies of its essential template, became in Galton’s hands a representation of nature’s virtue in producing a wide and varying range of individuals.
Galton chose statistics as the best method for predicting the relationship between a parental trait—say, height or intelligence—and the same trait in offspring. He devised the concept of a correlation coefficient—a number that would express the relationship between two variables. His concept of correlation developed because his eugenic concerns ‘‘made possible a more general treatment of numerical variability’’ (Mackenzie 1981; Porter 1986). Subsequent developers of statistics, especially Karl Pearson (who invented the chi square and contingency tests) and R. A. Fisher (who invented the analysis of variance tests often used today), were also devotees of eugenics and, as with Galton, their concerns about human heredity drove their statistical discoveries. See Mackenzie 1981 for a fascinating discussion of the political implications of the chi-squared test and the way Fisher’s concerns with eugenics led him to significantly narrow the scope ofevolutionary theory. The field ofmod – ern biology has been importantly shaped by the eugenic commitments of a large number of biologists working in the first third of the twentieth century.
74. The process does not involve drawing such a curve; the information can be dealt with entirely through numbers. I invoke the curve here to help the reader visualize what is being done.
73. For a discussion of the limitations of the uses of ANOVA, see Lewon – tin 1974 and Wahlsten 1990. Lewontin writes: ‘‘What has happened in attempting to solve the problem of the analysis of causes by using the analysis of variation is that a totally different object has been substituted. . . . The new object of study, the deviation of phenotypic value from the mean, is not the same as the phenotypic value itself’’ (p. 403).
76. This test takes into account sample size, the degree of variation around the male mean, and the degree of variation around the female mean. Many of the workers in this dispute acknowledge the wide variability for both sexes in CC shape.
77. Both means testing and ANOVAs were used by various groups.
78. Allen etal. 1991.
79. Latour (1990) calls these graphs, tables, and drawings ‘‘inscriptions,’’ and comments on their place in the scientific paper: because ‘‘the dissenter [in this case that would be me—the highly skeptical reader] can always escape and try out another interpretation. . . much energy and time is devoted by scientists to corner him and surround him with ever more dramatic visual effects. Although in principle any interpretation can be opposed to any text and image, in practice this is far from being the case; the cost of dissenting increases with each new collection, each new labeling, each new redrawing’’ (p. 42; emphasis in the original).
80. Allen et al. 1991 ,p. 933; emphasis in the original.
81. 1Ы^ p. 937.
82. In the first quarter of this century, Pearson developed the X2 method to establish the validity of a correlation between two or more qualitative variables. But other methods also contested for this privilege. See Mackenzie 1981, pp. 133—183 for an analysis of a dispute between Pearson and his student G. Udny Rule over the best way to analyze such data. Rule studied social policy requiring a yes or no answer. Did, for example, a vaccine against a particular disease save lives during an epidemic? Rule invented a statistic— which he called Q—which could tell him whether there was a relationship between treatment and survival. Pearson not only wanted a yes or no answer, he wanted to study the strength or degree of any association. The motivation for this ‘‘strength of correlation’’ approach came directly from his wish to develop a practical program of eugenics—’’to alter the relative fertility of the good and the bad stocks in the community’’ (Mackenzie 1981 ,p. 173). Pearson needed a mathematical theory in which knowledge of a person’s ancestry could enable him to predict an individual’s abilities, personality, and social propensities. In the 1890s, when Pearson first began working on problems of descent, there was no accepted way to study the heredity of unmeasurable characteristics such as color or mental ability. Pearson needed to extend the theory of correlation to measure the strength of inheritance of traits that had no units of measurement. Pearson solved his problem by collecting data on intelligence—based on teachers’ estimates of a child’s abilities—from over 4,000 pairs of siblings in the schools. He then asked: If one brother was rated highly intelligent, what was the likelihood that the other one would as well? His method of calculating correlation for these conditions convinced him that human character traits were strongly inherited. ‘‘We inherit,’’ he wrote ‘‘our parents’ tempers, our parents’ conscientiousness, shyness and ability, even as we inherit their stature, forearm and span’’ (quoted in Mackenzie, p. 172). Rule criticized Pearson for making an unverifiable assumption—that the numbers used to calculate the X2 were distributed in a bell-shaped curve. Pearson attacked Rule’s Q because it could not measure the strength of correlation. Their positions were unreconcilable because they had designed their tests to accomplish different goals. The controversy between Rule and Pearson never really ended. Today both methods are used. According to Mackenzie, Rule’s Q is most popular among sociologists, while Pearson’s correlation coefficient is more in vogue among psychometricians. For additional analysis of the issues raised by this dispute, see Gigerenzer et al. 1989.
83. This is not an attack on Allen et al. Indeed, this is one of the strongest papers in the CC collection. Rather, I use them to illustrate the tactics scientists use to stabilize and draw meaning from the CC.
84. That is, the type of story I explicated when discussing nineteenth- century disputes on brain laterality (see notes 68—73 and 82 on the social history of statistics).
A related and helpful theoretical approach would be to think of the CC as a boundary object, in this case a standardized form that ‘‘inhabits several intersecting social worlds and satisfies] the informational requirement of each’’ (Star and Griesemer 1989, p. 393). Boundary objects can take on different meanings in each social world, but they must be easily recognizable and thus provide a way to translate among different groups. The social worlds in this case can be read from figure 3.6. They include research areas with overlapping but differing foci, as well as social and political groupings—educational reformers, feminists, gay rights activists, and the like.
83. For some current theories of CC function, see Hellige et al. (1998), who suggest that larger CC size may reflect a greater functional isolation of the two hemispheres. Moffat et al. (1998) suggest that males (there were no females in this study) whose speech and handedness functions are located in different brain hemispheres may require increased interhemispheric communication and thus a larger CC. (Note the difference with the previous citation.) Nikolaenko and Egorov (1998) note that there is no commonly accepted model of brain asymmetry. They present a thesis in which the CC is the key to integrating dynamically interacting brain hemispheres. The nerve fibers that course through the CC certainly have different functions, some excitatory and others inhibitory. Some types of CC activity will surely inhibit information flow, and other types will enhance it. The level of subtlety needed to understand the mechanisms involved in brain cognition and their relationship to CC function are not currently available. See, for example, Yazgan et al. (1993), who write: ‘‘The corpus callosum is composed of fibres with excitatory and inhibitory functional effects, the proportions and distributions of which are unknown in the CC’s of these particular subjects’’ (p. 776). The same may be said of all the subjects in all the human CC studies. For an extended treatment of hemispheric asymmetry, see Hellige 1993.
86. Allen etal. 1994. O’Rand (1989) applies the idea of a thought collective to beliefs about brain morphology and cognitive abilities. Star (1992) writes that a conclusion about the function of a particular region of the brain ‘‘is really a report about the collective work of a community of scientists, patients, journal publishers, monkeys, electrode manufacturers, and so on, over a period of some 100 years’’ (pp. 207—208).
87. Cohn (1987) discusses how entering into a linguistically defined community—in her case, defense intellectuals—imposes a particular mode of thought. To communicate within the community, one must use their language. But in choosing their language, one gives up other ways of seeing the world. See also Hornstein 1988.
88. See, for example, Aboitiz et al. 1992. Nobody knows whether size differences in CC subdivisions result from denser packing of neurons, a change in the relative proportions of differently sized neurons, or a reduction in the number of many different kinds of neurons. For attempts to answer some of these questions, see Aboitiz et al. 1992 and i998a, b. In animals, researchers identify and trace individual nerve fibers from their origins in the cerebral cortex to their passage through the CC after injecting a dye in the cortex. Individual nerve fibers absorb the dye and conduct it along their axons. (An axon is the long end of a nerve fiber that conducts electrical impulses from the cell’s point of origin to its connection to another nerve cell or a muscle cell.) When researchers later isolate the CC, they can find the dye and see which part of the CC contains axons originating from the region of the cerebrum they injected. In one study of this sort on rats, researchers confirmed that the splenium was comprised in part of axons originating in the visual cortex (the region of the brain involved with enabling vision). Some of the axons running through the CC were coated with an insulating substance called myelin, while others were bare nerve fibers. There were no sex differences in overall area of the CC or of the splenium. The total density of unmyelinated axons (number of fibers per mm2 in certain subdivisions of the splenium) differed in male and female rats (female>male). Male rats, however, had the advantage in myelinated axons. Simply counting axons of all types buried the more subtle differences. The size of both fiber types was the same in males and females (Kim et al. 1996). This level of detail—currently unattainable in humans—is what is minimally necessary to relate functional consequences to structural differences. In humans, very careful dissection has revealed some of the general topographical features of connections between particular regions of the human cerebral cortex and particular regions of the corpus callosum (de Lacoste et al. 1985; Velut et al. 1998).
89. For a single volume that shows the density and diversity of this node, see Davidson and Hugdahl 1995. There are literally thousands of research articles on handedness, brain asymmetry, and cognitive function. This is one definition of node density. By diverse, I mean the range of questions (or number of subnodes) subsumed within this knot. Articles in the Davidson volume cover the following topics: hormonal influences on brain structure and function, brain anatomy, theories of visual processing, theories of aural processing, discussions of handedness, theories of learning, links to other medical questions such as sudden cardiac death, links to emotional aspects of behavior, evolution of brain asymmetry, development of brain asymmetry, learning disabilities, and psychopathology.
90. See, for example, Bryden and Bulman-Fleming 1994 and Hellige et al. 1998.
91. Note the paper title of Goldberg et al. 1994. For an evaluation of methods used in laterality studies, see Voyer 1998.
92. See, for example, Bisiacchi et al. 1994; Corballis 1994; and Johnson et al. 1996.
93. For an up-to-date view of the debate about handedness, laterality, cognition, lateralization, sex differences, and much more, see Bryden et al. 1994 and the papers that reply to all, found in vol. 26 (1994) of Brain and Cognition. See also Hall and Kimura 1993.
94. For a recent study, see Davatzikos and Resnick 1998.
93. Whether one finds differences in performance on specialized tests of particular cognitive tasks may well depend on what sample one uses (e. g., a large general sample versus a sample of gifted children) and when and how one does the test. Although many previously reported differences have begun to diminish or even disappear, a few are stable with time. This does not, of course, mean that they are biological in origin, only that if they are social, they have not been modified by social change in the past twenty to thirty years. The number of different types of tests on which sex differences continue to appear and are of the same magnitude as they were twenty-five years ago is now small. The social import of any such differences, of course, remains in hot dispute. For discussions of meta-analyses of studies of gender differences in cognition see Voyer et al. 1993; Halpern 1997; Richardson 1997; and Hyde and McKinley 1997. For a discussion of the meaning and interpretation of differences on cognitive tasks, see Crawford and Chaffin 1997 and Caplan and Caplan 1997.
96. Fausto-Sterling 1992; Uecker and Obrzut 1994; Voyer et al. 1993; Hyde and McKinley 1997.
97. Gowan 1983.
98. Some of these conflicting theories are discussed in Clarke and Zaidel 1994.
To get a taste of the varying viewpoints and research on gifted children and the incorporation of findings on the corpus callosum, see Bock and Ack – rifl 1993.
99. Evidence that the human CC continues to develop into at least the third decade of life is reviewed by Schlaug et al. 1993. The implication of postnatal development is that environment (in this case, musical training) can influence brain anatomy. These researchers report that musicians who began their musical training before the age of seven had larger anterior CC size than controls. They find their results to be ‘‘compatible with plastic changes of components of the CC during a maturation period within the first decade of life, similar to those observed in animal studies’’ (p. 1047). Note the invocation of animal studies.
100. Allen et al. i99i, p. 940.
101. Some scientific papers, however, explicitly raise the possibility. Cowell et al. (1993) link laterality, hormones, and sex differences in the frontal lobe while Hines (1990) floats the idea of hormonal effects on the human corpus callosum.
102. Halpern (1998) writes: ‘‘For obvious ethical reasons, experimental manipulations ofhormones that are expected to alter the brain are conducted with nonhuman mammals. . . . Researchers assume that the effects in humans will be similar. . . but not identical. . . .Conclusions. . . are corroborated with data from. . . naturally occurring abnormalities. . . such as girls with congenital adrenal hyperplasia’’ (p. 330). Note how the hormone nodule always links back at some point to intersexuality. A similar approach to drawing strength from association with other arenas may be found in Wisniewski (i998).
103. Sociologist Susan Leigh Star and the psychologist Gail Hornstein describe this as a shell game that has played itself out in earlier disputes about the brain when ‘‘uncertainties from one line of work were ‘answered’ in the public construction of the theory by drawing on results from another domain. In triangulating results across domains, accountability to the anomalies in any single domain was never required’’ (Hornstein and Star 1994, p. 430).
104. Efron (1990) has written an extensive critique of the concept of hemispheric lateralization and of the experimental methods, such as the use of tachistoscopes and dichotic listening devices, which support claims of lateralization. Uecker and Obrzut (1994) question the interpretation of right – hemisphere male superiority for spatial tasks. Chiarello (1980) suggests there is no conclusive evidence that the CC is needed for lateralization of certain functions. Clarke and Lufkin (1993) find that variations in callosal size do not contribute to individual differences in hemispheric specialization. Jancke et al. (1992) critique interpretations of dichotic listening tests for cerebral lateralization. Gitterman and Sies (1992) discuss nonbiological determinants of language organization in the brain, while Trope et al. (1992) question the generalizability of the analytic/holistic distinction between left and right brain hemispheres.
103. Writing about the skeleton debate, the historian Londa Schiebinger notes: ‘‘Since the Enlightenment, science has stirred hearts and minds with its promise of a ‘neutral’ and privileged viewpoint, above and beyond the rough and tumble of political life’’ (Schiebinger 1992, p. 114).
106. Latour considers objects of knowledge to be hybrids. Reading his account of the history of natural and political science as efforts to stabilize the nature/nurture dichotomy by denying the hybrid nature of scientific facts was an illuminating experience for me (Latour 1993).
107. I have not exhausted the analysis. I don’t consider, for example, the institutional resources available to different research groups. Allen et al., for example, work at UCLA and have access to a large collection of MRI’s taken for other medical purposes. The researchers skeptical of sex differences, such as Byne et al. (1988), did not have institutional access to such a large database. Allen et al. can swamp out Byne and colleagues’ finding of no difference by the sheer size of their database. Ruth Bleier’s (she is the leader of the research group of Byne et al.) personal history as a political radical and feminist leads her to be more marginal in terms of her access to databanks. It is likely that, politically or otherwise, marginal people always have a harder time mobilizing counter data and getting their mobilized data heard.
Nor have I produced a detailed analysis of conventional rhetoric. For example, Allen et al. use the word dramatic to describe the sex difference in splenial shape, when in fact they had to use a rather tortured process to render the difference visible. The use of emphatic words is, of course, part of the rhetoric of calling attention to a particular finding.
108. This point really becomes clear when we think about homosexuality. In the early part of the century and currently, many liberal thinkers were/are genetic determinists. They believe(d) that homosexuality is ‘‘genetic,’’ and that one social implication is that gay people should have equal civil rights. Religious conservatives, on the other hand, argue that homosexuality is a ‘‘choice’’ and that, since it is also a sin, homosexuals should choose to become straight. They use the ability to choose to argue against equal civil rights. Sandwiched in between, in the middle of the century, are the practices of Nazi Germany. Nazis believed that homosexuality is ‘‘genetic,’’ but saw that as an argument for extermination.
109. Halpern 1997, p. і,098.
110. Hyde and McKinley 1997 ,p. 49. What is meant by this goal is often unclear. Many conceptualize equal opportunity to mean no more than the absence of overt discrimination. Hyde’s view is that it should involve active efforts to level the cognitive playing field. Furthermore, my argument assumes that when they appear, group differences in cognition are small enough that the right combination of skills training and encouragement could eliminate them. I am aware of the counterargument—that it would take extreme measures (cost too much, push girls too hard against their ‘‘natural’’ inclinations, etc.) to equalize group differences, or that perhaps equalizing group differences in cognition by training and remediation is simply not possible. (Currently, we offer remedial reading and verbal training. These are areas where group differences favoring girls often appear.) A further assumption underlying this argument is that known group differences in cognition actually account for subsequent professional achievement. My own view is that this is probably not a good assumption. I suspect that unacknowledged gender schema do abetter job of explaining such difference. (See Valian i998a, b for a full statement of this argument.)
iii. I know from experience that some will read my position as antimaterialist regardless of my protestations, but reaffirming my materialist belief system is worth a shot.