Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 Sex Affects Behavior and Perception ABSTRACT Basic genetic and physiological differences, in combination with environ- mentaZ factors, result in behavioral and cognitive differences between males and females. Sex differences in the brain, sex-typed behavior and gender identity, and sex differences in cognitive ability should be studied at aZZ points in the life span. Hormones play a role in behavioral and cognitive sex differences but are not solely responsible for those differences. In addition, sex differences in perception of pain have important clinical implications. Research is needed on the natural variations between and within the sexes in behavior, cognition, and perception, with expanded investigation of sex differences in brain structure and function. The purpose of this chapter is not to review all the evidence about the nature and determinants of sex differences in behavior or any other char- acteristic but to describe how basic genetic and physiological differences between males and females might produce phenotypic differences throughout the life span. SEX DIFFERENCES IN BEHAVIOR AND COGNITIVE ABILITIES Behavioral sex differences may originate in events that begin in the womb. The fetal environment, particularly hormones present during de- velopment, affects aspects of later behavioral and cognitive sex differ- ences. Sex differences in behavior are important in their own right, but 79
80 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH also suggest ways in which prenatal influences can contribute to sex dif- ferences in nonbehavioral traits, including those associated with health and illness. The information presented in this section should not be inter- preted to mean that all behavioral sex differences are caused by hormones during prenatal development but, rather, should serve as an illustration of the potential role of prenatal hormones in producing phenotypic sex differences. No single factor produces sex differences in any one behavioral or cognitive trait, let alone in all of them. Until recently, it has been popular to focus on cultural or experiential causes of these differences. Thus, for example, sex differences in the occurrence of depression have been con- sidered to reflect women's greater social orientation (which is itself as- sumed to be cultural) or stresses associated with women's multiple social roles (as also mentioned in Chapter 3~. In the past 10 years, however, there has been increasing appreciation of the fact that genetic and physi- ological differences between males and females might also influence be- havioral sex differences. Although some might argue that the pendulum has swung too much in favor of genes and physiology (Fausto-Sterling, 2000), there is considerable interest in examining the joint effects of genes, physiology, and experiences. In particular, there is recognition that the environment is not independent of the individual (Scarr and McCartney, 1983~. Individuals actively construct their environments and are re- sponded to by others in their environments. The effects of imposed envi- ronments are not the same for everyone. When one considers sex differ- ences, one must also remember that females and males "inhabit" different cultures and that some behavioral sex differences are more marked when people are in social groups than when they are alone. Thus, questions about sex differences concern not just differences between individual males and females but also differences between male and female cultures (Maccoby, 1998~. Psychosexual Differentiation Studies with nonhuman vertebrate species suggest that the sexual role adopted at maturity is determined by the hormonal environment in early life. As for other aspects of sex differentiation, there appears to be a predisposition for individuals to develop female sexual postures. The de- velopment of male patterns of sexual behavior in nonhuman species is influenced to a large extent by exposure to androgens in particular, tes- tosterone during the prenatal and perinatal periods. This organizing ca- pacity of testosterone administered at a critical stage of development has been localized to specific areas of the brain. Sexually dimorphic organiza- tions of target cell nuclei detected during behavior-related events in other species are the result of local aromatization (conversion) of testosterone to
SEX AFFECTS BEHAVIOR AND PERCEPTION 81 estradiol in the central nervous systems of these species. In humans, mas- culinization of the central nervous system does not appear to result from aromatized estradiol but appears to result from forms of testosterone (Grumbach and Auchus, 1999~. Sex Differences in the Central Nervous System and Brain Sex differences in the central nervous system extend beyond func- tions and structures traditionally associated with reproduction. These dif- ferences might be better understood if they were studied in the context of new and exciting conceptualizations of how the brain works, which en- compass notions of lifelong plasticity, ensemble processing and distrib- uted networks, and the brain's role as an endocrine organ. The classic examples of sex differences in the brain involve neuroana- tomical differences that are developmentally programmed. In several spe- cies, sex differences in the patterns of synaptic innervation are observed in the preoptic area and are influenced by the perinatal hormone environ- ment but not by hormonal conditions in the adult animal (Gorski et al., 1978; Nottebohm and Arnold, 1976; Raisman and Field, 1971~. These early studies reveal the effects of castration of males and the administration of testosterone to females early in development and established the idea that differences in the wiring of the brain are programmed at birth. There are now many documented sex differences in a wide range of species, includ- ing primates (Forger, 1998~. In canaries and zebra finches, for example, differences in singing behavior between males and females have been correlated with differences in the sizes of three vocal control areas in the brain (Nottebohm and Arnold, 1976), but, importantly, the young male bird must hear the adult male song to initiate its own repertoire. There are also sex differences in the human brain, including the higher cognitive centers. These differences have been observed in adults, and the nature and origins of these differences are subjects of active investigation. Recent studies suggest sex differences in brain structure size as the brain develops in children (Giedd et al., 1987; Lange et al., 1997~. It is important to remember that these differences are not absolute and that it is currently not possible, nor may it ever be, to look at a brain or a brain image and know the sex of its owner. The principles that have emerged from studies with nonhuman spe- cies have generally been confirmed in humans, although differences in details exist. For example, androgens act as masculinizing agents in all species, but they appear to do so through different metabolites. Another important principle that has emerged from studies with animals and that has been confirmed in humans is that the central nervous system remains plastic throughout the life span. Finally, former notions that discrete brain regions have specific and static functions have been modified by work on
82 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH ensemble neuronal activity (Laubach et al., 2000) and distributed net- works (Sanes and Donoghue, 2000~. Areas that have not been traditionally thought to be sexually dimor- phic may be involved in sexually dimorphic behavior. Some examples are (1) dopamine functions within the striatum and nucleus accumbens (Becker, 1999~; (2) the responsiveness of neurons in the gracile nucleus to stimulation of skin and pelvic organs (Bradshaw and Berkley, 2000) (neu- ronal responsiveness and activity in the two regions vary with the estrous cycle and hormonal manipulation in a manner that correlates with lordo- sis and other reproductive behaviors; and (3) modulation of functions in the hippocampus, inferior olive, and cerebellum (Smith et al., 2000~. The Brain as an Endocrine Organ A great deal of evidence indicates that the brain functions as an endo- crine (hormone-secreting) organ. Throughout life, there are profound sex differences in the brain's responsiveness to sex hormones, some of which are established early in development and which have implications for later behavior, including cognitive function. The brain is also involved in the regulation of other hormones that show sex differences and that are involved in both reproductive and non- reproductive behaviors. For example, aggression in male mice is consid- erably more intense than that in female mice, and this difference is known to be influenced by testosterone. Recent studies suggest that the story may be more complex. Nitric oxide, a compound that participates in cell- to-cell signaling, may be involved. The neural form of nitric oxide is mea- sured by changes in nitric oxide synthase (nNOS) and plays an important role in the expression of aggressive behavior in males (Nelson, 1997~. This was discovered when nNOS knockout mice were created, and informal observations indicated that nNOS -/- male mice (where -/- indicates the absence of the gene on both chromosomes) were hyperaggressive but that female nNOS knockout mice were not (Nelson et al., 1995~. Inappropriate aggressiveness was never observed among the nNOS -/- female mice. When given an opportunity to defend their pups, nNOS -/- mice were very docile, unlike their wild-type sisters. These studies suggest that ni- tric oxide from neurons has important but opposite effects in the media- tion of aggression in male and female mice (Nelson and Chiavegatto, 2000~. In the rat brain, the ventromedial hypothalamus is important in the regulation of reproductive behavior such as lordosis. The estrogen-induc- ible progesterone receptors in the ventromedial nucleus appear to play a role (Parsons et al., 1984; Schumacher et al., 1992~. Estrogens have also been shown to induce receptors for oxytocin in the hypothalamus, and blockage of oxytocin receptors interferes with the expression of lordosis
SEX AFFECTS BEHAVIOR AND PERCEPTION 83 behavior. Estrogens also cause the formation of new synaptic connections between ventromedial hypothalamic neurons in the hypothalamus. Rats display a characteristic set of motor behaviors following activa- tion of serotonin receptors or elevation of synaptic serotonin levels after treatment with L-tryptophan. Both males and females exhibit this "sero- tonin behavioral syndrome," but females display signs of the syndrome at much lower doses than males. Fischette and colleagues (1984) have shown that androgens, via androgen receptors, modulate the reduced sensitivity of male rats to the tryptophan drug challenge. Sex-Typed Behavior and Gender Identity Discussions about the determinants of human sex-typed behavior, especially gender identity, have recently become highly visible because of scientific and popular accounts of a prominent case (Colapinto, 2000; Dia- mond and Sigmundson, 1997~. The case challenged the established belief that individuals are born with the potential to develop male or female gender identity and that the specific gender identity can be determined exclusively by sex of rearing (Hampson and Hampson 1961; Money and Ehrhardt, 1996; Money et al., 1955; reviewed in Grumbach and Conte, 1998~. For detailed reviews and discussions, see Bradley et al. (1998), Colapinto (2000), Diamond and Sigmundson (1997), Fausto-Sterling (2000), Kessler (1998), Wilson (1999), and Zucker (1999~. The case involved a boy (46,XY karyotype) with male-typical devel- opment whose penis was ablated after a mishandled circumcision and whose gender was subsequently reassigned and reared as a female. Con- trary to early reports, the child never adjusted to the female assignment, despite having no knowledge of his early history. Sex reassignment was ~ 1 1 e1 · 1 · · 1 1 · ~ 1 ~ 1 · ~ 11 1 requested, and the Individual Is now reported to live successfully and happily as a man. Because this individual is a normal genetic male who was exposed to male-typical hormones in prenatal and early neonatal life, this case lends credence to the view that gender identity is determined by early hormones that act on the developing brain and argues against the view that rearing sex is the main determinant of gender identity (Dia- mond and Sigmundson, 1997; Grumbach and Conte, 1998~. The conclusion, however, must be considered in light of other details of this case and other cases. The individual described above (Diamond and Sigmundson, 1997) was reared unequivocally as a boy at least until age 7 months, when the accident occurred, and perhaps longer, because the final decision about female reassignment was not made until his sec- ond year and surgery was not completed until age 21 months. Further- more, the outcome for another individual with an ablated penis was very different: after an accident at age 2 months, another child was reassigned as a female at age 7 months and has reportedly adapted well to this
84 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH identity. As an adult, she shows no evidence of gender dysphoria, although she has a male-typical occupation and a bisexual orientation (Bradley et al., 1998~. Ongoing studies with boys with cloacal exstrophy (malformed or ab- sent penis with normal testes) who are reared as girls should help to provide systematic evidence about the determinants and malleability of gender identity. These boys are usually reassigned as girls because of concerns about adjustment problems associated with inadequate male genitalia. Preliminary reports from an ongoing systematic study (Reiner, 2000) indicate that more than half of these female sex-assigned XY chil- dren identify as boys, consistent with their male-typical prenatal andro- gen exposure, and not with their female-typical rearing. Interestingly, however, some of these children continued to accept their female assigned sex, so it will be important to determine what differentiates children with male identity from those with female identity, despite their common 46,XY chromosome constitutions. This is clearly an area deserving of further investigation. Other Sex Differences in Human Behavior Although identification as male or female is the most obvious psycho- logical sex difference, it is far from the only one. A variety of important human behaviors covering a range of domains are more common or occur at higher levels in one sex than in the other. The behaviors that have received the most attention include aspects of normal social behavior and cognition, such as childhood play behavior and related activities and in- terests, personality (such as aggression and interest in babies), nonverbal communication, sexuality, and cognitive abilities (Hall and Carter, 1999; Halpern, 2000; Maccoby, 1998; Ruble and Martin, 1998~. Activities related to these behaviors are performed at different frequencies by males and females in most cultures studied (Daly and Wilson, 1990~. Again, the goal of this chapter is not to provide an exhaustive review of behavioral sex differences but to illustrate some of the differences and to indicate how they might be influenced in part by sex hormones. There are also sex differences in health-related behaviors, such as frequency of visits to health professionals and use of complementary medicine, but these have not been well studied. There are also sex differ- ences in the incidence and course of some mental disorders and substance abuse (National Institutes of Health, Office of Research on Women's Health, l999b). These differences in mental health may also produce dif- ferences in physical health.
SEX AFFECTS BEHAVIOR AND PERCEPTION Cognitive Function 85 A large body of research has now converged to indicate that there are sex differences in specific areas of cognitive function. Although there has been some controversy over the proverbial question of which sex is the smarter one, a reasonable conclusion reached by many scientists is that there are no meaningful differences in intelligence between males and females (Halpern, 2000~. A more probing question asks if there are par- ticular areas of thinking or problem solving in which males and females differ; such cognitive abilities are referred to as "sexually dimorphic be- haviors." Before reviewing the research findings, it is important to bear in mind several factors. (l) In general, there is a marked overlap in the abilities of males and females. In some cases, the sex differences are most marked at the extreme ends of a particular ability, for example, among those who are the most skilled (Figure 4-l) (Hampson, in press; Hampson and 0.4~ 0.3 - ~ 0.2- IL 0.1 - o.o o m-\ \ \ / / \ \ , \ \ / / \ \ / / \ \ / / \ \ / / / / \ \ / - / / /// / \ \ \ \ \ \ \ \\ \ ·\ \ it"" 1 1 1 \ 20 40 60 80 100 Test Score Women Men FIGURE 4-1 Frequency distribution of scores on a hypothetical cognitive test plotted separately by sex. As a consequence of the differences in the means, the number of individuals scoring above a given point will differ for the two sexes; for example, the mean for women is higher than the mean for men such that only 25 to 30 percent of males score above the mean score for females. Source: Hamp- son and Kimura (1992, Figure 12-1) Reprinted, with permission, from l. B. Becker, S. M. Breedlove, and D. Crews, Behavioral Endocrinology, Cambridge, MA: The MIT Press, 1992, p. 359.
86 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH Kimura, 1992~. Although there may be slight but significant differences between the mean scores for males and females on some tests, they are invariably smaller than the differences between the highest- and lowest- scoring males (or females) on the same tests. (2) When differences are noted, they may apply only to individuals at a specific age or stage of life. (3) Finally, how an ability is measured may affect the results, for example, whether the response is multiple choice, fill in the blank, short essay, or oral. Cognitive abilities can be subdivided and considered in any number of ways. Maccoby and lacklin (1974) prepared a useful classification in which they delineated three general cognitive domains demonstrating sex differences: verbal, quantitative, and visuospatial abilities. Although for ease of presentation the report refers to these three main groups of cognitive abilities, these encompass heterogeneous areas of function, with each one representing several different functions. Furthermore, the spe- cific cognitive processes of interest may be assessed quite differently, often leading to conflicting results. Despite these caveats, it should be noted that a reasonable consensus has emerged relating sex differences to specific patterns of cognitive func- tion: in general, women most often demonstrate an advantage in verbal abilities particularly verbal fluency, speech production, the ability to decode a language, and spelling; perceptual speed and accuracy; and fine motor skills whereas men frequently show an advantage on tests of spatial abilities, quantitative abilities, and gross motor strength (Hampson, in press; Hampson and Kimura, 1992~. The following sections summarize data that support this general statement. Verbal Abilities Although it is often stated that females demonstrate better verbal abilities than males, it is important to note, as Halpern (2000) has, that "the term verbal abilities is not a unitary concept. The term applies to all components of language usage: word fluency, which is the ability to gen- erate words (both in isolation and in a meaningful context), grammar, spelling, reading, writing, verbal analogies, vocabulary, and oral compre- hension. The size and reliability of the sex differences depends on which of these aspects of language is being assessed" (pp.93-94~. Sex differences have been demonstrated for some but not all of these verbal abilities; however, when there is a difference, it invariably favors females. Two aspects of language showing perhaps the most consistent sex differences are verbal fluency and speech production, both of which share the need to have the ability to quickly access and to produce speech sounds and words. Verbal fluency (Hampson and Kimura, 1992; Hines, 1990; Hyde and Linn, 1988) is tested by having a subject name as many
SEX AFFECTS BEHAVIOR AND PERCEPTION 87 words as rapidly as possible according to either a phonological or sound- based cue (words that begin with a particular letter) or rhyming with a specific sound or by having the subject name words that belong to a certain category such as food or plants. In studies investigating sex differ- ences in verbal abilities, the largest difference (effect size [D] = 0.33) is typically found for speech production (Hyde and Linn, 1988), a measure, as discussed later, that is closely related to both reading and reading disability. Reliable sex differences have also been reported for spelling, another verbal ability closely related to reading; however, reports of sex differences in other areas of verbal ability such as vocabulary or reading comprehension have been inconsistent and are not considered reliable (Hampson and Kimura, 1992~. Sex differences have also been noted in tests of memory, particularly in tests of working memory (the ability to hold in memory information intended for temporary use). This is a particularly important ability be- cause it affects many aspects of a person's everyday life, for example, remembering a phone number given by the information operator, where the keys were just put down, or a message on the answering machine. Females have an advantage over males in remembering both verbal and nonverbal information. Females' superiority in verbal memory has re- ceived much attention, although their skill in remembering visual details, for example, spatial locations, has often been overlooked. As summarized below, males outperform females in visuospatial abilities when the task requires the manipulation of the spatial information; females, however, remember visual information better (Halpern, 2000; Hampson and Kimura, 1992~. Articulatory Skills, Manual Fine Motor Skills, and Perceptual Speed and Accuracy Females generally perform articulatory tasks or fine motor tasks more quickly and more adroitly than males. These skills all depend on the coordination of a sequence of movements. Articulatory skills are assessed by having the subject quickly repeat several syllables, for example, "pub tab huh, pub tab huh, pub tab huh," for 1 minute or try to say a tongue twister such as "sweet Susie swept sea shells" as rapidly as possible. Females also outperform males in carrying out fine hand movements such as rapidly placing pegs in small holes or in carrying out a simple sequence of hand movements (Hampson and Kimura, 1992~. In addition, females tend to perform better than males on tasks requiring perceptual speed and accuracy. This ability is assessed by asking subjects to quickly scan an array of symbols or figures and to indicate which one matches a previously indicated stimulus; for example, in the "random A's test," the
88 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH subject rapidly scans letters scattered over a page and is asked to cross out only the letter "A." Spatial and Quantitative Abilities Males demonstrate an advantage on tests of visuospatial ability (as reviewed by Maccoby and lacklin  and more recently by Halpern [2000~. According to Halpern (2000), this refers to the ability "to imagine what an irregular figure would look like if it were rotated in space or the ability to discern the relationship between shapes and objects" (p. 98~. Kerns and Berenbaum (1991) noted that a major issue is how to define and measure spatial ability. In a comprehensive meta-analysis, Linn and Petersen (1985) focused on three categories of spatial ability: spatial per- ception, mental rotation, and spatial visualization. The most consistent sex differences occur with measures of skills referred to as "spatial per- ception" and "mental rotation" (Linn and Petersen, 1985~. In particular, the mental rotation task has demonstrated the most sensitivity at detect- ing sex differences in spatial ability (Sanders et al., 1982~; here, a subject is asked to imagine how a figure would appear if it were rotated in a two- or three-dimensional space. Sex differences in quantitative abilities have also been reported. Here, it is important to ask "what" particular abilities and in "which people." Quantitative abilities refer to a heterogeneous group of abilities; depend- ing on the specific ability tested, males or females will have an advantage. For example, males seem to outperform females on tests of geometry, measurement, probability, and statistics as well as on tests of spatial and mechanical reasoning (Stones et al., 1982; Stumpf and Stanley, 1998~. Some have suggested that the male advantage in quantitative abilities reflects the male's use of visuospatial approaches for problem solving. In con- trast, females perform better on measures of calculation and also on tests in which the problem requires much reading. Perhaps the most important finding from the various research studies is that differences in math ability are much smaller toward the middle of the distribution, where most males and females are represented, and are most pronounced at the upper end of the distribution. Males consistently outperform females on tests of quantitative ability, for example, the math- ematics portion of the Scholastic Aptitude Test (SAT). Competitions among seventh and eighth grade boys and girls held to identify math- ematically precocious youth on the basis of scores on the mathematics portion of the SAT greatly favor boys. A consistent finding on these tests is that differences between boys and girls tend to increase at the higher levels of performance. Thus, boys outscore girls 2:1 at scores of 500 and above, 5:1 at scores of 600 and above, and 17:1 at the highest scores, 700 and above (Benbow, 1988, Stanley and Benbow, 1982~. One problem in
SEX AFFECTS BEHAVIOR AND PERCEPTION 89 interpreting such results is that the best predictor of performance on such standardized mathematics tests is experience. That is, most of the stu- dents who score the highest are enrolled in high-level mathematics courses. The data also indicate that many more males than females are enrolled in these high-level mathematics courses (Iones, 1984~. However, even when one controls for the number of advanced mathematics courses, males continue to have an advantage, albeit a much smaller one (Meece et al., 1982~. Newer studies are shedding light on the nature of sex differences in quantitative abilities. A recent analysis (Gallagher et al., 2000) indicated that males performed better on various types of mathematics questions that had in common a dependence on a strategy to "construct and men- tally transform a mental representation" (Halpern, 2000, p.117~. This sug- gests that it is not the type of mathematics problem that is important in evaluating sex differences but the kind of strategy required to solve it that is critical in determining whether males or females have ability. Reviews of the relationship between quantitative skills and spatial ability find that spatial ability is an important factor in predicting performance on ad- vanced mathematics tests and that this relationship is especially strong at the highest levels of mathematics performance (Halpern, 2000~. EFFECTS OF HORMONES ON BEHAVIOR AND COGNITION Prenatal Androgens and Sex Differentiation of Human Behavior There is now good evidence that human behavioral sex differences are influenced by sex hormones present during prenatal development, confirming findings from studies with other mammalian species (de- scribed in Chapter 3~. These hormones act by "organizing" neural sys- tems that mediate behavior later in life. Much of the evidence about the behavioral effects of prenatal sex hormones comes from individuals with clinical conditions that alter these hormones (so-called experiments of nature), although in recent years there has been confirming evidence from studies with individuals with circulating concentrations of hormones in the normal range. The following section provides an illustration of work done in this area; for detailed reviews of hormonal influences on human behavior, see Berenbaum (1998), Collaer and Hines (1995), Hampson and Kimura (1992), and Wilson (1999~. Prenatal androgens alone do not determine behavioral sex differences. Social and environmental factors undoubtedly contribute to differences between males and females, but the focus of this section is on genetic and physiological factors. Rather than considering physiological-hormonal and social explanations as being mutually exclusive, however, it is impor- tant to think about how they might operate in concert to produce behav-
So EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH ioral sex differences. For example, biologically influenced traits may af- fect an individual's response to the environment or the way that the indi- vidual is treated by others; such effects have been demonstrated in other species (Clark and Galef, 1998; Fitch and Denenberg, 1998~. Studies of Females with Congenital Adrenal Hyperplasia A, , The evidence for hormonal influences on human behavior is illus- trated with findings from studies of females with congenital adrenal hy- perplasia (CAH), a genetic disease in which the fetus is exposed to high levels of androgens beginning early in gestation (Grumbach and Conte, 1998; Miller, 1996; White et al., 1987~. If sex differences in human behavior are affected by the levels of androgens that are present during early de- velopment, then females with CAH should be behaviorally more mascu- line and less feminine than a comparison group of females without CAH (the best comparison group consists of the unaffected sisters of females with CAH because they provide a control for general genetic and envi- ronmental backgrounds). Indeed, females with CAH do differ from their sisters in sex-typed behavior (for a detailed description, see Berenbaum [2000~. One of the largest differences between females with CAH and their unaffected sisters is in their activities: it is characteristic of girls with CAH to play with boys' toys in childhood and to be interested in boys' activities in adolescence (Berenbaum, 2000; Berenbaum and Snyder, 1995; Ehrhardt and Baker, 1974~. For a variety of other behaviors, the differences between females with CAH and their unaffected sisters are almost as large as the differences between typical males and typical females. This includes in- terest in babies (Leveroni and Berenbaum, 1998), reported likelihood of using aggression in conflict situations (Berenbaum and Resnick, 1997), and spatial ability (Hampson et al., 1998; Resnick et al., 1986~. Other dif- ferences between females with CAH and unaffected females are smaller relative to the difference between typical males and typical females. For example, most girls with CAH prefer girls as playmates (Berenbaum and Snyder, 1995), and most women with CAH are exclusively heterosexual in terms of their sexual fantasy and arousal characteristics (Zucker et al., 1996), although some do prefer boy playmates and some have bisexual fantasy and arousal characteristics. The differences are even smaller for gender identity: only a very small minority of females with CAH have male-typical gender identity or are gender dysphoric (Ehrhardt and Baker, ~ Masculine and feminine are empirically defined and refer to a person's relative position on traits that show sex differences.
SEX AFFECTS BEHAVIOR AND PERCEPTION 91 1974; Meyer-Bahlburg et al., 1996; Zucker et al., 1996~. This is consistent with the idea that male-typical gender identity requires a higher level or different timing of exposure to androgen than is characteristic of that for females with CAH, rearing as a male, or exposure to other genetic or hormonal factors unique to or more common in males, for example, the SPY gene. Limitations of Studies of Females with CAH Females with CAH do not provide a perfect test of the behavioral effects of exposure to androgens early in gestation because they differ from their unaffected sisters in a number of ways that might affect behav- ior. Of particular importance from a social perspective is the fact that females with CAH have masculinized genitalia, and it is possible that their masculinized behavior results from the treatment of these girls by their parents in response to their physical appearance (Quadagno et al., 1977~. Recent evidence, however, renders this explanation unlikely: the amount of time that girls with CAH spend playing with boys' toys is linearly related to their degree of prenatal androgen excess, as inferred from the degree of genetic mutation, and it decreases (rather than in- creases) when a parent is present (Nordenstrom et al., 1999; Servin, 1999~. Convergence of Evidence Given the limitations of studies of females with CAH, it is important to seek a convergence of evidence across methods for the behavioral ef- fects of hormones. Findings from studies of females with CAH have been confirmed from other "experiments of nature" and from studies of samples of typical (non-CAH) individuals. For example, girls who were exposed to masculinizing hormones because their mothers took medica- tion (androgenizing progestins) during pregnancy are more likely than their unexposed sisters to report that they would use aggression in con- flict situations (Reinisch, 1981~. Males with reduced androgen levels be- cause of an endocrine condition called idiopathic hypogonadotropic hy- pogonadism (IHH) have lower levels of spatial ability than controls; within the group of men with IHH, spatial ability correlated with testicu- lar volume and did not improve with androgen replacement therapy (in- dications that the low level of spatial ability was associated with low levels of androgen early in development and not at the time of testing) (Hier and Crowley, 1982~. Converging evidence for these special cases has come from studies of normal individuals with typical variations in prenatal hormone levels. For example, 7-year-old girls who had high levels of testosterone in utero (determined from measurements of the concentrations in amniotic fluid
92 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH at 14 to 16 weeks of gestation) had faster mental rotation (an aspect of spatial ability) capabilities than girls who had low levels of prenatal tes- tosterone (Grimshaw et al., 1995), and females with a male twin appear to be more masculine than females with a female twin on several traits, including sensation seeking (Resnick et al., 1993), spatial ability (Cole- Harding et al., 1988), and auditory characteristics (McFadden, 1993~. Gen- der-specific behavior in young adult women has been suggested to be related to their exposure to sex hormones during the second trimester of fetal development (Udry et al., 1995~. Although some of these studies are imperfect, the limitations are different from those of studies of individu- als with CAH. Understanding How Prenatal Androgens Might Affect Behavior (and Other Traits) The evidence presented above suggests that androgens present early in life do affect a variety of sex-typed behaviors and that the effects are complex (Wilson, 1999~. It is unclear what might account for variations in a given behavior and among different behaviors among individuals. For example, it is not yet known what accounts for the fact that most girls with CAH play with boys' toys but that the majority of women with CAH are heterosexual. Investigators do not know the genetic, physiological, or stochastic factors that differentiate women with CAH who are aroused by members of the same sex from those who are not. It is not yet known why some individuals with male-typical prenatal hormone levels develop fe- male-typical gender identity when they are reared as females but others develop male-typical gender identity in that environment. The develop- mental course of the behavioral effects of prenatal androgens is also not known. Finding the answers to these questions provides an opportunity to integrate hormonal explanations for behavioral sex differences with other explanations and to arrive at a more complete understanding of why males and females differ in their behavioral characteristics. What Accounts for Variations? Evidence from studies of other species indicates that differences in the nature or extent of hormone exposure can have implications for be- havior. First, the timing of hormone exposure is important. The early prenatal period has generally been thought to be the crucial time for the organizational effects of hormones on mammalian brain development and later behavior, but other periods may be important, too. In primates, for example, there appear to be several distinct periods during which behavior is sensitive to the effects of androgen, with different behaviors masculinized by exposure early versus late in gestation (Goy et al., 1988~:
SEX AFFECTS BEHAVIOR AND PERCEPTION 93 female rhesus macaques exposed to androgen early in gestation (and thus with virilized genitalia) show increased mounting behavior, whereas those exposed late in gestation (with no genital virilization) show in- creased rough play. Second, different aspects of physical and behavioral sexual differen- tiation may be affected by different forms of masculinizing hormones. For example, dihydrotestosterone is responsible for differentiation of the ex- ternal genitalia (Siiteri and Wilson, 1974~; in monkeys, dihydrotestoster- one and testosterone propionate have different effects on learning (Bachevalier and Hagger, 1991~. In some species, male-typical develop- ment results from estradiol metabolized from androgen in the brain, al- though aromatized estrogens do not appear to play a role in masculiniz- ing the human brain or behavior (Grumbach and Auchus, 1999~. Third, the effects of specific hormones may be modified by other hormones (Goy and McEwen, 1980~. There is increasing recognition of the importance of ovarian estrogens for both physical and behavioral sexual differentiation, and it seems likely that androgens are modified by the effects of ovarian estrogens. Fourth, the behavioral effects of hormones may be modified by indi- vidual differences in hormone receptor sensitivity. In both human and nonhuman species, sex differences and the effects of hormones on behavior are influenced by the social environment, with hormones having their greatest effects on behaviors that show consistent sex differences among affected individuals (Wallen, 1996~. The theories and evidence regarding the role of cognitive and social factors in produc- ing sex differences have recently been reviewed by others (e.g., Bussey and Bandura , Lytton and Romney , Maccoby , and Ruble and Martin [1998~. It seems likely that although prenatal hormones contribute to some behavioral sex differences, they do not act alone and do not produce all sex differences. For all behaviors studied, the differences between females with CAH and unaffected females are less than the differences between typical males and typical females. Although rearing as a female is not always sufficient to produce female gender identity (Diamond and Sigmundson, 1997; Wilson, 1999), in some cases it may be (Bradley et al., 1998~. It is unclear how much these results can be explained by sex-related socialization and how much may ultimately be explained by hormonal factors that have not been delineated, such as ovarian estrogens. Mechanisms by Which Androgens Affect Behavior Although there is good evidence that at least some of the sex differ- ences in behavior are influenced by the prenatal androgens operating on the developing brain, it is not at all clear how that happens. For example,
94 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH it is not known what parts of the brain are particularly susceptible to organizational changes induced by high levels of androgen or what basic behavioral mechanisms cause someone who is exposed to high prenatal levels of androgen to play with boys' toys. Hormone exposure might also affect behavior through the selection and interpretation of the social environment, particularly those aspects related to sex. Recent studies of gender development in typical children suggest that the social environment is actively constructed and interpreted in ways that reinforce sex differences through the use of gender identity and gender labels (for a review, see Ruble and Martin [1998~. Similar studies with children with variations in sexual differentiation (Table 3-2) have the potential to provide information about factors that affect gender cognition and interpretation of the social environment. Such studies should also help provide an understanding of the development of behav- ioral sex differences, that is, how differences between boys and girls arise as a result of maturation and transactions with the physical and social environments. A Counterview: The Artificial Separation of Biology from Environment The concept that physiological and behavioral characteristics are pre- natally organized by hormones, as first proposed by Phoenix et al. (1959), has been fundamental to understanding the development of sex differ- ences in mammals. As is true for many overarching theoretical constructs, it has been significantly modified (Arnold and Breedlove, 1985) and has recently been reviewed critically (Fausto-Sterling, 2000; Wilson, 1999~. Some of the anatomical, physiological, and biochemical changes resulting from the early organizational effects of hormones have been revealed, particularly in the central nervous system (Breedlove, 1994; Gorski et al., 1978~. The effects of certain genes, such as the gene coding for the classic estrogen receptor alpha, on body and brain development depend on the sex of the individual in which it is expressed. McCarthy and colleagues (1993) have been able to explain some of the organizational effects at the molecular level by showing that neural estro- gen receptors are essential during the prenatal period for masculinization and defeminization of rats. Using estrogen receptor knockout (ERKO) mice, Simerly and colleagues (1997) have also demonstrated estrogen re- ceptor-dependent sexual differentiation. Such research is just beginning to reveal the mechanisms by which the effects of hormones early in gestation have long-term consequences. Among the mechanisms underlying the organizational concept are that the development of particular behaviors can be mediated and affected by a number of postnatal environmental experiences (Moore and Rogers,
SEX AFFECTS BEHAVIOR AND PERCEPTION 95 1984; reviewed by Fausto-Sterling [2000~. The "organizational effect hy- pothesis" thus serves as a useful framework on which to attach facts as they become available and to stimulate additional research on the com- plex array of factors that produce sex and gender differences. Two members of the committee, however, raised concerns about the use of the term organizational effect. Although is clear that the organiza- tional effects of prenatal hormones on the later development of particular behaviors is mediated and affected by a large number of organismal fac- tors and postnatal effects, it is often not clear what such an effect is or might mean at the cellular level. Two members of the committee argued that continuing to use the phrase organizational effect as an explanation could preclude experiments that might reveal the actual mechanisms by which hormones, genes, and a variety of postnatal experiences produce the sex and gender differences of interest. Two members of the committee believed that reliance on analyses that divide variance into main effects and smaller contributing effects sidetracks other biologically appropriate analysis, such as pursuing de- velopmental understanding of the emergence of cognitive skills. It also does not enable researchers to see how experience and biology work to- gether to produce difference. What is called for here, argued the two committee members, is an approach that examines the mutual construc- tion of cognition by physiology and by experience during key periods of development. There are animal models that successfully integrate the study of hor- mones and experience as they contribute to cognitive and spatial abilities. For example, variations in maternal care in rats (specifically, the amount of licking and grooming) contribute to the development of spatial memory and learning. The effect is mediated by synaptogenesis in the hippocam- pus (Liu et al., 2000~. Maternal licking can, in turn, be affected by a variety of factors, including an odor developed in the pup in response to an individual pup's testosterone levels (Moore, 1990; Moore and Rogers, 1984, Moore et al., 1992~. This example illustrates part of a process by which a particular capacity emerges. Nature is not distinct from nurture, since maternal behavior responds to both pup odor and other inputs and directly influences pup brain development and, hence, the pup's behavior as an adult. Such effects are transmitted across generations. There are other examples of this sort in the literature on studies with animals (Crews, 2000; Gottlieb, 1997~. A variety of studies have shown postnatal effects on the development of hormonally influenced behaviors (summarized schematically in Figure 4-2~. For further discussion see Fausto-Sterling (2000) and Tobet and Fox (1992~.
96 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH in' Behavioral Development in Rodents NEURAL COMPONENTS: BRAIN AND CNS I ~ (At on stages and in all components of the nervous system) SENSORY INPUT MOTOR INPUT 'Hormones: Maternal/fetal (prenatal)! IPerinataI I IJuvenilel IPubertall Dim ~ , _ . |LIVIn9 Conditions| 4_ |Social Rearing ~ |Adult Behaviors| Dispersed or in territories Lab cages Isolated or not Stress Levels Cage Size Stimulation (physical and intellectual) ~ ·AII male `.AII female Same age? With mother? ·Multigenerational? Juvenile period . Gene expression (also affected by species and strain genotype) multi-animal choices? attest site familiarity? Attest site structure attest duration/frequency Observe in wild or semi-wild conditions? FIGURE 4-2 Behavioral development in rodents. Source: Fausto-Sterling (2000, p. 228~. Reprinted, with permission, from A. Fausto-Sterling. 2000. Sexing the Body: Gender Politics and the Construction of Sexuality. New York: Basic Books. Copyright 2000 by Basic Books, a Member of the Perseus Books Group. An Approach to Studying Multicausal Development Two members of the committee supported a new approach to study- ing multicausal development that has emerged in recent years. The gen- eral claims of developmental systems theory are as follows. 1. Some aspects of development are self-organizing, but these self- organizing aspects are rarely sufficient for the emergence of a trait or a behavior. For example, early in the embryogenesis of retina-brain nerve connections, nerve cells begin to fire spontaneously. This spontaneous nervous activity is, in turn, required for an initial "rough targeting" of cells between the thalamus and the cortex (Catalano and Shatz, 1998~. However, the rough embryonic connections still require postnatal visual experience for fully functional visual circuits to develop (Katz and Shatz, 1996~. 2. The initial spontaneous and apparently random activities of cells, such as nerve or muscle cells, eventually stabilize (via self-organizing properties and because of feedback from outside the body), leading to periods of great stability or stasis. Thus, certain features of development emerge as apparently fixed. Some features of development are more stable
SEX AFFECTS BEHAVIOR AND PERCEPTION 97 than others, but as researchers come to understand the systems that gen- erate and maintain them, they will come to understand both stasis and its loss. By overly focusing on the "main effect" rather than examining a developmental system out of which differences emerge and are main- tained, important data may be missed. 3. The emergence of sex and gender differences, then, "can be seen in dynamic terms as a series of states of stability, instability and phase shifts"(Thelen, 1995. p. 84) These ideas have been applied at length to cognitive development in general in a National Research Council and Institute of Medicine report entitled From Neurons to Neighborhoods (2000~. What remains is serious thought about how these ideas should be applied to the development of sex and gender. Cognitive Effects of Sex Hormones on Adults Sex hormones not only act in prenatal life, to organize the brain for later behavior, but also continue to exert effects later in life. This is clearly seen with respect to cognitive abilities. One of the areas most carefully studied has been the relationship between sex hormones in females and cognitive abilities, especially verbal skills. Studies of this relationship have found that performance on particular cognitive tests varies with changing hormone levels, demonstrating, too, that even mature neural systems in adult brains are responsive to the influence of sex hormones. Thus, in addition to their early influences on brain development, sex hormones may also exert influences later on. In addition, as noted below, sex hor- mones affect neural systems in adult women during their active repro- ductive years and postmenopausal years. During adult life, women's hormone levels fluctuate monthly with the menstrual cycle, and some studies have shown that these variations to some degree affect performance on certain tests of cognitive abilities (al- though the sizes of the effects were quite small). In three studies, Hampson (199Oa,b) and Hampson and Kimura (1988) tested women during their menses and during the preovulatory phase of their menstrual cycles to specifically compare states of minimal and maximal estrogen secretion, respectively. As hypothesized, the preovulatory phase, a time of rela- tively high estrogen levels, was found to be associated with modest de- creases in spatial ability and improved ability on tests of manual coordina- tion and articulatory skills. These findings have been confirmed by subsequent researchers (e.g., McCourt et al. , Moody , Phillips and Silverman , and Silverman and Phillips [1993~. A few studies, however, failed to detect any menstrual cycle effects (Gordon and Lee, 1993; Peters et al., 1995), although methodological differences could account for the different find- ings. For example, in the study by Peters and colleagues (1995), data on
98 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH the menstrual cycle were not carefully verified, and verbal self-reports can be very inaccurate. Epting and Overman (1998) also failed to replicate the performance fluctuations across the menstrual cycle. However, the women in the study of Epting and Overman (1998) were younger than women typically used for menstrual cycle studies. For example, in contrast, Hampson (199Oa,b) and Hampson and Kimura (1988) excluded women younger than age 21 years. Even though menstrual cycles can seem to be regular in young women, young women have higher incidences of anovulatory cycles and lower levels of ovarian output than women in their 20s and beyond. Thus, it is possible that the cognitive effects of menstrual cycle changes are genuinely weaker in women in their teens than in women over the age of 21 years. A particularly intriguing bit of evidence supporting the role of female sex hormones on language-related behaviors comes from studies of Koko, a female gorilla that has been trained to communicate using American Sign Language (Patterson et al., 1991~. Both the number of discrete signs used and the total number of signs per day rose in the follicular phase of her reproductive cycle, when Koko's estrogen concentrations were raised. The investigators speculate that the increase in manual and verbal output at midcycle could serve to enhance the possibility of conception through more effective signaling to the male as a part of the proceptive behavior complex. Taken together, these studies indicate that female sex hormones ap- pear to enhance performance of those skills usually performed better by females, whereas they cause a decrement in performance of those skills usually performed better by males. Hormonal levels in women also change during menopause, when the levels of the hormone estrogen undergo dramatic declines after the cessa- tion of cyclic ovarian function. Given the demonstrated sex differences in cognitive function that favor verbal abilities in females and the associa- tion of better performance of these skills during phases of the menstrual cycle when estrogen levels are high, there has been great interest in the effects of hormone replacement therapy (exogenous estrogen) on these cognitive abilities. Although evidence suggests that estrogen positively affects basic neural processes and cognitive function in animals (McEwen and Alves, 1999), the influence of estrogen on cognitive function in hu- mans, especially postmenopausal women, has been much more difficult to establish. Results to date from observational studies and clinical trials with women receiving hormone replacement therapy are far from consis- tent (Barrett-Connor, 1998b; Haskell et al., 1997; Rice et al., 1997; Sherwin, 1997; Yaffe et al., 1998~. Such inconsistency may reflect differences in the ages of the women studied. For example, at midlife, estrogens tend to have a positive effect
SEX AFFECTS BEHAVIOR AND PERCEPTION 99 on cognitive function (Shaywitz et al., 1999; Sherwin, 1997~. Studies with older populations have more varied results, with some indicating a posi- tive influence of estrogen on cognitive function (Jacobs et al., 1998; Resnick et al., 1997; Steffens et al., 1999) but others failing to show such an effect (Barrett-Connor and Kritz-Silverstein, 1993; Matthews et al., 1999~. Fur- thermore, it may be that the effects of estrogen on cognitive function are observed most strongly when the agent is first used, an effect noted in studies with animals (Miranda et al., 1999~. Nevertheless, when effects are observed they invariably tend to be positive influences on verbal func- tion, particularly verbal memory and verbal fluency. Most recently, estrogen has been found to improve the oral reading ability of postmenopausal women (S. E. Shaywitz et al., submitted for publication). The notion that estrogen has a positive influence on cogni- tive function received further support from another recent study in which postmenopausal women were monitored for a 6-year period, with their cognitive function measured initially and after 6 years. The amount of free circulating estrogen was also measured. The investigators found that women with the highest levels of hormone were those least likely to show signs of cognitive decline on testing (Yaffe et al., 1998~. Possible Mechanisms Influencing Effects of Estrogen on Cognitive Functions Is it reasonable to suppose that estrogen may have salutary effects on certain cognitive abilities? A large body of evidence supports the notion that estrogen has significant effects on neuronal function and affects a range of neural activities in mature animals (McEwen and Alves, 1999~. Thus, it is logical to suggest that estrogen has potent effects on central nervous system functioning, including cognitive functions. The next ques- tion relates to how estrogen may affect the specific cognitive functions demonstrated to be sensitive to its actions. To address this question, some investigators have turned to another large body of evidence, but in this instance the studies related to the process of reading (Shaywitz et al., 1998; Shaywitz et al., submitted.) Reading and Language Processes This discussion should be appreciated in the context that reading is related to language (as determined from a newer understanding of the reading process) and that within language the reading process is related to phonological processing. Interestingly, the key cognitive functions af- fected by estrogen (verbal fluency, verbal memory, and articulation) are the same cognitive processes that are deficient in individuals who have difficulty reading. These processes share a dependence on the need to
100 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH access the basic sound structure of words, that is, phonological process- ing. Synthesis of the findings from the literature on hormonal influences on cognitive function and from the literature on reading has led to the novel hypothesis that hormones, specifically estrogen, influence phono- logical processing, which in turn influences the development of verbal fluency, speech production, and reading skills. Current theory supported by substantial empirical data supports the belief that the same processes that serve language also serve reading. Although speaking is a universal behavior, not everyone learns to read; speaking is automatic, whereas reading must be learned. These observa- tions have led to the belief, from an evolutionary perspective, that the biological systems that serve reading are not newly developed but, rather, represent a modification of those biological processes already in place to serve language. One hypothesis suggests that the language apparatus forms a distinct biological system or module (Fodor, 1983) that is served by specific brain mechanisms and structures (Liberman and Mattingly, 1989; Liberman, 1989~. In this view, the same processes and processors that serve language also serve reading, the major difference being that speaking is automatic but reading must be learned. Much has been learned about the nature of the reading process and the component skills neces- sary for the acquisition of reading skills, particularly the importance of phonological processes (Casts, 1986; Shaywitz, 1996~. The Phoneme Reading and language share the same basic elemental unit, the pho- neme, the smallest unit of sound that gives meaning to a word. The pho- neme represents the sound structure that underlies all words, written or spoken. For words to be understood, spoken, read, stored, or retrieved, they must first be segmented into phonemes. Deficits in phonological processing have been intimately related to reading disability, and a large body of literature now indicates that a phonological core deficit is respon- sible for the difficulties that dyslexic children have in learning to read. Areas of impaired phonological processing consistently related to reading disability represent the same areas of language most sensitive to the actions of estrogen, that is, word fluency (e.g., naming) and speech production (e.g., speed of articulation). For example, deficits in naming colors, common objects, and numbers consistently characterize reading disability (Denckla and Rudel, 1976; Wolf, 1984; Wolf and Goodglass, 1986~. Problems with naming are conceptualized as reflecting problems with registering, storing, or retrieving phonemes (Casts, 1986~. Similarly, children with reading disabilities have difficulties with speech produc- tion (Casts, 1986, 1989~; in this case, poor readers have difficulty either selecting, ordering, or articulating phonemes during speech (Casts, 1986~.
SEX AFFECTS BEHAVIOR AND PERCEPTION Reading and the Actions of Estrogen 101 This very brief summary has reviewed information that supports the well-accepted belief that reading and reading disability are related to language and, in particular, to phonological processing. Disruptions in phonological processing predict, characterize, and explain many of the difficulties experienced by poor readers. These data relating phonological processing to reading and reading disability converge with the consistent findings that within the domain of verbal ability, the specific measures that are sensitive to the positive effects of estrogen are those very same tasks that tap specific components of phonological processing disrupted in those with reading disabilities. Such converging findings from two different areas of investigation have important implications: they suggest that there may be commonali- ties to the mechanisms of reading and reading disability and to the ac- tions of the female sex hormone, estrogen. Both reading and reading dis- ability reflect language, and estrogen's strongest influence is on verbal abilities. Within the language system, phonological processing has been identified as the critical component relating to reading and reading dis- abilities; and among the actions of estrogen on language, the areas most sensitive to hormonal effects are verbal fluency, naming, and speech pro- duction-speed of articulation, those areas most related to phonological processing. Together, these findings suggest that the deficit in reading ability and the action of estrogen on specific verbal skills may have a common base: phonological processing. These findings are exciting because they repre- sent a link between studies of the influence of sex hormones on cognitive function and studies relating specific cognitive subskills to the reading process. They indicate that the notion that reading may be influenced by female sex hormone is consonant with current theories of reading. Fur- thermore, it is of particular interest that Hampson (199Oa,b) reported that a surge in estradiol levels enhanced performance on tests of specific ver- bal skills, including color naming and syllable repetition. Similar results have also been found with exogenously administered hormones in stud- ies of the effects of estrogen replacement therapy in postmenopausal women; for example, Kimura (1995) reported that another measure of phonological processing, speeded articulation, was performed better by postmenopausal women in the on phase of their estrogen replacement therapy. Thus, the specific verbal abilities that are sensitive to change in estradiol levels are those verbal skills that also reflect phonological pro- cessing and that have been implicated in the reading process. Evidence relating sex hormones to reading comes from epidemiologi- cal and developmental studies suggesting that hormonal changes during puberty in females may influence phonological processing skills and read-
102 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH ing. These data suggest that although sex ratios for dyslexia are compa- rable during childhood (Shaywitz et al., 1990), these ratios may change as children mature into adults. Thus, studies of adults in the same family have consistently indicated sex ratios for dyslexia of 1:1.5 to 1:1.8 in favor of males (DeFries et al., 1991), whereas investigations of compensated dyslexics (adults who were dyslexic as children but who are able to read with some degree of accuracy as adults) report that preponderance (72 percent) are females (Lefly and Pennington, 1991~. Thus, reports of equal sex ratios for children with dyslexia, in contrast to a sex ratio favoring males in studies of adults in the same family, may reflect hormonal influ- ences associated with puberty. Similarly, reports of an increased prevalence of females in the group of dyslexic readers who have become more accurate readers but who are not automatic readers may reflect the positive effects of female sex hor- mones on dyslexic readers. Such findings could be interpreted to suggest that as young women progress through puberty they improve their lin- guistic skills and that these improved linguistic skills allow them to com- pensate, to some degree, for their reading disability. This explanation is appealing because it is parsimonious, accounting both for the observed difference in prevalence ratios for children compared with those for adults and for the known positive effects of female sex hormones on the cogni- tive and linguistic skills that underlie reading. More recent studies, using sophisticated imaging technology, demon- strate sex differences for language, specifically for phonological process- ing, and add further evidence that supports the notion that estrogen may exert its effects on cognitive function through its actions on phonological processing (Shaywitz et al., 1995~. Thus, researchers interested in localiz- ing the neural systems used for reading, particularly those engaged by phonological processing, studied a group of 19 men and 19 women using functional magnetic resonance imaging (fMRI). fMRI captures changes in blood flow associated with cerebral activity and acts to identify those regions of the brain used to perform a cognitive task. The men and women in that study were asked to sound out nonsense words, for example, "fete" and "jeat," and to indicate if the pairs of words rhymed or not. To carry out this task, the subject must sound out the words, that is, rely on phonological processing. The results of the study were remarkable: they indicated that men and women carry out phonological processing using different neural systems; that is, men rely on the left inferior frontal gyros (Broca's area), whereas women use both the left and the right inferior frontal gyri (Figure 4-3) (Shaywitz et al., 1995~. These findings of a sex difference in brain systems underlying phono- logical processing stimulated efforts to understand the origins of these differences, including the possible influence of female sex hormones. In one investigation, Shaywitz and colleagues (1998) studied a group of post-
SEX AFFECTS BEHAVIOR AND PERCEPTION 103 FIGURE 4-3 Composite images of the distribution of activations upon perfor- mance of rhyme-case tasks (phonological processing) for 19 males (left image) and 19 females (right image). Males show unilateral activation, primarily in the left inferior frontal gyros. In females, phonological processing activates both the left and right inferior frontal gyri. (Color dots represent pixels for which the mean value for the split t statistic from the average for the 19 subjects was greater than 0.4. Dark red dots represent mean values close to 0.4; yellow dots represent mean values that approach 1.0) Source: Shaywitz et al. (1995) Reprinted, with permission, from B. A. Shaywitz, S. E. Shaywitz, K. Pugh, R. T. Constable, P. Skudlarski, R. K. Fulbright, R. A. Bronan, l. M. Fletcher, D. P. Shankweiler, L. Katz, and l. C. Gore. 1995. Sex differences in the functional organization of the brain for language. Nature 373:607-609. Copyright 1995 by Nature. menopausal women (mean age, 50 years) while they were on and off estrogen replacement therapy. Each woman's brain was imaged as she was given a series of tasks that tested her verbal and nonverbal working memory. Examination of the fMRI scans of the women on and off estro- gen replacement therapy showed a significant influence of estrogen on the neural systems for memory. Overall, these results demonstrated that brain plasticity continues into midlife and that "functional brain organi- zation in women (and, we assume, men) is neither fixed nor immutable" (Shaywitz et al., 1998, p. 1201~. More specifically, prior studies had indi- cated different patterns of neural organization for memory in older men and women compared with those in younger men and women. Shaywitz and colleagues (1999) found that estrogen usage in older individuals was associated with brain activation patterns different from those in younger
104 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH individuals. Interestingly, the inferior parietal lobule, a region of the brain known to be engaged by phonological processing, was activated while a woman was on estrogen, supporting the notion that estrogen influences phonological processing. To this large body of evidence linking estrogen to phonological pro- cessing is a new report indicating that estrogen improves the reading ability in postmenopausal women (Shaywitz et al., submitted). Thus, a range of evidence taken together indicates that estrogen may exert its actions through its influence on a fundamental component of the lan- guage system, phonological processing, which is critical for speaking, remembering, and reading. This hypothesis brings together seemingly disparate data and provides a reasonable explanation for at least one group of sex differences in cognitive function that have been observed. Future Work Future studies are needed to focus on the possible relationship be- tween male sex hormones and reported male advantages in spatial and quantitative functions. A good beginning has been made in a recent in- vestigation that used fMRI to study brain organization during the perfor- mance of navigational skills (Gron et al., 2000~. In that study, a group of males and females were imaged as they searched their way out of a three- dimensional, virtual-reality maze. The investigators noted sex differences in the regions of the brain activated; males activated the left hippocampus but females activated the right parietal cortex and right prefrontal cortex during the same navigational task. Similar to the findings of sex differ- ences in brain organization for observed differences in language skills, specifically, phonological processing, these findings now provide a neu- ral basis for observed sex differences in spatial performance. One inter- pretation of these findings is that females rely mostly on landmark cues (Sandstrom et al., 1998) and that activation of the prefrontal region re- flects efforts to hold these cues in working memory. On the other hand, males depend on both landmark and geometric cues so that they activate the hippocampus, which allows them to process geometric cues. The advent of newer brain imaging technologies should now provide more information on the underlying neural substrate of a range of ob- served sex differences in cognitive function. It is hoped that, with time, new hypotheses will also emerge that offer possible explanations of the mechanisms that underlie the observed sex differences in spatial perfor- mance and other areas of cognition and the observed sex differences in brain organization.
SEX AFFECTS BEHAVIOR AND PERCEPTION SEX DIFFERENCES IN PERCEPTION OF PAIN 105 The issue of sex differences in pain during adulthood has been the subject of considerable research and meta-analysis, much of which has recently been reviewed (Berkley, 1997a,b; Berkley and Holdcroft, 1999; Derbyshire, 1997; Fillingim, 2000; Fillingim and Maixner, 1996; Riley et al., 1998; Unruh, 1996~. In general, the data consistently show that females are more sensitive than males to nociceptive (potentially or frankly dam- aging) stimuli, including those that occur in internal organs (Giamberardino, 2000~. Added to this greater sensitivity of females is the higher prevalence of many painful disorders in females (Table 4-1~. On the other hand, perhaps because there is a more permissive atmosphere for women to acknowledge the threat of injury (i.e., pain) and perceive dysfunction (Taylor et al., 2000), women seek more and more varied forms of health care than men, making use of it in a more positive, multidimen- sional manner, thereby deriving more relief than men (Affleck et al., 1999; Robinson et al., 2000; Unruh et al., 1999~. Also evident in all of the reviews, however, is frustration with apply- ing to individuals the generality that females are more sensitive to pain because exceptions abound. Whether it be humans or nonhuman animals, the existence or even the direction of sex-related differences in pain have been shown to vary with different situations, for example, as one ages, by testing paradigm or setting, by type or location of pain, by subject demo- graphics, by reproductive status, by genetic profile, by treatment utiliza- tion behavior, by the way in which pain is measured (and by whom), by analgesic, and by responses to different treatments. Thus, one is faced with what may be two separate problems: determining what factors un- derlie what appears to be a general greater female vulnerability to pain over the female's lifetime versus understanding how being female or male contributes to individual and circumstantial variations in pain and re- sponses to treatment. Some constructive answers with potential applica- tion to human health are beginning to emerge from the latter approach as a result of research with both laboratory animals (mostly rodents) and humans, with implications for either sex. Two examples follow. Genes, Nociception, and Sensitivity to Analgesics Neuroscientist Mogil (2000) has recently published an elegant series of studies on the responses of different strains of rodents to noxious stimu- lation and to analgesics. He summarizes this work as follows: "Recent findings in my laboratory strongly suggest that the modulatory effect of either of these organismic variables (genetic variation, sex) on pain-re-
106 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH TABLE 4-1 Sex Prevalences of Some Common Painful Syndromes and Potential Contributing Causes Female Prevalence Male Prevalence Head and Neck Migraine headache with aura Chronic tension headache Postdural puncture headache Cervicogenic headache Tic douloureux Temporomandibular disorder Occipital neuralgia Atypical odontalgia Burning tongue Carotodynia Temporal arteritis Chronic paroxysmal hemicrania Carpal tunnel syndrome Raynaud's disease Chilblains Reflex sympathetic dystrophy Chronic venous insufficiency Piriformis syndrome Peroneal muscular atrophyC, d Esophagitis Gallbladder diseased Irritable bowel syndrome Interstitial cystitis Proctalgia fugax Chronic constipation Fibromyalgia syndrome Multiple sclerosis,Tg Rheumatoid arthritis, T Acute intermittent porphyriaC Lupus erythematosus, T Migraine without aura Cluster headache Posttraumatic headache Paratrigeminal syndromes Limbs Internal Organs General Thromboangiitis obliteransb Hemophilic arthropathyC Brachial plexus neuropathy Pancoast tumbrel Pancreatic disease Duodenal ulcer Postherpetic neuralgia a Raeder's syndrome. b Buerger's disease. c Sex-linked inheritance is a potential contributory cause. d Charcot-Marie-Tooth disease. e Bronchogenic carcinoma. f Lifestyle is a potential contributory cause. g T. autoimmune. SOURCE: Berkley and Holdcroft (1999~. Sex prevalence information is mainly from Merskey and Bogduk (1994) and was cross-checked by using MedLine and other search sources.
SEX AFFECTS BEHAVIOR AND PERCEPTION 107 fated traits can only be understood in the context of the other. That is, sex differences vary with, and are specific to, the particular genetic back- ground in question, and genetic differences (between strains) can some- times only be observed in one sex but not the other" (Mogil, 2000, p. 26~. It is important to understand that in these studies the effects were re- vealed by using a specific set of experimental tests of nociception (tail withdrawal from a 49°C hot plate) and antinociception (reduction in nociception with systemic morphine or a K-opioid or cannabinoid recep- tor agonist). As Mogil readily admits, the results of such tests can be influenced by many factors, such as time of day, the type of stimulus (mechanical versus thermal), diet, pre- and postnatal stress, housing (in a group versus in isolation), current or prior injury, reproductive status of the comparison females, and more (Berkley,2000~. Thus, Mogil's observa- tions herald a huge potential for the emergence of individual differences in phenotype as genotypic influences are further affected by life's accu- mulating circumstances. Mechanisms of Analgesia, Sex Steroid Hormones, and Central Sensitization An exciting series of findings from research with rodents is that sex differences emerge from complex interactions between stress and endog- enous analgesia. In other words, it may be that there are more potent sex differences in mechanisms of pain and analgesia than in measured pain behaviors. The differences seem to lie in how sex steroid hormones exert their effects (Aloisi, 2000; Gintzler and Liu, 2000; Sternberg and Wachterman, 2000~. Thus, stress gives rise to an analgesia mediated by a nonopioid, N-methyl D-aspartate (NMDA), that is present primarily in males but that is also present in some females: those who have been ovariectomized or who were neonatally exposed to testosterone. Stress also gives rise to an estrogen-dependent, nonopioid, non-NMDA-medi- ated analgesia present only in intact females, the mechanisms of which are unknown. Furthermore, the hormonal milieu of pregnancy creates an antinociception involving 6- and K-opioid systems but not ,u-opioid sys- tems. When such an analgesia is created artificially by hormone treatments in gonadectomized rats, in females the analgesia results from a synergis- tic combination of spinal K-opioid, 6-opioid, and oc2-noradrenergic path- ways but not ,u-opioid pathways, whereas in males the analgesia results from independent additive contributions of spinal K- and ,u-opioid path- ways but neither the 6-opioid nor the oc2-noradrenergic pathway. Finally, estrogen can influence cardiovascular responses (e.g., promo- tion of vasodilatory or spasmodic effects) and neuronal responses (e.g., expression of the trkA gene) to injury, thereby influencing nociception
108 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH differently in females and males. Some of these findings may relate to recent studies with humans showing that K-opioids are more effective analgesics in young adult women than in young adult men who have undergone molar tooth extraction (Gear et al., 1996, 1999~. Significance for Human Health Assuming that the two sets of observations just described are appli- cable to humans, what might their significance be for health? One obvious area is in the development of analgesic medications. Is it possible that at some time in the foreseeable future analgesics will be prescribed on the basis of an individual's genotype, sex, and reproductive status? Given the first discussion on genotype, such a strategy would likely be pursued only with great care and only in special circumstances (Mogil et al., 2000~. For example, individuals with mutations that lead to altered functioning of the cytochrome P450 2D6 enzyme are likely to be prescribed some analgesic other than codeine because they are unable to transform co- deine into morphine (Sindrup and Brosen,1995~. Drug development must take into consideration both the sex and the reproductive status of the research subjects not only during all phases of clinical trials but also dur- ing the drug development stages of basic research with animals. On the other hand, before concluding that a specific drug may even- tually be prescribed on the basis of the sex of the individual or the repro- ductive or hormonal status of the patient, it also seems important to con- sider how stress exerts its cumulative effects over the life span of an individual. Of relevance here is the plasticity of neural function: the abil- ity of neural elements to change their phenotype, to "learn." Considerable research on these changes in the context of pain has led to the discovery of what is called "central sensitization," which is an enhanced responsive- ness of central nervous system neurons induced by intense stimulation or injury or by a stressor that, importantly, continues long after the initial noxious event has resolved (Dubner and Ruda, 1992; McMahon et al., 1993~. Thus, if the different complex modulatory mechanisms of endog- enous sex steroids discovered in female and male rats also exist in human females and males, it is likely that how they influence pain behaviors and the effects of analgesics will change in an ever more complicated manner as the different sociocultural stressors in human females and males exert themselves across their life spans. It may therefore be that one of the most important clinical insights from these two disparate areas of research (mechanisms of endogenous analgesia and central sensitization) is real- ization of the importance of understanding the chronology and sociocul- tural context of stressor events for each individual, with that individual's being female or male forming only one of many components considered for drug prescription and therapeutic strategies. Two examples follow.
SEX AFFECTS BEHAVIOR AND PERCEPTION Sex Differences in Efficacy of ,u-Opioids in Clinical Setting 109 Miaskowski and colleagues (2000) have carried out an extensive re- view of the clinical literature and have concluded that ,u-opioid analgesics are more effective in human females than in human males. Verification of such a conclusion might lead toward research on the development of different analgesics or combinations of analgesics for use as treatments for males. However, it is important to consider the basis for this conclu- sion. As pointed out by those investigators, the effects have been mea- sured mainly by determination of the amount of ,u-opioid medication that females and males consume postsurgically. In most studies males con- sume more medication than females (when the levels are measured) to achieve comparable levels of pain reduction. The question of whether the consumption of larger amounts of ,u-opioids postsurgically by males indi- cates that they have lower levels of efficacy in males then arises. One possible way to interpret the finding of greater ,u-opioid usage by males is to consider the results of other studies demonstrating that fe- males and males make use of different strategies to reduce pain. As re- cently reviewed by Robinson and colleagues (2000), females bring a greater variety of coping strategies to bear on their pains than males; that is, females make greater use of what might be called self-polytherapy than males (Berkley and Holdcroft, 1999) (Table 4-2~. It is therefore pos- sible that females use smaller amounts of ,u-opioids because they are able to engage other forms of positive coping strategies, thereby reducing their need for opioids, and that males use more ,u-opioids because that is the only relief they can find. Thus, efficacy depends not simply on whether the drug user is female or male but, rather, depends on sociocultural factors. Such an hypothesis can be tested. Is it in fact the case that in the postoperative setting females engage more coping mechanisms than males? On the other hand, do individuals who have learned to engage multiple coping measures, regardless of their sex, use smaller amounts of opioid medication than others? If so, could opioid usage be reduced over- all if individuals were encouraged and educated on how to engage addi- tional constructive coping mechanisms? Impact of Menstrual Cycle on Pain Along with genetic and developmentally programmed sex differences in neural organization and physiology, the entire nervous system is po- tently influenced by the hormonal milieu of the individual (McEwen, 1999; McEwen and Alves, 1999~. One arena in which this influence be- comes evident is the ovarian cycle (one should keep in mind, however, that the basis for ovarian cyclicity in any realm of physiology or behavior may not necessarily be entirely due to the hormonal milieu). Several stud-
110 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH TABLE 4-2 Growing List of Therapies for Pain Somatic Interventions Drugs Situational Approaches Primary analgesics Nonsteroidal anti- inflammatory agents A cet amin op hen Opioids Other analgesics a2 Agonists p-Adrenergic antagonists Antidepressants Antic onvuls ants Antiarrhythmics Calcium channel blockers Cannabinoids Corticosteroids Cox-2 inhibitors y-Aminobutyric acid type B agonists Serotonin agonists Adjuvants Antihistamines Laxatives Neuroleptics Routes Topical, transdermal, oral Buccal, sublingual, intranasal Vaginal, rectal Inhalation Intramuscular, intraperitoneal Intravenous Epidural, intrathecal In tr av e nt ri cu. l a r Simple Heat or cold Exercise Massage Vibration Relaxation Minimally invasive Physical therapy Traction Manipulation Ultrasound Transcutaneous electrical nerve stimulation Acupuncture Local anesthetics Invasive Radiation therapy Dorsal column stimulation Nerve blocks Neurectomy Local ganglion blocks Sympathectomy Rhizotomy Dorsal root entry zone lesions Punctate midline myelotomy Limited myelotomy Commissural myelotomy Cordotomy Brain stimulation Brain lesions Clinician Education Attitude Clinical setting and arrangement Self Education Meditation Diet Art, music, poetry, performing arts Sports, gardening, hobbies Humor Aroma therapy Religion Pets Interactive Hypnosis Biofeedback Support groups Advocacy groups Networking Self-help groups Structured settings Group therapy Family counseling Job counseling Cognitive therapy Behavioral therapy Psychotherapy Multidisciplinary clinic Hospice SOURCE: Berkley (2000~. NOTE: Women are more likely than men to take advantage of most of the therapies listed in the "simple" and "minimally invasive" sections of the middle column and nearly all of the therapies listed in the "situational approaches" column (Berkley and Holdcroft, 1999~.
SEX AFFECTS BEHAVIOR AND PERCEPTION 111 ies with rodents have shown the powerful impact of the ovarian (estrous) cycle on the functioning not only of the parts of the brain associated with reproductive functions but on other regions of the brain as well, such as (so far) the hippocampus, striatum, inferior olive, cerebellum, and dorsal column nucleus (Becker, 1999; Bradshaw and Berkley, 2000; Smith and Chapin, 1996a,b; 1998; Woolley and McEwen, 1993; Xiao and Becker, 1994~. Importantly, these changes are not always predictable according to the hormonal milieu (Bradshaw and Berkley, 2000~. Given that brain imaging studies show that many parts of the brain are engaged when the subject is in pain (Ingvar and Hsieh, 1999), it is not surprising that numerous studies have found that pain can vary with the menstrual cycle, especially pain that occurs when noxious stimuli are delivered to healthy individuals under certain tightly controlled experi- mental conditions (Riley et al., 1999~. One consequence of this situation is that results of studies comparing pain in young adult females and young adult males may depend on the time of the menstrual cycle in which the women's pain was assessed. The clinical significance of these findings, however, is unclear be- cause the existence and pattern of the menstrual effects that have been reported are not consistent, especially for painful clinical conditions (Berkley, 1997a,b; Fillingim and Ness, 2000~. Part of the inconsistency across studies may be due to technical factors, such as how different parts of the menstrual cycle are classified and the manner in which the analysis has been made. Given that brain imaging studies, however, are beginning to show that the brain regions engaged while an individual is under painful conditions vary with the individual (Davis et al., 1998; Gelnar et al., 1999), it is relevant to consider other factors. For example, a recent study compared skin and muscle pain thresholds in the lower abdomen and limbs across the menstrual cycle in women with severe menstrual pain (dysmenorrhea) and women without dysmenorrhea and across the month in similarly aged young men (Giamberardino et al., 1997~. For the men, limb pain threshold did not vary across the month, but abdominal thresholds could not be measured because of the men's extremely high sensitivity (all refused further testing of this region after the first set of trials). For women, the presence of dysmenorrhea gave rise to a general- ized muscle (but not skin) hyperalgesia and a significant enhancement of the different patterns for skin and muscle across the menstrual cycle. Comparison of the limb pain thresholds in men and women showed no differences between the men and nondysmenorrheic women, regardless of the time of the month, but did show a higher threshold for both groups compared with that for the dysmenorrheic women. Although these results highlight the complexity of the issue of differ- ences in pain by sex and time of the menstrual cycle, they point to several potentially important clinical issues. First, the results suggest that dys-
2 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH menorrhea might enhance the severity and cyclicity of other visceral con- ditions ("viscero-visceral interactions". This hypothesis is being tested in parallel studies with animals with endometriosis and ureteral stones and with humans with dysmenorrhea and ureteral stones. So far the results show significant interactions between the two conditions that have impli- cations for diagnosis and treatment in both females and males (Giamberardino, 2000; Giamberardino et al., 1999~. Second, a number of painful clinical disorders vary significantly with the menstrual cycle in some women but not others, such as certain types of headache, irritable bowel syndrome, interstitial cystitis, temporoman- dibular disorder, and fibromyalgia (Bradley and Alarcon, 2000; Fillingim and Maixner, 2000; Holroyd and Lipchik, 2000; Mayer et al., 1999; Naliboff et al., 2000~. It is possible that the women with cyclical pains also suffer from dysmenorrhea, a possibility that can be tested experimentally. If so, it is also possible that treatment directed at the dysmenorrhea might alle- viate those women's other pains, and this is also testable. Furthermore, an analysis of what factors reduce the pains during certain phases of the menstrual cycle might yield clues about the mechanism of the pain and treatments that could be applied to men with similar conditions. Third, what might be the basis for the surprising extreme abdominal sensitivity exhibited by the men, and what implications does this sensitiv- ity have for symptom reporting and clinical testing? Summary Overall, the results from research on sex differences in pain mecha- nisms and responses to treatment provide good examples of a construc- tive approach toward understanding the mechanisms of other sex differ- ences. This approach highlights the importance of considering how sex differences in genetic, hormonal, psychosocial, and stressful environmen- tal circumstances interact and evolve across the life span to give rise to an individual's ever-changeable "pain phenotype" at any particular time of her or his life (Berkley and Holdcroft, 1999; LeResche, 1999~. ANIMAL MODELS OF CEREBROVASCULAR AND CARDIOVASCULAR DISEASES Sex-specific responses to experimental traumatic or ischemic brain injury have been reported and are summarized in Table 4-3. The role of sex in behavioral outcomes after traumatic brain injury has also been studied. Clinical studies report improved outcomes for fe- male patients with head injuries compared with those for male patients with head injuries, as determined by the ability of patients with head injuries to return to their preinjury work levels (Groswasser et al., 1998~.
SEX AFFECTS BEHAVIOR AND PERCEPTION TABLE 4-3 Sex-Specific Responses to an Experimental Traumatic or Ischemic Cerebral Insult 113 Animal Model Results Gerbil 3-h carotid occlusion Permanent carotid occlusion Rat Permanent bilateral carotid occlusion 2-h MCAO Impact / acceleration closed-head injury Traumatic brain injury Progesterone after traumatic injury Entorhinal cortex injury Ovariectomy, global ischemic insult Post menopausal, MCAO M have more CAT hippocampal and cortical neuronal loss (Hall et al., 1991) M have more strokes (Berry et al., 1975) M have higher rates of mortality and larger numbers of brain lesions (Sadoshima et al., 1988) M have larger infarcts (Alkayed et al., 1998; Belayev et al., 1996; Zea-Longa et al., 1989) M have worse rates of survival (Roof and Hall, 2000a) M have more cerebral edema (Roof et al., 1993a) Equally beneficial effect on edema in both M and F (Roof et al., 1993a) M perform worse in maze test (Roof et al., 1993b) Ovariectomized F have greater neurological dysfunction than intact F (Wang et al., 1999) M and F similar in infarct size (Alkayed et al., 2000) Estradiol pretreatment of Increased survival and decreased ovariectomized F. temporary ischemic area in treated versus MCAO nontreated F (Simpkins et al., 1997) MCAO, estrogen treatment of M Prognosis improves in estrogen-treated MCAO, estrogen receptor antagonist Mice Unilateral carotid occlusion Unilateral carotid occlusion in SOD overexpressers M (Toting et al., 1998) Ischemia increases in F but not in M (Sawada et al., 2000) Larger lesion in M (Roof and Hall, 2000b) M protected by overexpression of SOD (Roof and Hall, 2000b) NOTE: MCAO, middle cerebral artery occlusion; M, male; F. female; SOD, superoxide dismutase. In studies with rats, sex-specific neuroprotection was lost when fe- male rats were ovariectomized, suggesting that circulating gonadal hor- mones are responsible for the sex differences (Simpkins et al., 1997~. Sev- eral reports demonstrate that estrogen and progesterone treatment has a neuroprotective effect. This area of research has recently been reviewed (Roof and Hall, 2000b). Results of experiments with rats suggest that
114 . EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH osteogenic neuroprotection is not sex specific and is not affected by tes- tosterone. The mechanisms by which female sex or by which estrogen or proges- terone attenuates brain damage are complex. Estrogen could preserve autoregulation or antioxidant activity, affect leukocyte adhesion, or up- regulate nitric oxide synthase. Estrogen modulates leukocyte adhesion in the cerebral circulation during resting conditions as well as after transient forebrain ischemia. Leukocyte adhesion and infiltration have been linked to the neuropathology in the brain; estrogen's neuroprotective effects may be due to modulation of this inflammatory pathway (Santizo et al., 2000~. In a model of the rate of progression of atherosclerosis in rabbits fed a high-cholesterol diet, the concentrations of lipids (total cholesterol, high- density lipoprotein cholesterol, and triglycerides) in serum were the same in males and females; however, the rate of progression of disease as deter- mined by histological examination of the thoracic aorta differed (greater in males than in females). Estrogen administration to oophorectomized rabbits fed high levels of cholesterol resulted in a reduced degree of ath- erosclerosis (Haarbo et al., 1991~. The inflammatory response that occurs during atherogenesis involves adhesion of monocytes to endothelial cells and migration across endothelial cells (Nathan et al., 1999~. Adhesion of monocytes to endothelial cells is slower in females. In addition, the level of VCAM-1 protein expression in aortas from oophorectomized rabbits fed an diet enriched in cholesterol was increased and was attenuated by the ischemia. These sex differences in VCAM-1 expression in this model suggest an estrogen-mediated anti-inflammatory mechanism. Transgenic (TNF1.6) mice with cardiac-specific overexpression of tu- mor necrosis factor alpha (TNF-oc) develop ventricular hypertrophy, car- diac dilatation, interstitial infiltrates, massive pleural effusion, and fibro- sis and die from congestive heart failure (Kubota et al., 1997~. The 6-month survival rate was significantly better in females. The marked sex differ- ences in survival cannot be the result of differences in the levels of expres- sion of TNF-oc since at both the transcript and the protein levels the levels of expression of TNF-oc was the same in males and females. Rather, male TNF1.6 mice had higher steady-state levels of messenger RNAs encoding both TNF-oc and -p receptors. The investigators (Kubota et al., 1997) dem- onstrated the physiological relevance of this increased level of expression of TNF receptors in male mice by looking at ceramide production, a TNF- dependent process, from myocardial tissue (male transgenic mice pro- duced more ceramide than females). These results suggest that enhanced survival in female mice in the presence of TNF overexpression may be attributable to sex-related differences in TNF receptor levels. The etiology of this differential regulation of TNF receptors remains unknown. In hu- man patients with heart failure, women live significantly longer than men (Becker et al., 1994; Greenland et al., 1991; Steingart et al., 1991~.
SEX AFFECTS BEHAVIOR AND PERCEPTION 115 Animal models provide an important research tool for the study of pathophysiological mechanisms of disease and therapeutic approaches. Male animals have predominantly been used in such animal models, how- ever, on the basis of the assumption that the results obtained from studies conducted with male animals could be extrapolated to female animals. Furthermore, the inclusion of female animals in preclinical studies in- creases the complexity of a study because of the female estrous cycle and the need to control for the associated hormonal fluctuations (Panetta and Srinivasan, 1998~. Thus, the roles of sex and sex hormones in mechanisms of disease outcome have not been routinely studied in animal models. It is not clear whether estrogen's effects are mediated via receptor-based or nongenomic mechanisms. However, continuing efforts to tease apart the mechanisms of sex-based differential vulnerability to traumatic and is- chemic brain injuries and cardiovascular diseases could lead to improved understanding of the pathophysiologies of these injuries and diseases and may suggest new mechanistic approaches to their treatment. FINDINGS AND RECOMMENDATIONS Findings Sex hormones do not act alone. No one factor is responsible for sex differences; rather, a number of genetic, hormonal, physiological, and experiential factors operating at different times during development re- sult in the phenotype called an individual. To better understand the influ- ences and roles of factors that may lead to sex differences, the committee makes the following recommendations. Recommendations RECOMMENDATION 4: Investigate natural variations. · Examine genetic variability, disorders of sex differentiation, re- productive status, and environmental influences to better understand human health. · Naturally occurring variations provide useful models that can be used to study the influences and origins of a range of factors that influ- ence sex differences. RECOMMENDATION 5: Expand research on sex differences in brain organization and function. New technologies make it possible to study sex-differential environ- mental and behavioral influences on brain organization and function and
116 EXPLORING THE BIOLOGICAL CONTRIBUTIONS TO HUMAN HEALTH to recognize modulators of brain organization and function. Explore in- novative ways to expand the availability of and reduce the cost of new technologies. Also see Recommendation 3 (Chapter 3) for a discussion of the need to mine cross-species information.