Arch Sex Behav (2009) 38:6–15 DOI 10.1007/s10508-007-9291-z ORIGINAL PAPER Prenatal Sex Hormones (Maternal and Amniotic Fluid) and Gender-related Play Behavior in 13-month-old Infants Cornelieke van de Beek Æ Stephanie H. M. van Goozen Æ Jan K. Buitelaar Æ Peggy T. Cohen-Kettenis Received: 28 June 2006 / Revised: 8 October 2007 / Accepted: 28 October 2007 / Published online: 13 December 2007 Ó Springer Science+Business Media, LLC 2007 Abstract Testosterone, estradiol, and progesterone levels were measured in the second trimester of pregnancy in maternal serum and amniotic fluid, and related to direct observations of gender-related play behavior in 63 male and 63 female offspring at age 13 months. During a structured play session, sex differences in toy preference were found: boys played more with masculine toys than girls (d = .53) and girls played more with feminine toys than boys (d = .35). Normal within-sex variation in prenatal testosterone and estradiol levels was not significantly related to preference for masculine or feminine toys. For progesterone, an unexpected significant positive relationship was found in boys between the level in amniotic fluid and masculine toy preference. The mechanism explaining this relationship is presently not clear, and the finding may be a spurious one. The results of this study may indicate that a hormonal basis for the development of sex-typed toy preferences may manifest itself only after toddlerhood. It may also be that C. van de Beek (&) Department of Child and Adolescent Psychiatry, VU University Medical Center, Postbus 303, 1115 ZG Duivendrecht, Amsterdam, The Netherlands e-mail: [email protected] S. H. M. van Goozen School of Psychology, Cardiff University, Cardiff, UK J. K. Buitelaar Department of Psychiatry, St. Radboud University Nijmegen Medical Center, Nijmegen, The Netherlands P. T. Cohen-Kettenis Department of Medical Psychology, VU University Medical Center, Amsterdam, The Netherlands 123 the effect size of this relationship is so small that it should be investigated with more sensitive measures or in larger populations. Keywords Sex hormones Maternal hormones Amniotic fluid Play behavior Gender role Sex differences Introduction Toy preference is one of the earliest manifestations of gender-related behavior. Girls have been found to prefer playing with toys such as dolls and household supplies, whereas boys prefer playing with toys such as vehicles and weapons (Hines, 2004). Several studies have demonstrated that sex differences in play behavior are present in the second year of life (e.g., Caldera, Huston, & O’Brien, 1989; Fagot, 1974; O’Brien & Huston, 1985). A study by Servin, Bohlin, and Berlin (1999) showed that girls and boys chose different toys as early as the age of 12 months. The fact that parents engage in some form of sex typing of their infant’s play behavior in the first year of life (Fagot, 1995; Fisher-Thompson, 1993; Lytton & Romney, 1991) suggests that socializing influences already may play a role. However, there are also indications that the child himself or herself may have certain innate preferences, i.e., the tendency to be focused on specific aspects of objects, like movement, color or form (Alexander, 2003), which may prime them to prefer specific toy categories. The findings of sex differences in toy preferences in non-human primates (Alexander & Hines, 2002), in which the influence of social learning and cognitive concepts or beliefs on toy preference can be considered nil, give additional reason to assume that there is a biological basis for the development of sex-typed toy preferences. Arch Sex Behav (2009) 38:6–15 Among the biological determinants of play behavior, sex hormones are likely candidates. Prenatal sex hormones can permanently influence postnatal human sex-typed behaviors by altering fetal brain development (for review and discussion, see Berenbaum, 1998; Cohen-Bendahan, Van de Beek, & Berenbaum, 2005; Collaer & Hines, 1995; Gooren & Kruijver, 2002; Kawata, 1995). Moreover, since there is also much within-sex variation in gender-typical behaviors, such as tomboy behavior (Bailey, Bechtold, & Berenbaum, 2002; Morgan, 1998), sex hormones are probably not only responsible for behavioral sex differences, but also for within-sex variations in gender-related behavior. Animal studies have shown that there are a few critical periods in development in which the brain regions that are responsible for the regulation of the sex-typed behavior are highly sensitive to the effects of sex hormones (Hines, 2004). In humans, the period between weeks 8 and 24 of gestation may be particularly important for sexual differentiation because this is the period when the male fetus shows a peak in serum testosterone (at its highest at about week 16) (Smail, Reyes, Winter, & Faiman, 1981). It has been clearly shown that prenatal androgens can have masculinizing and defeminizing effects on postnatal human behaviors (Hines, 2004). Very little is known about the function of ‘‘female-typical’’ hormones in sexual differentiation and their effects on human behavioral differentiation. Although rodent studies indicate that estradiol is very important in behavioral masculinization (Collaer & Hines, 1995; Goy & McEwen, 1980) and defeminization (Fitch & Denenberg, 1998), its role in humans is less obvious. Findings in both girls and boys, exposed to diethystilbestrol (DES) (a synthetic estrogen), suggest a general lack of influences on childhood gender typical play (Hines, 2004). Studies that have investigated the behavioral effects of prenatal progestagens mostly focused on exogenously administered synthetic progestins. The interpretations of the findings are complicated by the fact that different types of progestins were used. Prenatal exposure of girls to androgen-based synthetic progestins seemed to result in more male-typical behavior, such as more tomboyism and a stronger preference for male-typical toys, whereas exposure of girls to progesterone-based synthetic progestins produced fewer and opposite effects. Similar but less pronounced effects of these two types of exogenous progestins were found in boys (for review, see Collaer & Hines, 1995). Less is known about the effects of normal circulating progesterone levels during pregnancy, although an antiandrogenic effect on the fetus has been proposed (Ehrhardt, Meyer-Bahlburg, Feldman, & Ince, 1984). Currently, most evidence of a prenatal effect of androgens on human play behavior comes from studies in girls with congenital adrenal hyperplasia (CAH). CAH is a genetic condition that results in the production of high adrenal androgens beginning very early in gestation. Girls with CAH 7 show both masculinization and defeminization of behavior: they have a stronger preference for traditionally masculine toys and activities than unaffected control girls (relatives or matched comparisons) (Berenbaum & Hines, 1992; Pasterski et al., 2005). Indications for a dose–response relationship have also been found: more severely affected girls with CAH were more interested in masculine careers and toys than less affected girls (Servin, Nordenström, Larsson, & Bohlin, 2003). Furthermore, the timing of exposure on behavior has been demonstrated: prenatal androgen exposure, but not early or later postnatal exposure, was related to more maletypical play behavior (Berenbaum, Duck, & Bryk, 2000). In boys with CAH, elevated prenatal adrenal androgen exposure does not seem to result in more masculine play behavior (Berenbaum & Hines, 1992). There is also evidence that normal variations in androgen levels are systematically related to early sex differences and within-sex variations in gender-typical play behavior of children without clinical conditions. This evidence comes from studies using maternal blood to infer hormonal effects on the fetus. Udry, Morris, and Kovenock (1995) found a relationship between prenatal maternal sex hormone binding globulin (SHBG, which is responsible for binding testosterone and therefore inversely related to the amount of biologically active testosterone) and gender role behavior in 27- to 30year-old women. In correspondence with the idea that prenatal sexual differentiation of the human brain takes place during the testosterone peak, in the second rather than the first or third trimester, maternal SHBG levels in the second trimester predicted gender role behavior. Unexpectedly, no significant relationship was found between maternal testosterone and the women’s postnatal gender role behavior. In a very large sample, however, Hines, Golombok, Rust, Johnston, and Golding (2002) found higher maternal testosterone levels, as measured between weeks 5 and 36 of gestation, to be related to more masculine-typical preschool activity interests (motherreported) in 3.5-year-old girls. This relationship was not found in boys, which may indicate that the threshold for masculinization of behavior by androgens is exceeded in healthy males such that variability within the normal range would not be associated with individual differences in their gender role behavior. Both Hines et al. (2002) and Udry et al. (1995) used maternal blood as an indicator of prenatal androgen exposure of the fetus. Recently, a positive correlation was found between maternal and fetal blood testosterone (Gitau, Adams, Fisk, & Glover, 2005). This suggests that testosterone may cross the placenta, but it may also reflect a genetic relationship. However, maternal androgens do not appear to come from the fetus, as several studies have failed to find a difference in serum second trimester testosterone levels between women carrying a male and those carrying a female fetus (Forest, Ances, Tapper, & Migeon, 1971; Glass & Klein, 1980; Meulenberg & Hofman, 123 8 1991; Rivarola, Forest, & Migeon, 1968; Van de Beek, Thijssen, Cohen-Kettenis, Van Goozen, & Buitelaar, 2004). Another approach to investigate the organizational effects of prenatal sex hormones is used by amniotic fluid studies. Amniotic fluid can only be obtained from amniocentesis conducted for purposes of diagnosing fetal anomalies. Coincidentally, this medical intervention usually takes place around week 16 of gestation, a time that appears to correspond well to the male testosterone peak. Amniotic fluid seems to provide information about the sex steroid production by the fetus since several studies have found large sex differences in amniotic androgens (Dawood & Saxena, 1977; Finegan, Bartleman, & Wong, 1989; Judd, Robinson, Young, & Jones, 1976; Nagamani, McDonough, Ellegood, & Mahesh, 1979; Robinson, Judd, Young, Jones, & Yen, 1977; Rodeck, Gill, Rosenberg, & Collins, 1985; Van de Beek et al., 2004). However, currently there is no hard evidence of a direct relationship between amniotic testosterone and fetal serum testosterone. The mostly low and/or non-significant correlations between androgens measured in amniotic fluid and in maternal serum suggest that these measures reflect something different (Van de Beek et al., 2004). In addition, there is also another methodological difference between these two measures: amniotic fluid measures are only available for a specific population, which contains generally older mothers, while maternal blood measures can be relatively easily measured in a representative sample of the general population. Therefore, both measures have their advantages and disadvantages, and might provide convergent evidence of the relationship between prenatal hormones and postnatal behavior. Until now, there are two groups of investigators that have examined the relationship between early androgen exposure by investigating prenatal amniotic hormones and postnatal behavior. One research group mainly looked at measures of cognition and cerebral lateralization (Finegan, Niccols, & Sitarenois, 1992; Grimshaw, Bryden, & Finegan, 1995; Grimshaw, Sitarenios, & Finegan, 1995). The other group of investigators primarily focused on aspects of early social development (Knickmeyer, Baron-Cohen, Raggatt, & Taylor, 2005; Lutchmaya, Baron-Cohen, & Raggatt, 2001, 2002), but also included one aspect of gender role behavior in their design, namely game participation (Knickmeyer et al., 2005). They did not find amniotic testosterone to be related to individual differences on the Masculinity and Femininity scales of the Children’s Play Questionnaire at age 5/6 years, despite the fact that this measure showed large sex differences in the same sample. However, this parent report instrument might not have been sufficiently sensitive to within sex variability in behavior, as opposed to between-sex differences. Also, the very small sample size may have increased the chance of negative results. In the current study, we focused on toy preference observed in a laboratory situation. Rather than using parental reports, play behavior was scored by trained observers. Our first goal 123 Arch Sex Behav (2009) 38:6–15 was to replicate previously reported sex differences in masculine and feminine toy preference in normally developing 13-month-old infants. Furthermore, we investigated whether prenatal testosterone, estradiol, and progesterone levels, as assessed in amniotic fluid and maternal serum, were related to sex differences and within-sex variations in gender-related play behavior. Since the literature suggests most clearly an effect of prenatal androgens on play behavior, we hypothesized that higher levels of testosterone would be related to more masculine and less feminine play behavior. Previous studies on children with CAH indicate that these effects are clearly observed in girls (Berenbaum & Hines, 1992; Pasterski et al., 2005), but are absent in boys (Berenbaum & Hines, 1992). In the maternal blood study, Hines et al. (2002) reported the same pattern in a normative population. Therefore, we predicted that this relationship would be sex-specific, i.e., more clearly present in girls than boys. We did not have clear predictions with respect to the behavioral effects of variations in estradiol and progesterone. Method Participants All participants were enrolled in a prospective longitudinal project on the effects of prenatal sex hormones on gender development. Participants were recruited from a consecutive series of referrals, between January 1999 and August 2000, to the Department of Obstetrics at the University Medical Centre in Utrecht (UMCU) to undergo an amniocentesis because of prenatal diagnostic screening. All participants lived in relative close vicinity (30 km) of the UMCU. Amniotic fluid samples were provided by 153 participants with a normal healthy singleton pregnancy. Each participant gave informed consent to the procedure and the UMCU Medical Ethical Committee approved the study. For the majority of the women (96%), the reason for amniocentesis was a higher risk of age-related (36 years or older) genetic changes and their implications. The others had an amniocentesis because of their medical history (e.g., a previous child with Down syndrome or radiation cancer treatment) (3.3%) and one participant had a high result on the triple test (indicating an increased risk for Down syndrome) (0.7%). Seven participants reported that they did not become pregnant spontaneously; n = 4 had ovulation induction of which one also had donor insemination, n = 1 had artificial insemination, and n = 2 had in vitro fertilization (IVF). For 18 participants, data could not be collected at follow-up. Three children could not be tested because of illness or pregnancy of their mothers. Two children were excluded when their mother’s command of the Dutch language appeared to be insufficient to complete questionnaires. One child was excluded Arch Sex Behav (2009) 38:6–15 because of extreme prematurity (gestational period less than 35 weeks) at birth. Furthermore, we lost participants (n = 12) because they moved abroad (n = 1) or were no longer interested to participate in the study (n = 11). In total, 135 children were seen at the age of 13 months. Two girls were so timid that behavioral observations were not possible. The observations of three other children were not reliable because of illness or fatigue and the play behavior of four children could not be scored because of technical problems. In the end, the data of 126 children (63 boys and 63 girls) were used for further analyses. The mean age of the children was 56.4 weeks (SD = 1.2, range, 53.6–60.6). The mean age of the mothers at amniocentesis was 37.5 years (SD = 2.1, range, 28–45). The mean parental education score was 9.8 (SD = 3.1, range, 2–14). With respect to the presence of older siblings, there were: no older brother (n = 69), one older brother (n = 43), two or more older brothers (n = 10), no older sister (n = 86), one older sister (n = 31), and two or more older sisters (n = 5). None of these variables differed significantly between boys and girls. 9 immediately following the amniocentesis. The hormonal analyses were conducted as described above except for testosterone. In maternal serum, free testosterone, the biologically active part of this sex hormone, was calculated using existing procedures (Dunn, Nisula, & Rodbard, 1981). These equations can be used to describe the relation between the bound and free fraction of testosterone with other steroids and with several binding proteins. We made calculations for a system of two steroids (estradiol and testosterone) and two binding proteins (albumin and SHBG). The value of albumin in maternal serum was fixed at 43 g/l for measurements in the second trimester and 35 g/l in the third trimester (Ashwood, 1994). Sex hormone binding globulin (SHBG) was measured by immunometric assay (SHBG Immulite, Euro/DPC, Gwynedd, UK); interassay variation was 6.9% at a concentration of 93 nmol/l. One girl with an extremely high (because of skewed distributions, a criterion of >4 SD was used) amnion testosterone level (2.50 nmol/l) and whose mother had an extremely high progesterone level (2,860.0 nmol/l) and one girl whose mother had an extremely high plasma testosterone level (42.40 nmol/l) were not included in the data analyses. Measures Sex Hormones in Amniotic Fluid The amniotic fluid samples were collected between week 15.3 and 18.0 of pregnancy (M = 16.3, SD = .46). The length of gestation was determined by the last menstrual period and/or ultrasonic measurement of crown-rump length (CRL) (Daya, 1993). Because we had to adjust to the time schedule of the clinic, it was not possible to standardize the hormone sampling time completely; however, all samples were taken between 8:00 am and 13:00 pm. The material was stored at -30°C until assayed. The lab employees who analyzed the samples were masked to the behavioral data. Testosterone was determined by radioimmunoassay (RIA) after extraction with diethyl ether (Van Landeghem et al., 1981). Interassay coefficient of variation was 8.8% at a testosterone level of 0.75 nmol/l and 9.4% at a level of 2.55 nmol/l. The measured T level in amniotic fluid includes exclusively free T (FT) and therefore needed no conversion. Estradiol (E2) (total) and progesterone were determined using the Axsym of Abbott (Abbott Park, IL). Sera were diluted with phosphate buffered saline (0.01 M at pH 7.0, containing 0.5% bovine serum albumin) (estradiol: second trimester serum 209; progesterone: 109). Interassay variation was 5.1% at 1,060 pmol/l for estradiol and 5.0% at 18 nmol/l for progesterone. Sex Hormones in Maternal Serum We were able to collect serum in 115 mothers-to-be (57 boys, 58 girls). The maternal serum samples were collected Play Observations In a structured toy play session, nine different toys that previously had been classified by parents and non-parents as masculine, feminine, or neutral were used (Campenni, 1999; Servin et al., 1999). Each toy category was equally presented. Neutral toys were a plastic friction dog, a wooden puzzle, and a stacking pole with rings. Masculine toys were a trailer with four cars, a garbage truck, and a set of three plastic pieces of equipment. Feminine toys were a teapot with a cup, a soft doll in a cradle with a blanket, and a doll with beauty set (brush, comb, and mirror). The toys were arranged in a standard order in a semicircle (from left to right: tea set, dog, trailer, doll in cradle, rings, truck, doll with beauty set, puzzle, and equipment). The mother was asked to place the child in the center of the semicircle, at the same distance from all toys, and then take a seat in the chair just outside the semicircle (1.5 m). She was instructed to let the child play on his or her own, but was allowed to give neutral verbal reactions if the child specifically asked her attention. When giving verbal reactions, she was asked to avoid naming the objects and guiding the child in its actions. The child was videotaped for 7 min, starting by the first touch of a toy. If play behavior was not present for longer than 30 s (e.g., if the child started to cry and sought comfort with his/her mother), this time was added to the original 7 min. In this way, every child had 7 min real playing time. The play session took place at the beginning of a more extensive testing session, after a short (5– 10 min) talk with the mother. The entire visit took approximately 1 h. 123 10 For each toy, we recorded the number of seconds the child played with that particular toy. Each ‘‘play action’’ was scored from the moment that physical contact started until the physical contact stopped, unless there was still obvious ‘‘involvement’’ (e.g., pushing a car and crawling behind it). ‘‘Involvement’’ was defined as looking at, pointing at, and moving behind the object. If the child played with several toys simultaneously, physical contact time was scored for each toy separately. Thus, the total playing time of a child (all time spent with the different objects in sum) sometimes was longer than the 7 min of play observation. For each toy category, the total amount of time was counted and the percentage of the total playing time was calculated, resulting in the variables ‘‘% masculine play,’’ ‘‘% feminine play,’’ and ‘‘% neutral play.’’ Two participants with a deviant total playing time (>3 SD) were excluded; one boy showed an extremely high (730 s) and another boy an extremely low (36 s) total playing time. The videotapes were scored by trained observers who were masked to the hormonal values. The inter-rater reliability was high. Kendall’s tau correlations ranged from .95 to 1.00 for masculine play, from .94 to .99 for feminine play, and from .99 to 1.00 for neutral play. Developmental Assessment A home visit took place within a week after the play session. During this visit, the Bayley Scales of Infant Development II (BSID II) (Van der Meulen, Ruiter, Lutje Spelberg, & Smrkovsky, 2002) was administered to assess motor and mental development. Background Variables Information was collected on several social and demographic variables by means of a short questionnaire. The following variables were used in the analyses: age of the mother at the time of amniocentesis, parental educational level, and number of older brothers and sisters. For each parent, the educational level was scored on a 7-point scale, with a score of 1 indicating ‘‘no formal qualifications’’ to 7 representing ‘‘a university degree.’’ The scores of both parents were combined. Statistical Analyses To standardize not normally distributed variables, a transformation was used. A square-root transformation was carried out on the variables ‘‘% masculine play’’ and ‘‘% feminine play.’’ The distribution of all prenatal sex hormones was standardized by means of a logarithmic transformation. Student’s t tests were employed in statistical group comparisons 123 Arch Sex Behav (2009) 38:6–15 (one-tailed when comparing boys and girls on masculine and feminine play, two-tailed in the other comparisons) and Pearson correlation coefficients were calculated to map the relationships among variables. Backward stepwise linear regression analyses (removal criterion p =.10) were used to examine the variance associated with relevant independent and background variables that showed to be related (at p < .10) to gender-related play. In the analysis including both sexes, the dummy variable ‘‘sex’’ was coded 0 for boys and 1 for girls. Because sex and amniotic testosterone were so highly correlated (r = -.78, df = 121, p < .001), it was not possible to disentangle their separate relationships with play behavior by means of regression analyses in the whole population (multicollinearity). Therefore, for this hormone, we could only analyze the data for boys and girls separately. Results Prenatal Sex Hormones There were no significant sex differences in maternal serum (see Table 1). In amniotic fluid, male fetuses had significantly higher testosterone levels, t(120) = 13.20, p < .001 (d = 2.39), and female fetuses had significantly higher estradiol levels, t(118) = -2.39, p < .05 (d = .43). No significant sex difference was found with respect to amniotic progesterone levels. No significant relationship was found between maternal age and the hormone levels, and the same applied for gestational age at sampling and maternal hormones. In male fetuses, but not in female fetuses, there was a gestational age effect with respect to the amniotic data. With the progression of pregnancy, a significant decrease was found in testosterone (r = -.33, df = 121, p < .05), progesterone (r = -.36, df = 121, p < .05), and estradiol (r = -.22, df = 119, p < .10). To control for these effects, we used in further analyses the standardized residual of the calculated regression lines instead of the actual hormone levels. Play Behavior The total sample had a mean total playing time of 415 s (SD = 87, range, 181–692). There was no significant sex difference in the total time that the toys were handled. Figure 1 shows the percentages of the total time spent with the different toy categories for boys and girls. Boys (M = 40.96, SEM = 3.59) spent a significantly higher percentage of time playing with the masculine toys than girls (M = 27.10, SEM = 2.87), t(120) = 2.92, p = .002 (d = .53). The percentage of time with feminine toys was significantly higher for girls (M = 45.74, SEM = 3.43) than for boys (M = 36.62, Arch Sex Behav (2009) 38:6–15 Table 1 Sex differences in prenatal sex hormone levels 11 Hormones (nmol/l) Male M Female SD n M p SD n Maternal serum Testosterone Estradiol Progesterone 6.78 3.91 56 7.15 3.81 56 ns 21.94 8.58 56 21.95 8.53 56 ns 176.36 49.90 56 173.71 54.06 56 ns 1.41 .36 61 .68 .28 61 <.001 .88 .26 61 1.01 .36 59 <.05 275.74 89.89 61 286.48 111.06 61 ns Amniotic fluid Testosterone Estradiol Progesterone SEM = 3.56), t(120) = -1.92, p = .029 (d = .35). Boys and girls did not differ significantly with respect to neutral play. Gender-related Play and Non-hormonal Variables In boys, no clear relationships were found between the age of mother, parental education, the number of older sisters, and mental and motor development, on the one hand, and masculine and feminine play behavior, on the other, except for a marginally significant positive relationship between parental educational level and masculine play behavior. For feminine play, a marginally significant relationship was found for boys, but in the opposite direction. In addition, the percentage of masculine play behavior tended to be lower, and feminine play higher, with an increase in the number of older brothers (see Table 2). In girls, no significant relationships were found between any of the background variables and gender-related play, except for a significant negative relation between percentage of feminine play and number of older sisters (see Table 2). 60 boys girls % of total playing time 50 * 40 ** 30 20 10 0 Prenatal Hormones and Gender-related Play No significant relations were found between maternal sex hormones levels and masculine or feminine play behavior in either boys or girls (see Table 2). With regard to the sex hormones measured in amniotic fluid, a positive correlation between progesterone and time spent with masculine toys (r = .30, df = 121, p < .05) was observed in boys. No other significant relationships with masculine or feminine play were found in either boys or girls. Regression Analysis Within the Sexes Because there were no significant relationships between prenatal hormones and feminine play in boys or girls, and because there was no relationship between prenatal hormone levels and masculine play behavior in girls, we further explored the relationship between hormones and masculine behavior in boys only. First, the variables that were related to masculine play (at a p < .10 level) were entered in the model simultaneously. These variables were progesterone in amniotic fluid, parental educational level, and the number of older brothers. Although amniotic testosterone was not significantly correlated with masculine play, we included it in the regression, because amniotic testosterone and progesterone were positively related (r = .36, df = 121, p <.01). In the final model (see Table 3), only progesterone in amniotic fluid significantly and positively predicted masculine play. With respect to the non-hormonal variables, masculine play was predicted by the number of older brothers (less masculine play with more older brothers) and parental educational level (less masculine play with lower educational level). Regression Analysis Including both Sexes feminine masculine neutral Fig. 1 Play with gender-specific toy categories ** p < .01, * p < .05 Using a similar procedure, we also entered sex, and the interactions between sex and amniotic progesterone and testosterone, respectively, in the model (see Table 4). Again, progesterone 123 12 Arch Sex Behav (2009) 38:6–15 Table 2 Correlations between independent variables and gender-specific play categories Masculine play Feminine play Boys Girls r Boys Girls n r n r n r n Hormones maternal serum Testosterone .09 56 -.01 56 -.17 56 .09 56 Estradiol Progesterone -.13 .12 56 56 .06 .22 56 56 .05 -.07 56 56 -.05 -.05 56 56 .11 61 .00 61 .01 61 .03 61 -.05 61 -.04 59 .04 61 .14 59 .30** 61 .07 61 -.19 61 .00 61 Mother’s age .10 61 .03 61 -.11 61 .17 61 Parental education .22* 61 .18 61 -.24* 61 -.20 61 Infant’s BSID II mental score .02 60 .01 60 .03 60 -.01 60 Infant’s BSID II motor score -.14 60 .05 60 .17 60 .11 60 No. of older brothers -.23* 61 -.15 61 .23* 61 -.12 61 No. of older sisters -.02 61 .21 61 61 -.31** 61 Hormones amniotic fluid Testosterone Estradiol Progesterone Background variables -.04 ** p < .05, * p < .10 Table 3 Regression model for predicting masculine play in 13-month-old boys (n = 61) Predictor B SE B Standardized b Partial r t p Progesterone in amniotic fluid .70 .28 .30 .32 2.53 .014 Parental education .19 .10 .24 .26 2.00 .050 -.71 .36 -.23 -.25 -1.96 .054 Older brother(s) 2 R = .194, F(4, 56) = 4.59, p = .006 Table 4 Regression model for predicting masculine play in all participants (n = 122) at age 13 months Predictor B SE B Standardized b Partial r t p Sex -1.19 .39 -.26 -.27 -3.05 .003 Progesterone in amniotic fluid .35 .20 .15 .16 1.75 .083 Parental education .14 .06 .18 .20 2.15 .034 -.51 .27 -.16 -.17 -1.91 .058 Older brother(s) 2 R = .395, F(4, 117) = 5.40, p = .001 positively predicted masculine play, but now at a trend level. Sex of child strongly predicted masculine play, as did parental educational level. Finally, at a trend level, an inverse relation was found between the number of older brothers and masculine play. Discussion In the present study, we observed significantly higher amniotic testosterone levels in male pregnancies, significantly higher 123 amniotic estradiol levels in female pregnancies, and no sex difference in maternal plasma sex hormone levels. This is in line with previous findings (Dawood & Saxena, 1977; Finegan et al., 1989; Glass & Klein, 1980; Hines et al., 2002; Judd et al., 1976; Nagamani et al., 1979; Robinson et al., 1977; Rodeck et al., 1985). Furthermore, sex differences in toy preference were clearly present at the age of 13 months. Boys spent significantly more time playing with masculine toys than girls and girls played significantly more with feminine toys than boys. For both sexes, no significant relations were found between amniotic testosterone or other hormones and masculine or Arch Sex Behav (2009) 38:6–15 feminine play behavior, except for an unexpected positive relationship between amniotic progesterone and masculine play. This effect was clearly present in boys and was present at a trend level in the total sample (girls and boys together). We can think of several reasons for the absence of a significant relationship between amniotic testosterone and play behavior. In our study, there are several methodological gaps that may have led to insufficient experimental power. First, it may be that one single sample of hormones does not represent actual individual differences in fetal hormone exposure. Little is known about circadian rhythms of sex hormones during pregnancy, but it has been reported that several hormones show fluctuations within a day and across days, even in fetuses (Séron-Ferré, Ducsay, & Valenzuela, 1993; Walsh, Ducsay, & Novy, 1984). Furthermore, the relationship between testosterone measured in amniotic fluid and fetal blood is not established, so it is still uncertain what the amniotic level really represents. Second, it may be that, at the age of 13 months, sex differences in play behavior were too small to study hormone–play relationships in the currently used sample size. At 13 months of age, we found clear sex differences in masculine and feminine toy preference, but the effect sizes (Cohen, 1992) were small (feminine play: d = .35) to moderate (masculine play: d = .53). This is much smaller than what is usually reported in studies in somewhat older pre-school children (e.g., d = 1.92 for masculine toys and d = 1.23 for feminine toys in 3-yearolds (Servin et al., 1999). However, other hormone studies did find a relationship between amniotic testosterone and behaviors that show sex differences of moderate effect size, i.e., eyecontact (d = .53) (Lutchmaya et al., 2002) and vocabulary size (d = .67) (Lutchmaya et al., 2001). Therefore, we expected to find at least significant results for masculine play. Third, in a study on hormone–behavior relationships, an instrument is needed that captures behavioral differences between boys and girls, but that is also sensitive to within sex differences. Despite the fact that we did find sex differences, it may be that our method was not sensitive enough to measure within sex variability. Although the mothers in our study were instructed not to interfere, it may be that their presence has caused some error. The positive and unexpected relationship between amniotic progesterone and masculine play behavior in boys may have reflected a Type I error, considering the multiple comparisons. Yet, in the Stanford Longitudinal Study (Jacklin, Maccoby, & Doering, 1983; Jacklin, Maccoby, Doering, & King, 1984; Marcus, Maccoby, Jacklin, & Doering, 1985), which used umbilical cord blood assessments, some significant relationships were found between progesterone and traits that can be labeled masculine. However, it is questionable whether umbilical cord blood levels are a good index of organizational effects of sex hormones (Cohen-Bendahan et al., 2005) and little is known about the relationship between 13 progesterone levels from cord blood at birth and progesterone levels in second trimester amniotic fluid. Finally, some attention should be paid to the few potentially relevant non-hormonal variables we incorporated in our study to control for their possible effects. Although the design of the study does not allow for definite conclusions, the results indicate that, already at 13 months of age, variables such as parental education and the number of older brothers are related to gender-related play behavior. In boys, there were indications for a tendency of masculine play to be less frequent, and feminine play to be more frequent, when the child had more older brothers. Although having one older brother is associated with more masculine gender role behavior (Rust et al., 2000), studies that included families with more than one sibling suggest that having two or more brothers is related to a decrease in masculine behavior (Leventhal, 1970; SuttonSmith & Rosenberg, 1970). Rather than considering sibship as a social factor some researchers consider it as a biological factor. Blanchard (2001) proposed that each succeeding male fetus leads to the production of antibodies that can pass through the placental barrier, enter the fetal brain, and impede the sexual differentiation of the brain in the male-typical direction. Although both biological and social factors play a role in gender role development (Ruble, Martin, & Berenbaum, 2006), there is no real consensus as to if and how the different factors interact and whether there is a specific point in time at which social or cultural factors might, in fact, overrule potential biological influences. Parental behavior may strengthen biologically based differences, augmenting small differences present at 13 months. However, parents may also modify individual preferences, and overrule biological predispositions, for example in the case of very feminine boys. It is important to follow up the participants of this study and establish whether the studied relationships between prenatal hormones and genderrelated play behavior increase or change over time or only appear later in development. Acknowledgments We wish to thank Lieve Christiaens, Gerard Visser, and the other members of the Department of Obstetrics for their cooperation and Inge Maitimu and the employees of the Laboratory of Endocrinology for performing the hormonal analyses. We wish to thank the research assistants and students who helped to collect the behavioral data. And above all, we are grateful to all mothers and children who participated in the study. This research was funded by the Netherlands Organization for Scientific Research NWO Grant No. 575-25-010. References Alexander, G. M. (2003) An evolutionary perspective of sex-typed toy preferences: Pink, blue, and the brain. Archives of Sexual Behavior, 32, 7–14. Alexander, G. M., & Hines, M. (2002). Sex differences in response to children’s toys in nonhuman primates (Cercopithecus aethiops sabaeus). Evolution and Human Behavior, 23, 467–479. 123 14 Ashwood, E. R. (1994). Clinical chemistry of pregnancy. In N.W. Tietz (Ed.), Tietz textbook of clinical chemistry (p. 2112). Philadelphia: Saunders. Bailey, J. M., Bechtold, K. T., & Berenbaum, S. A. (2002). Who are tomboys and why should we study them? Archives of Sexual Behavior, 31, 333–341. Berenbaum, S. A. (1998). How hormones affect behavioral and neural development: Introduction to the special issue on ‘‘Gonadal hormones and sex differences in behavior’’. Developmental Neuropsychology, 14, 175–196. Berenbaum, S. A., Duck, S. C., & Bryk, K. (2000). Behavioral effects of prenatal versus postnatal androgen excess in children with 21hydroxylase-deficient congenital adrenal hyperplasia. Journal of Clinical Endocrinology and Metabolism, 85, 727–733. Berenbaum, S. A., & Hines, M. (1992). Early androgens are related to childhood sex-typed toy preferences. Psychological Science, 3, 203–206. Blanchard, R. (2001). Fraternal birth order and the maternal immune hypothesis of male homosexuality. Hormones and Behavior, 40, 105–114. Caldera, Y. M., Huston, A. C., & O’Brien, M. (1989). Social interactions and play patterns of parents and toddlers with feminine, masculine, and neutral toys. Child Development, 60, 70–76. Campenni, C. E. (1999). Gender stereotyping of children’s toys: A comparison of parents and nonparents. Sex Roles, 40, 121–138. Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159. Cohen-Bendahan, C. C. C., Van de Beek, C., & Berenbaum, S. A. (2005). Prenatal sex hormone effects on child and adult sex-typed behavior: Methods and findings. Neuroscience and Biobehavioral Reviews, 29, 353–384. Collaer, M. L., & Hines, M. (1995). Human behavioral sex differences: A role for gonadal hormones during early development? Psychological Bulletin, 118, 55–107. Dawood, M. Y., & Saxena, B. B. (1977). Testosterone and dihydrotestosterone in maternal and cord blood and in amniotic fluid. American Journal of Obstetrics and Gynecology, 129, 37–42. Daya, S. (1993). Accuracy of gestational age estimation by means of fetal crown-rump length measurement. American Journal of Obstetrics and Gynecology, 168, 903–908. Dunn, J. F., Nisula, B. C., & Rodbard, D. (1981). Transport of steroid hormones: Binding of 21 endogenous steroids to both testosteronebinding globulin and corticosteroid-binding globulin in human plasma. Journal of Clinical Endocrinology and Metabolism, 53, 58–68. Ehrhardt, A. A., Meyer-Bahlburg, H. F. L., Feldman, J. F., & Ince, S. E. (1984). Sex-dimorphic behavior in childhood subsequent to prenatal exposure to exogenous progestogens and estrogens. Archives of Sexual Behavior, 13, 457–477. Fagot, B. I. (1974). Sex differences in toddlers’ behavior and parental reaction. Developmental Psychology, 10, 554–558. Fagot, B. I. (1995). Parenting boys and girls. In M. H. Bornstein (Ed.), Handbook of parenting: Vol. 1. Children and parenting (pp. 163– 183). Mahwah, NJ: Lawrence Erlbaum Associates. Finegan, J. A., Bartleman, B., & Wong, P. Y. (1989). A window for the study of prenatal sex hormone influences on postnatal development. Journal of Genetic Psychology, 150, 101–112. Finegan, J. A. K., Niccols, G. A., & Sitarenois, G. (1992). Relations between testosterone levels and cognitive abilities at 4 years. Developmental Psychology, 28, 1075–1089. Fisher-Thompson, D. (1993). Adult toy purchases for children: Factors affecting sex-typed toy selection. Journal of Applied Developmental Psychology, 14, 385–406. Fitch, R. H., & Denenberg, V. H. (1998). A role for ovarian hormones in sexual differentiation of the brain. Behavioral and Brain Sciences, 21, 311–352. 123 Arch Sex Behav (2009) 38:6–15 Forest, M. G., Ances, I. G., Tapper, A. J., & Migeon, C. J. (1971). Percentage binding of testosterone, androstenedione and dehydroisoandrosterone in plasma at the time of delivery. Journal of Clinical Endocrinology, 32, 417–425. Gitau, R., Adams, D., Fisk, N. M., & Glover, V. (2005). Fetal plasma testosterone correlates positively with cortisol. Archives of Disease in Childhood, 90, 166–169. Glass, A. R., & Klein, T. (1980). Changes in maternal serum total and free androgen levels in early pregnancy: Lack of correlation with fetal sex. American Journal of Obstetrics and Gynecology, 140, 656–660. Gooren, L. J. G., & Kruijver, F. P. M. (2002). Androgens and male behavior. Molecular and Cellular Endocrinology, 198, 31–40. Goy, R. W., & McEwen, B. S. (1980). Sexual differentiation of the brain. Cambridge, MA: MIT Press. Grimshaw, G. M., Bryden, M. P., & Finegan, J. K. (1995). Relations between prenatal testosterone and cerebral lateralization in children. Neuropsychology, 9, 68–79. Grimshaw, G. M., Sitarenios, G., & Finegan, J. A. K. (1995). Mental rotation at 7 years: Relations with prenatal testosterone levels and spatial play experience. Brain and Cognition, 29, 85–100. Hines, M. (2004). Brain gender. Oxford: Oxford University Press. Hines, M., Golombok, S., Rust, J., Johnston, K. J., & Golding, J. (2002). Testosterone during pregnancy and gender role behavior of preschool children: A longitudinal, population study. Child Development, 73, 1678–1687. Jacklin, C. N., Maccoby, E. E., & Doering, C. H. (1983). Neonatal sexsteroid hormones and timidity in 6–18-month-old boys and girls. Developmental Psychobiology, 16, 163–168. Jacklin, C. N., Maccoby, E. E., Doering, C. H., & King, D. R. (1984). Neonatal sex-steroid hormones and muscular strength of boys and girls in the first three years. Developmental Psychobiology, 17, 301–310. Judd, H. L., Robinson, J. D., Young, P. E., & Jones, O. W. (1976). Amniotic fluid testosterone levels in midpregnancy. Obstetrics and Gynecology, 48, 690–692. Kawata, M. (1995). Roles of steroid hormones and their receptors in structural organization in the nervous system. Neuroscience Research, 24, 1–46. Knickmeyer, R., Baron-Cohen, S., Raggatt, P., & Taylor, K. (2005). Foetal testosterone, social relationships, and restricted interests in children. Journal of Child Psychology and Psychiatry, 46, 198–210. Knickmeyer, R. C., Wheelwright, S., Taylor, K., Raggatt, P., Hackett, G., & Baron-Cohen, S. (2005). Gender-typed play and amniotic testosterone. Developmental Psychology, 41, 517–528. Leventhal, G. S. (1970). Influence of brothers and sisters on sex-role behavior. Journal of Personality and Social Psychology, 16, 452–465. Lutchmaya, S., Baron-Cohen, S., & Raggatt, P. (2001). Fetal testosterone and vocabulary size in 18- and 24-month-old infants. Infant Behavior and Development, 24, 418–424. Lutchmaya, S., Baron-Cohen, S., & Raggatt, P. (2002). Fetal testosterone and eye contact in 12-month-old human infants. Infant Behavior and Development, 25, 327–335. Lytton, H., & Romney, D. M. (1991). Parents’ differential socialization of boys and girls: A meta-analysis. Psychological Bulletin, 109, 267–296. Marcus, J., Maccoby, E. E., Jacklin, C. N., & Doering, C. H. (1985). Individual differences in mood in early childhood: Their relation to gender and neonatal sex steroids. Developmental Psychobiology, 18, 327–340. Meulenberg, P. M. M., & Hofman, J. A. (1991). Maternal testosterone and fetal sex. Journal of Steroid Biochemistry and Molecular Biology, 39, 51–54. Morgan, B. L. (1998). A three generational study of tomboy behavior. Sex Roles, 39, 787–800. Nagamani, M., McDonough, P. G., Ellegood, J. O., & Mahesh, V. B. (1979). Maternal and amniotic fluid steroids throughout human Arch Sex Behav (2009) 38:6–15 pregnancy. American Journal of Obstetrics and Gynecology, 134, 674–680. O’Brien, M., & Huston, A. (1985). Development of sex-typed play behavior in toddlers. Developmental Psychology, 21, 866–871. Pasterski, V. L., Geffner, M. E., Brain, C., Hindmarsh, P., Brook, C., & Hines, M. (2005). Prenatal hormones and postnatal socialization by parents as determinants of male-typical toy play in girls with congenital adrenal hyperplasia. Child Development, 76, 264–278. Rivarola, M. A., Forest, M. G., & Migeon, C. J. (1968). Testosterone, androstenedione and dihydroepiandrosterone in plasma during pregnancy and at delivery: Concentration and protein binding. Journal of Clinical Endocrinology, 28, 34–40. Robinson, J. D., Judd, H. L., Young, P. E., Jones, O. W., & Yen, S. S. C. (1977). Amniotic fluid androgens and estrogens in midgestation. Journal of Clinical Endocrinology and Metabolism, 45, 755–761. Rodeck, C. H., Gill, D., Rosenberg, D. A., & Collins, W. P. (1985). Testosterone levels in midtrimester maternal and fetal plasma and amniotic fluid. Prenatal Diagnostics, 5, 175–181. Ruble, D. N., Martin, C. L., & Berenbaum, S. A. (2006). Gender development. In W. Damon & R. M. Lerner (Series Eds.) and N. Eisenberg (Vol. Ed.), Handbook of child psychology (Sixth ed.). Vol. 3: Social, emotional, and personality development (pp. 858– 932). New York: Wiley. Rust, J., Golombok, S., Hines, M., Johnston, K., Golding, J., & The ALSPAC Study Team. (2000). The role of brothers and sisters in the gender development of preschool children. Journal of Experimental Child Psychology, 77, 292–303. Séron-Ferré, M., Ducsay, C. A., & Valenzuela, G. J. (1993). Circadian rhythms during pregnancy. Endocrine Reviews, 14, 594–609. Servin, A., Bohlin, G., & Berlin, L. (1999). Sex differences in 1-, 3-, and 5-year-olds’ toy choice in a structured play-session. Scandinavian Journal of Psychology, 40, 43–48. 15 Servin, A., Nordenström, A., Larsson, A., & Bohlin, G. (2003). Prenatal androgens and gender-typed behavior: A study of girls with mild and severe forms of congenital adrenal hyperplasia. Developmental Psychology, 39, 440–450. Smail, P. J., Reyes, F. I., Winter, J. S. D., & Faiman, C. (1981). The fetal hormone environment and its effect on the morphogenesis of the genital system. In S. J. Kogan & E. S. E. Hafez (Eds.), Pediatric andrology (pp. 9–19). The Hague, Netherlands: Martinus Nijhoff. Sutton-Smith, B., & Rosenberg, B. G. (1970). The sibling. New York: Holt, Rinehart & Winston. Udry, J. R., Morris, N. M., & Kovenock, J. (1995). Androgen effects on women’s gendered behavior. Journal of Biosocial Science, 27, 359–368. Van de Beek, C., Thijssen, J. H. H., Cohen-Kettenis, P. T., Van Goozen, H. M., & Buitelaar, J. K. (2004). Relationships between sex hormones assessed in amniotic fluid, and maternal and umbilical cord blood: What is the best source of information to investigate the effects of fetal hormonal exposure? Hormones and Behavior, 46, 663–669. Van der Meulen, B. F., Ruiter, S. A. J., Lutje Spelberg, H. C., & Smrkovsky, M. (2002). Bayley Scales of Infant Development (BSID) II. Dutch version. Lisse, Netherlands: Swets Test Publishers. Van Landeghem, A. A. J., Poortman, J., Deshpande, N., Di Martino, L., Tarquini, A., & Thijssen, J. H. H. (1981). Plasma concentration gradient of steroid hormones across human mammary tumors in vivo. Journal of Steroid Biochemistry, 14, 741–747. Walsh, S. W., Ducsay, C. A., & Novy, M. J. (1984). Circadian hormonal interactions among the mother, fetus, and amniotic fluid. American Journal of Obstetrics and Gynecology, 150, 745–753. 123
© Copyright 2026 Paperzz