EARLY ANDROGENS ARE RELATED TO CHILDHOODSEX-TYPED TOY PREFERENCES

Sheri A. Berenbaum I and Melissa Hines2I Department of Psychology and Department of Psychiatry and Behllvioral Sciences. University of Health

Sciences/The Chicago Medical School. and 2Department of Psychiatry and Biobehllvioral Sciences. UCLA School ofMedicine. University of California. Los Angeles

Abstract-Girls with cong~nital adr~nalhyp~rplasia (CAR) who w~r~ ~xpos~d tohigh '~v~ls 01 andro,~n in th~ pr~nataland ~arly postnatal p~riods show~d in-cr~as~dp/Qy with boys' toys and r~duc~dplay with girls' toys compar~d with th~irun~xpos~d I~ma/~ r~/atives at a,~s 3 to8. Boys with CAR did not differ fromth~ir ma/~ r~/atives in play with boys' orgirls' toys. Th~se r~sults sugg~st thatearly hormon~ exposur~ in/~males has amasculinizing ~ff~ct on s~x-typ~d toypr~/erences.

the patient or control status of subjects.We hypothesized that CAH girls wouldshow 8feater preference for boys' toysthan their unaffected female relatives,and reduced preference for girls' toys.We did not predict effects for CAH boysbecause androgen treatment has incon.sistent effects in male experimental ani-mals (Baum &. Schretlen, 197.5; Dia.mond. Llacuna, & Wong, 1973).

METHOD

Sex differences in children"s toy pref-erences have been demonstrated repeat-edly" Boys prefer constrUction and trans-portation toys. whereas girls preferdolls. don fumishin,s. and kitchen sup-plies (Connor & Serbin. 1977; Liss.1981). These preferences appear to bepartially leamed. through modelina andreinforcement. For example. children atvarious stages of development emulatethe behavior of same-sex models in pref-erence to opposite-sex ones (Bussey &Bandura. 1984; Huston. 1983). Wepresent evidence that these sex-typedtoy preferences are also related to pre-natal or neonatal hormones (androlens).

Gonadal hormones playa major rolein the development of sex differences inbehavior and the brain in a variety ofspecies. including rodents, songbirds.and primates (Arnold &:. Gorski. 1984;Beatty, 1979; Goy & McEwen. 1980;Mac Lusky & Naftolin, 1981). A uniqueopponunity to study hormonal influ-ences on human sex-typed behavior isprovided by the genetic disorder congen-ital adrenal hyperplasia (CAH). Becauseof an enzymatic defect. individuals with

Illness Characteristics

Although behavioral changes in CAHgirls may be caused by androgen influ-ences on the developina brain, it hasbeen suggested that these changes mightinstead result from social or illness fac-tors (Quadagno, Briscoe, &. Quadagno,1977; Slipjer, 1984). For example, be-cause androgen levels are high in ult-rO,females have masculinized aenitalia.Suraical reconstruction is often neces-

Address correspondence to Sheri A. Be-renbaum, Department of Psychology. Univer-sity of Health Sciences/'l"be Chica,o MedicaJSchool, 3333 Green Bay Rd., North Chicaao.IL~.

203Copyrlpt C 1992 American Psychololical SocietyYOLo 3. NO.3. MAY 1992

CAH produce high levels of adrenal an-I drogens beginning in ut~ro. Postnatal

treatment with corticosteroids (and min-eralocorticoids for the 75% who are also

I salt-~osers) nonnalizes honnone levelsI (WhIte, New, &:. Dupont, 1987).I If early honnone exposure affects the

development of sex-typed behavior inhuman beings as it does in other species,then CAH girls should show behaviorsimilar to that of normal boys. Previous

I studies have suggested that CAH fe-males show intense physical eneray ex-penditure, "tomboyism," rough outdoorplay, preference for traditionally mascu-line toys and activities (Ehrhardt &:.Baker, 1974; Ehrhardt, Epstein, &:.

I Money, 1968), and greater spatial abilitythan their female relatives (Resnick, Be-renbaum, Gottesman, &:. Bouchard,1986). These findings parallel those re-poned in other samples with prenatal ex-

I posure to ~ul~nizing h.onnones due tomaternal Ingestion dunn, pregnancy I(Ehrhardt &:. Meyer-Bahlburg, 1979;Hines, 1982), and are especially interest-ing in light of recent repons of sex dif-

I ferences in human brain structure (Allen,Hines, Shryne. &:. Gorski. 1989; Hines &:.Green, 1991; Swaab &:. Fliers, 1985).

Although prior studies of CAH girlssuggest that honnones can influence

I childhood behavior, methodologicallim-

itations encouraged us to pursue this is-

sue. Specifically, in prior studies, (a) be-havior was assessed from interviewsrather than direct observation; (b) datawere collected with knowledge of the pa-

I tient or control status of the subject; (c)

behaviors were usually rated as presentor absent, rather than as continuoustraits; and (d) masculine and femininebehaviors were often not assessed sepa-

Irately, but instead treated as oppositeends of a sinaie continuum. Therefore,the present study of individuals withCAH focused on objective, quantifiablemeasures of sex-typed toy preferences,

I with assessments made by raters blind to

Subjects

We recruited 3- to 8-year-old childrenwith CAH from pediatric endocrine clin-ics at eight hospitals in the Midwest andCalifornia and tested 26 girls and IIboys. Because of the restricted age rangestudied and family constellations. only16 patients (43%) had a same-sex con-trol. Therefore. we combined relatives ofmale and female patients to obtain ourcontrol groups. The control groups con-sisted of 15 unaffected female relatives(10 sisters and 5 first cousins) and 18 un-affected male relatives (14 brothers and 4

I first cousins). Patients and controls didnot differ significantly in birth order orage. Mean ages in months were as fol-lows: female patients. 66.54 (ranle: 36-99); female controls. 61.80 (range: 36-93); male patients. 64.18 (range: 41-101);

I and male controls. 69.78 (range: 33-99).

PSYCHOLOGICAL SCIENCE

Hormones and Toy Preferences

the opposite preference. but the differ-ence was not significant. t( 14) = - 1.17.~

CAH Patient-ControlComparisons

the toys however he or she wanted. The12-min session was videotaped for laterscoring. The first 10 min of each sessionwere usually scored; the additional 2 minwere scored only if sections of the initial10 min were unscorable (e.g., if the childattempted interaction with the video-taper). Order of testing the patient andcontrol was random.

We scored the amount of time thechild played with each toy and thensummed the time spent in play with thetoys in each of the three categories, toproduce total scores for play with boys'toys, girls' toys, and neutral toys. Alltapes were rated by the same two raters.who had not tested the children and whowere blind to their patient or control sta-tus. Interrater reliability was very high:The median correlation was .99 for indi-vidual toys and .99 for total scores. Datareponed here represent the mean scoresof the two raters.!

sary, although postnatal treatment pre-vents further virilization. Parents maytreat their CAH daughters in a masculinefashion as a response to this masculineappearance at birth. Therefore, we askedparents to complete a questionnairewhich included the item "I encouragemy child to act as a girl should." Thequestionnaire was available for 24 CAHand 11 control girls.

We also examined characteristics ofthe child's disease: age at diagnosis, de-gree of genital virilization at diagnosis,and salt-losing status. This informationwas obtained from medical records,available on 25 girls and 10 boys. Medi-cal ratings were made by two researchassistants who had no knowledge of thechild's behavioral scores. Age at diagno-sis and salt-losing status were recordedwith perfect reliability. Most patientswere diagnosed in the early neonatal pe-riod: Median age at diagnosis was 11days for girls (range: 0 days to 64months) and 30 days for boys (range: 14days to 54 months). Degree of viriliza-tion was rated on a scale ranging from 0(normal female) to 6 (normal male), withintermediate values reflecting varyingdegrees of clitoral enlargement and fu~sion of the labia (Prader, 1954). Inter-rater reliability was .82; mean ratingswere used. AU girls had some degree ofgenital virilization: The mean Praderscore was 3.0 (range: 1-5). One girl hadbeen raised as a boy for the 1st month.

RESULTS

Statistical tests are one-tailed whenhypotheses are directional (sex differ-ences and femaJe CAH-control compar-isons for time spent in play with boys'and girls' toys) and two-tailed when nodifference is hypothesized (all compari-sons for play with neutral toys) or whenthe direction of the difference is notspecified (male CAH-control compari-sons).

Sex Differences

As expected. control boys and girlsdiffered in the amount of time theyplayed with boys' toys and girls' toys.but not neutral toys (see Fig. I). Themagnitudes of the differences (in stan-dard deviation units. or d; Cohen, 1977)are consistent with those reported byother investigators. Further. controlboys preferred boys' toys to girls' toys.1(17) = 4.58. P < .001. Control girls had

FemalesAs hypothesized, CAH girls spent

significantly more time playing withboys' toys than did control girls (d =.89), 1(37) = 2.66, p < .01, and about asmuch time as the control boys (see Fig.I). Further, like control boys, CAH girls Iplayed significantly more with boys' toys:then with girls' toys, 1(23) = 2.93, p <.01. This effect is also seen in the sub-sample of patients with a same-sex rela-tive control (N = II pairs): The meantime in play with boys' toys was 327.23 sfor CAH girls and 166.55 s for matchedgirls, 1(10) = 2.20, p < .05.

CAH girls also played less with girls'toys than did the control girls, 1(37) =- 2.02, p < .05, but this effect is smallerthan that for boys' toys (d = .65). Per-

haps because of low statistical power,the difference is not significant inmatched-pairs analysis of the II patient-control pairs, 1(10) = -1.09. Because

total play time was limited, play with oneset of toys tended to reduce play withanother set, so reduced play with girls'toys may have partly resulted from in-creased play with boys' toys. In fact, thecorrelations. among play with boys',girls', and neutral toys are all moderatelynegative, ranging from - .30 to - .69 (allps < .01, one-tailed). In this context,however, it is important to note thatthere were no differences between CAHand control girls in play with neutral toys(both matched and unmatched I-values< 1.0).

There were no significant differencesbetween CAH and control girls in par-ents' responses to the question "I en-courage my child to act as a girl should,"with 59.1% and 63.6% responding"yes," respectively. Amount of timespent playing with sex-typed toys wasnot significantly related to any disease

2. Because the distributions of play scoresare skewed. we conducted all analyses ondata transformed in various ways (e.g.. arc-sine. square root) and with nonparametricprocedures. All analytic procedures producedsimilar results.

I. Two children (both CAH Sirls) whoplayed with no toys durina their sessions wereexcluded from all analyses. These sirls aresimilar to the other subjects in disease char-acteristics. Analyses including these subjectsdo not change the interpretation of the results.

Toy Preference Materialsand ProcedureToy preference was measured by the

amount of time the child played withtoys shown by others to be preferred bygirls, by boys, or equally by the sexes(neutral). The boys' toys included trans-portation toys (a helicopter, two cars,and a fire engine) and construction toys(blocks and Lincoln Logs). The girls'toys included three dolls, kitchen sup-plies, a toy telephone, and crayons andpaper. Neutral toys, used as a control,included books, two board games, and ajigsaw puzzle. The toys were arranged ina standard order on the floor of the pe-diatric clinic or the child's home, in anarea approximately 8 ft by 10 ft sur-rounded by screens.

The child was brought into the playarea individually and told to play with

VOL. 3, NO.3, MAY 1992204

Sheri A. Berenbaum and Melissa Hines

c . Neutral ToysB. Girts' ToysA. Boys' Toys

-,.or

. .. ..:~

J~r

~.

.E.

~~i ~~

.~~:::;~;:'}

!..f

0-c.-- Q CN4 9po. III po.M)

c-. d c...d.,..,. "'."1,...

_Q-d -9-d.oa-9...d.. It"

Fig. 1. Time spent in play with sex-typed toys by female and male CAH patients and controls during 10 min of play. Scores arethe sum of play with individual toys and therefore may exceed 600 s. Bars represent group means; lines represent standard errors;points represent individual subjects; d = difference between group means/average standard deviation. Group differences wereevaluated by t test; .p < .05, ..p < .01, ...p < .001, one-tailed.

characteristic. For play with boys' toys.rs are - .10 for virilization. - .29 for ageat diagnosis. and .25 for salt-losing status(I = salt-loser. 0 = simple virilizer). Forplay with girls' toys. corresponding rsare - .09, - .14, and - .06. Note. how-ever. that the small sample size and re-stricted range (e.g., only 2 girls weresimple virilizers) reduce the power ofthese analyses.

MalesIn contrast to the differences ob-

served between CAH and control girls.there were no significant differences be-tween CAH boys and control boys onsex-typed play (ps > .20 for play withboys' and girls' toys). Examination ofmedical characteristics in relation to sex-typed play in boys is not meaningful be-cause of the lack of patient-control dif-ferences and limited variability in the

sample.

that masculinization of juvenile play isunrelated to genital virilization or mater-nal behavior (Goy, Bercovitch, & Mc-Brair, 1988). Nevertheless, given the dis-proportionate rate of return of question-naires for our CAH and control girls andother problems inherent in questionnairemeasures, direct observation of parentalbehavior would be valuable.

Our data indicate no changes in sex-typed toy preferences in CAH males,consistent with previous reports in non-human animals and people. CAH maleshave generally been reported not to dif-fer from controls in sex-typed activitiesor abilities (Ehrhardt & Baker, 1974;Resnick et al., 1986).

Although the data are consistent withan androgen influence on sex-typed toychoices, it is not necessary that hor-mones have a direct influence on thesechoices. Hormones may affect toychoices indirectly, perhaps through aninfluence on activity level, motor skills,abilities, or temperament. For example,CAH girls may be more active than con-trol girls, and boys' toys may facilitateactive play (O'Brien & Huston, 1985).

Results from this study may also berelevant to understanding the develop-ment of sex-typed toy preferences innormal children. Normal males havehigher testosterone levels than normalfemales from approximately week 8 toweek 24 of gestation and from approxi-mately the 1st to the 5th month of life(Smail, Reyes, Winter, & Faiman, 1981).These sex differences in hormones may

and behavioral ratings made withoutknowledge of patient or control status.The data also suggest a relative defemi-nization of toy preferences, although thisresult was less robust. Other studies ofhormone-exposed samples have alsobeen more likely to observe "masculin-izing" than "defeminizing" effects ofprenatal hormones, perhaps in part be-cause the studies have concentrated onthe former (Berenbaum, 1990; Ehrhardt& Meyer-Bahlburg, 1979; Hines, 1982).Although other hormones, such as 17-hydroxyprogesterone and corticoste-roids, are also abnormal prenatally inCAH, it is unlikely that they account forthe masculinized behavior in CAH girls,because these hormones have smallerand less consistent behavioral effectsthan androgen and may actually preventmasculinization (Erpino, 1975; Hull.Franz, Snyder, & Nishita, 1980).

It appears unlikely that the behavioralchanges are due to social or illness fac-tors. Our failure to find relationships be-tween sex-typed play and physical viril-ization is consistent with other data inCAH patients indicating no relationshipbetween degree of virilization and gen-der-role behavior (Slipjer, 1984). Al-though it is possible that any virilizationresults in different parental treatment,this argument is weakened by parents'retrospective reports that they did nottreat CAH girls in a "masculine" fash-ion, here and in other studies (Ehrhardt& Baker, 1974; Resnick, 1982), and bydata from rhesus macaques indicating

DISCUSSION

Large differences between CAH girlsand unaffected female relatives indicatemasculinization of toy preferences ingirls exposed to high levels of androgenduring early development. These resultsare consistent with data from animalmodels and with prior studies of CAHand other females exposed to masculin-izing honnones in utero. Our findingsstrengthen the conclusions of previousplay studies in CAH girls because weused objective. quantifiable measures

205YOLo 3. NO.3. MAY 1992

PSYCHOLOGICAL SCIENCE

Honnones and Toy Preferences

REFERENCEScontribute to subsequent sex differencesin behavior. such as those in toy prefer-ences. Further. natural variations in lev-els or availability of testosterone amongnonnaJ males and females might contrib-ute to individual differences in sex-typedbehavior.

Our data may also be relevant to ev-

idence that early sex-typed toy prefer-ences predict later behavior. includingsexual orientation (Green. 1987) and spa-tial ability (~e~.combe. Bandura. & Tay-lor. 1983; Shennan. 1967). and to evi-dence that spatial ability (Resnick et aJ.,1986) and ~exual orientation (Money.Sch~"anz. &: Lc~"is. 1984) are masculin-ized in CAH females (for discussion. seeBerenbaum. 19901. Specifically. it is pos-sible that hormonal innuences on adultbehaviors are mediated by childhoodsex-typed toy prefercnces. although it isalso possible that various sex-typed be-haviors are innucnced separately by hor-mone~ (Arnold &: Gorski. 1984; Goy &McE~.en. 1980; Goy ct al.. 1988). Fur-ther ~tudies of CAH and other endocrinesyndrome.. !ohould help us to understandnot only ~hcthcr ~Clnadal hormones in-fluence human t\Chavior. but also howthe)" do !oo.

Acknowlcdsmmc- Thi\ \tudy ~'as sup-poned b) Sational In"titules of HealthGrant\ HD19644. HD~4542. andSS~7, We thank the follo~'ing people~'ho contrit-uted to thi\ project: Drs.Stephen Duck. On'ille Green. Ora Pesco-,;u. Julio Santi~o. Jo Anne Brasel. Rob-en Clemon\. Genrude Costin. RichardFefferman. Lynda Fisher. FrancineKaufman. and Thomas Roe generouslypro,'ided aece~s to their patients and an-swered medical que~tions: Brenda Hend-enon. Andrt Black. Ruth Estes. ErinFoy. Kim Kern~. Deena Krumdick. Jen-nifer La"'rence. Naomi Lester. AnneMax~'ell. Kristie Nies. Ellen Rochman.Robyn Reed. Elizabeth Snyder. andChristopher Verbin helped ~;th data col-&ection. scoring. entry. and analysis: KimKerns establi~hed a ~uperb data-entry andtrackins system. "'e are particularlypateful to the subjects and their parentsfor their enthusiasm and cooperation. Wealso thank ~tichael Bailey. SusanResnick. ~tichael Taylor. and Gary Olt-mans for helpful comments on this paper.Portions of this paper ~'ere presented atthe meetins of the Society for Research inChild Development. Kansas City. April1989.

and neural ~lates of sex-typed bt;havionR,,'i, o[ 1'$)'('ltiGtry. 10. .536-.5.5.5.

Hull. E.M.. F~.I.R.. Snyder. A.M.." Nishita.I.K. (I_). PerinaW prQ8esterone and learn-ina. social and reproductive bt;havior in rats.I'IIrsiolo". GIld B~M"ior. 24. 2.51-2.56.

Huston. A.C. (1913). Sex-typina. In P.H. Mussen(Ed.). HGndbook o[ ('ftjld p$)'('hoIoO (Vol. IV.4th cd.. pp. 317-467). New York: Wiley.

Liss. M.B. (1911). Pattems of toy piay: An analysisof sex differences. S~z Rol~s. 7. 114)-1 ISO.

MacLusky. NJ.. a. Naflolin. F. (1911). Sexual dif-ferentiation of the central nervous system. S('j-~ftC~. 211. 1294-1303.

IMoney. J.. Schwanz. M.. a. Lewis. V.G. (1984).Adult eroIosexuai status and fetal honnonalmasculiniution and demasculinization: 46.XX conaenital virilwna adrenal hyperplasiaand 46. XV andrQIen-insensitivity syndromecompared. I'sy('IIoll~/I'o~lIdo(',illolo,y. 9.4OS-414.

Newcombt;. N.. Bandura. M.M.. a. Taylor. D.G.(1983). Sex differences in spatial ability andspatial activities. S~z Rol~s. 9. 377-386.

O'Brien. M.. a. Huston. A.C. (I9IS). Activity leveland sex stereotyped toy choice in toddler boysand lirls. }OIl,MI o[Grll~ti(' 1'$)'('IIolo".. 146.S27-S34.

Pradcr. A. (19S4). Der ,enitalbt;fund bt;im Pseudo-hermaphroditismus femininus des konleni-talen adrenoaenitalen Syndroms. H~/,.~ti(,GI'G~diGtri(,G A,('rG. 3. 231-248.

QuadaJ1lO. D.M.. Briscoe. R.. a. Qvada,no. I.S.(1977). Effects of perinatallonadal hormoneson selected nonsexual bt;havior patterns: Acritical assessment of the nonhuman and hu-man literature. I'sy('IIologica/ B/l//~till. 84. 62-80.

Resnick. S.M. (1982). 1'$)'('hologi('GI [/lII('tiolllll, illilldi,.id/lGIs ,,'ith (,oll'~lIirGI Gd'~MI ~),~'P/G-JIG: ED,,)' IIo""oMI illfl/lrll(,~s 011 ('ollliriO1lalld ~'solla/it).. Unpublished doctoral disser-tation. University of Minnesota. Minneapolis.

Resnick. S.M.. Berenbaum. S.A.. Goctesman. 1.1..a. Bouchard. TJ. (1986). Early hormonal in-fluences on coJnitive functionin, in con,enitaladrenal h)1'Crplasia. ~,'~/opm~lItGI 1'$)'('1Iol-00. n. 191-191.

Sherman. 1. (1967). Problem of sex differences inspace perception and aspects of intellectualfunctioning. I'syc-holo,ical R"'i~,,,. 74. ~299.

SJipjer. F.M.E. (1984). AndroJens and ,ender rolebt;havior in girls with conaenital adrenal hy-perplasia (CAH). In GJ. DeVries. J.P.C. De-Bruin. H.B.M. Uylinls. a. M.A. Comer(Eds.). I"al'~ss ill b,aill '~srGrch (Vol. 61.pp. 417-422). Amsterdam: Elsevier.

Smail. PJ.. Reyes. F.I.. W"mter.l.S.D.. a. Faiman.C. (1911). The fetal hormone environment andits effect on the morpholencsis of the ,enitalsystem. In SJ. Kosan a. E.S.E. Hafez (Eds.).l'~d;Gtric alldroioo (pp. 9-19). The Haaue:Martinus Nijhoff.

Swaab. D.F.. a. Fliers. E. (I9IS). A sexually dimor-phic nucleus in the human brain. SC;~IIC~. n8.1112-IIIS.

White. P.C.. New. M.I.. a. Dupont. B. (1987). Con-,enital adrenal hyperplasia. N~", EII,/Glld}OII,IIGI 0/ M~d;c;II~. 316. ISI9-1524.

(RECEIVED SnfJI; REVISION ACCEPTED 1~1)

Allen. L.S.. Hines. M.. Shrync. J.E.. &. Gorski.R.A. (1989). Two sexually dimorphic cellII'OUPS in the human brain. J_mal 0/ NrMro-s(';r~('r. 9. 497-506.

Amoid. A.P.. a. GORki. R.A. (1984). Gon8daJ ste-roid induction of structural sex differences inthe central nenrous system. Aft_al Rrvir.' 0/Nr_s('ir~('r. 1.413-442.

Baum. MJ.. a. SdlretJen. P. (I97S). Neuroendo-crine effects of perinatal androseniution inthe male ferret. progrrss ;ft Brain Rrsrarch.42. 343-3SS.

Beatty. W.W. (1979). Gon8daJ hormones and sexdifferences in nonreproductive behaviors inrodents: Oraanizational and acUvationai influ-ences. HomlOMS and Brha"ior. 12. 112-163.

Berenbaum. S.A. (1990). Conacnital adrenal hyper-plasia: Intellectual and psychosexual function.inc. In C. Holmes (Ed.). Ps,.choftrllrornd«,;-Ilol06)': Brain. b#ho,.;or. and hormonal inlrr-aCliolU (pp. 227-260). New York: Sprinser.Veriaa.

Bussey. K.. &. Bandura. A. (1984). Influence olacn-dcr constancy and social power on sex-linkedmodclinl. JOIImal of PrrsOllaliry and S«ialPsy('hololY.41. 1292-1302.

Cohen. J. (1977). SI4lisl;cal po..'rr a~a/ysis for Ihrbrhm'ioral s('irncrs (rev. ed.). New York: Ac-ademicPress.

Connor. J.M.. &. Serbin. L.A. (1977). Behaviorallybased masculine. and feminine.activity-prcfer-ence scales for preschoolers: Correlates withother classroom behaviors and coanitive tests.Child Dr"rlopmrlll. 48. 1411-1416.

Diamond. M.. Llacuna. A.. &. Wonl. C.L. (1973).Sex behavior after neonatal prosesterone. tes-tosterone. estrosen. or antiandrosens. Hor-monrs and Brha,'ior. 4. 73-88.

Ehrhardt. A.A.. &. Baker. S.W. (1974). Fetal andru-sens. human central nenrous system differen-tiation and behavior sex differences. In R.C.Friedman. R.M. Ricbart. &. R.L. Vande Wiele(Eds.). Srx dif!rrrll('rs ill brha,'ior (pp. 33-SI).New York: Wiley. .

Ebrhardt. A.A.. Epstein. R.. &. Money. J. (1968).Fetal androsens and female lender identity inthe early.treated adrenogenital syndrome.JohlU Hopkins Mrdical JOllma/. 122. 1~167.

Ehrbardt. A.A.. &. Meyer-Bablbura. H.F.L. (1979).Prenatal sex hormones and the developingbrain: Effects on psychosexual differentiationand colnitive function. Allnllal Rr,'ir.' ofMrdi('illr. JO. 417-430.

Erpino. MJ. (I97S). Androaen-induced aaressionin neonatally androsenized female mice. inhi-bition by prosesterone. Hormollrs and BrllDv-ior.6. 149-1Sg.

Goy. R.W.. Bcrcovitcb. F.B.. &. McBrair. M.C.(1988). Behavioral masculiniution is indepen-dent of lenital m&sculiniution in prenatallyandrogenized female rhesus macaques. Hor-mOllrs and BrllD,.ior. 22. SS2-S71.

Goy. R. W.. a. McEwen. B.S. (1980). Srxllal diffrr-rnlialion of 1M braill. Cambridle. MA: MITPress.

Green. R. (1987). Thr "sissy boy s,'lIdromr" and Ihrdr"rlopmrnl of homosrxlla/iry. New Haven.CT: Yale University Press.

Hines. M. (19g2). Prenatal JOnadal hormones andsex differences in human bebavior. Psy('holo,-i('al Bllllrlill, 92. S6-80.

Hines. M.. &. Green. R. (1991). Human bonnonal

YOLo 3, NO.3, MAY 1992206