Hormonal and experiential correlates of maternal responsiveness during pregnancy and the puerperium in human mothers

Hormones and Behavior 31, 145 – 158 (1997)
Article No. HB971376
Hormonal and Experiential Correlates of Maternal
Responsiveness during Pregnancy and the
Puerperium in Human Mothers
Alison S. FlemingDepartment of Psychology, University of Toronto, Erindale Campus,Mississauga, Ontario L5L 1C6, Canada Diane RubleDepartment of Psychology, New York University, New York, New York 10012 Howard Krieger and P. Y. WongToronto General Hospital, Toronto, Ontario M5G 2C4, Canada Two studies were undertaken (1) to determine whether
tachment to the infant across pregnancy. (3) However,
human mothers undergo a change in maternal respon-
the pattern of change in the ratio of estradiol to pro-
siveness during pregnancy before the birth of the baby,
gesterone from early to late pregnancy was related to
as shown for other mammalian species, and (2) to es-
postpartum attachment feelings. (4) Finally, hormonal
tablish whether a relation exists between changes in
correlates of attachment feelings may reflect effects
maternal feelings and attitudes and changes in hor-
both on feelings of nurturance directly and, indirectly,
mones. In both studies prospective first-time mothers
on mothers’ feelings of well-being.
᭧ 1997 Academic Press
completed an extensive set of questionnaires, cov-
ering a broad range of issues, including a set of 76-
to 100-item likert scales concerning attitudes toward
How new mothers care for their young is influ- infants, childbirth, pregnancy, caretaking, and other
enced by a variety of social and situational factors — interpersonal relationships. In the first cross-sectional
including their past experiences as infants (Suomi, study, mothers completed the questionnaires at one
1990; Kraemer, 1992; Main, Kaplan, and Cassidy, of seven time points, ranging from prior to pregnancy
1985; Fairbanks, 1989) and with infants (Pryce, 1993; to 3 months postpartum. In the longitudinal study,
questionnaires were completed repeatedly throughout
Ruppenthal, Arling, Harlow, Sackett, and Suomi, this same time period. In addition, blood was taken at
1976; Fleming, Ruble, Flett, and Shaul, 1988; see these same time points and assayed by RIA for plasma
Corter and Fleming, 1995), their life histories and concentrations of the steroids, estradiol, progester-
stresses, and their present social and physical con- one, testosterone, and cortisol. The primary findings
texts (see Corter and Fleming, 1995). Less obvious is are (1) feelings of nurturance grow during pregnancy
the role of biological factors in mothering and the and from pregnancy to postpartum; in the cross-sec-
process of becoming attached to one’s infant. How- tional study, for most of the factors relating to infants
ever, biological factors clearly contribute to moth- or mothering, pregnancy and postpartum responses
ering behavior in most other mammals that have were more positive than prepregnancy responses; in
been studied, including a number of primates the longitudinal study, many of these factors also
showed elevations across pregnancy itself, as well as
(Pryce, 1993, 1995) and there is no reason to assume further elevations with the birth of the infant. (2) Preg-
that humans are immune from these effects.
nancy hormones were not related to the growth of at-
Among many mammals, mothers show heightened 0018-506X/97 $25.00Copyright ᭧ 1997 by Academic PressAll rights of reproduction in any form reserved.
responsivity to offspring during the latter part of preg- not, however, occur in a vacuum. The mothers back- nancy, before the young are born, culminating in a ground experiences and experiences interacting with complete integrated response at parturition (see Corter the infant, in particular, also correlate with her feelings and Fleming, 1991; Maestriepi and Wallen, 1995; Gib- and attitudes and these may well accentuate, mask, or ber, 1986). Although we do not yet known which pre- interact with the hormonal correlates (Fleming, Ruble, cise hormones mediate responsiveness during gesta- tion, we know that the changes in hormones that occur The present set of studies is an extension of this ap- between mid- and late pregnancy or that accompany proach to the analysis of hormone – environment inter- parturition contribute to the rapid onset of nurturant actions in the regulation of maternal responsiveness in behavior that occurs with the birth of the young; in new mothers. In the present paper we report results general, the relevant pattern of hormonal changes in- which reflect our interest in hormonal factors during cludes a period with high levels of progesterone during pregnancy and across the parturitional period and their pregnancy followed by the decline of progesterone and relation to mothers’ maternal attitudes. In another pa- the associated rise in estrogen, prolactin, and oxytocin per (Fleming, Steiner, and Corter, in preparation), we (Rosenblatt, 1990; Bridges, 1990; Insel, 1990; Pryce, evaluate hormonal changes during the postpartum pe- 1993; Pryce, Doebli, and Martin, 1993; see Pryce, Mar- riod and their relation to mothers’ maternal behaviors.
Here we describe only a selected portion of two studies, However, hormonal influences are expressed only if one cross-sectional, the other longitudinal, each with the social and situational factors are favorable. Thus, its own extensive data set and both analyzed previously among some primates (e.g., marmosets) mothers will to address different questions (see Ruble, Brooks-Gunn, often not care for their offspring if they have not had Fleming, Fitzmaurice, Stangor, and Deutsch, 1990; Ru- prior juvenile experience allomothering — that is, caring ble, Fleming, Hackel, and Stangor, 1988). And in each for someone else’s offspring (Pryce, 1993; Pryce, Doebli, study we ask what is the evidence that reproductive and Martin, 1993; see Pryce, Martin, and Skuse, 1995).
state or hormones are related to maternal feelings and In another New World monkey (the red-bellied tama- attitudes during pregnancy and the puerperium in new rin) the effects of exogenous hormone replacement are mothers. A second goal of these studies is to simultane- also apparent only in previously experienced, multipar- ously assess the relation to mothering of mother’s prior ous animals (Pryce, Abbott, Hodges, and Martin, 1988).
experiences with infants; hence, we use hierarchical Among pigtail macaques social dominance status deter- multiple regression analyses to determine the variance mines the amount of interest shown during pregnancy in mothering explained by hormones and by previous in ‘‘foster’’ infants (Maestriepieri and Wallen, 1995).
Even in rats in which hormonal effects are clear anddistinct, they cannot override or inhibit the effects of aprior experience cannibalizing young or the effects of GENERAL METHODS
stress and overcrowding, etc. (see Fleming, Morgan,and Walsh, 1996).
The primary measure of maternal responsiveness in Not surprisingly, few studies have explored the role the present experiments was based on responses on of hormones in the regulation of human parenting, questionnaires. In addition, in the second, longitudinal and those that have report the counterintuitive results experiment, hormone concentrations were measured in that associate the adrenal hormones rather than the ovarian or placental steroids with parental respon-siveness (see Corter and Fleming, 1995; Fleming,Steiner, and Anderson, 1987). We found there was a Measures: Questionnaires (Experiments 1 and 2)
strong linear correlation between circulating postpar-tum concentrations of the adrenal steroid, cortisol, andthe intensity of mothers’ close contact approach behav- Women in both the cross-sectional and the longitudi- iors while interacting with their infants on Day 3 post- nal experiments completed questionnaires. The ques- partum. This relation was further enhanced if mothers tionnaire was quite large, requiring 1 – 112 hr to complete.
were also positively motivated toward infants, as ex- It consisted of almost 1000 items addressing a broad pressed in their pregnancy attitudes (Fleming, Steiner, range of issues including physical symptoms and and Anderson, 1987). These hormonal correlates do moods, eating behavior and alcohol consumption, baby temperament, demographics, and background, in addi- tency across time, (see Fleming, Ruble, Flett, and Shaul, tion to the maternal attitude items that represent the 1988; Fleming, Ruble, Flett, and Van Wagner, 1990).
Correlations between the CES and standardized mood One measure, the Childbearing Attitudes Question- scales (e.g., the MAACL (Zuckerman and Lubin, 1965) naire (CAQ), consisted of a 76-item scale concerning and BDI (Beck, Ward, Mendelson, Mock, and Erbaugh, attitudes toward the pregnancy and birth, self-esteem, 1961) are highly significant, ranging from 0.55 to 0.80 and other interpersonal relationships. The set of items in different studies (Fleming et al., 1988, 1990).
involved 19 different issues important to infants andchildbirth, consisting of 3 – 6 items each, counterbal-anced for direction to minimize response bias. Partici- EXPERIMENT 1: MATERNAL
pants completed this measure by responding to each ATTITUDES AT DIFFERENT STAGES
item on a 7-point scale ranging from disagree strongly OF PREGNANCY AND POSTPARTUM:
A CROSS-SECTIONAL ANALYSIS
The attitude factors used in the present report were derived from the set of attitude items by factor analysis (Ruble, Brooks-Gunn, Fleming, Fitzmaurice, Stangor,and Deutsch, 1990; see also Ruble, Fleming, Hackel, and Subjects
Stangor, 1988). In one analysis, based on responses of all667 subjects, 14 factors were extracted through iterated Subjects were recruited in a number of ways, includ- principal axis factor analysis, followed by a promax ing parent preparation, birthing, and exercise classes; oblique rotation (see Ruble et al., 1990). Of these, four newspaper and radio advertisements; and responses to will be discussed in the present report. The other analy- displays in the offices of obstetricians. Classes of vari- sis was based on a smaller Toronto sample of 246 ous types were the most frequent source of subjects, women (see Fleming, Ruble, Flett, and Shaul, 1988) who and to provide some measure of comparability across completed a 104-item questionnaire (including the orig- phases, pregnant women were randomly assigned to inal 76). In this analysis, the same basic factors were pregnant or postpartum (i.e., arrangements were made derived as in the larger analysis. However, in addition, to contact them at 1 or 3 months postpartum). Because 4 new clusters or factors (based on the additional items of the nature of the recruitment procedure, return rate included in the questionnaire) were derived and the was difficult to calculate precisely. Approximately 60% item intercorrelations were subjected to a principal- of the questionnaires sent to homes or left in doctors’ components factor analysis. Attitude clusters used in offices or classes were returned. Clearly, a self-selection the present set of studies assessed mothers’ feelings factor was operating and, unfortunately, it was not pos- of attachment or nurturance toward the growing fetus sible to determine on what dimensions responders and (fetus), attitudes toward preparing for the infant’s ar- rival (preparation), feelings of identification with the The cross-sectional sample was recruited from three pregnancy (pregnancy id), feelings of nurturance to- locations: Toronto, Seattle, and the greater New York ward own infant postpartum (attachment), feelings of area. The samples were recruited during the same time nurturance toward infants in general (other infants), period, which involved a term of approximately 2 years feelings about caretaking activities (caretake), feelings from 1981 to 1983. Approximately equal numbers of of adequacy in the maternal role (maternal adequacy), subjects in each childbearing phase participated at each feelings toward the partner or spouse (relationship with location.The two basic requirements for eligibility were partner), and feelings toward their own mothers (rela- that the woman was married (or cohabitating) and liv- tionship with mother). The factors fetus and prepara- ing with her husband (partner) at the time of the study tion were assessed only during the pregnancy. The and that the child was the woman’s first.
other attitude clusters were assessed at all time points.
The cross-sectional sample consisted of 667 women To determine mood state, mood factors derived from in 7 groups: (a) 113 women at 0 weeks who were plan- the current experience scale (CES) were used. These ning to conceive within 2 years (prepregnant); (b) 91 included factors that reflect feelings of anxiety, depres- women 1 – 17 weeks in their first trimester; (c) 91 sion, well-being, fatigue, and physical discomfort.
women 18 – 27 weeks in their second trimester; (d) 73 These mood factors have been found to be reliable and women 28 – 36 weeks in their third trimester; (e) 40 valid for the postpartum population and show consis- women at 37 weeks to the end of pregnancy; (f ) 129 women in their first month postpartum; and (g) 129 As shown in Figs. 1A and 1B, a significant stage effect women in their third postpartum month. The mean age was found for fetus (F(3, 290) Å 10.4, P õ 0.001), preg- for participants was 29, with a range of 18 to 42. The nancy identification (F(4, 381) Å 2.6, P õ 0.036), prepa- sample was 98% Caucasian and predominantly middle ration for the baby (F(3, 290) Å 29, P õ 0.000), maternal income, in social classes III, IV, and V (Hollingshead, adequacy feelings (F(6, 464) Å 7.6, P õ 0.001), and care- 1965). Over 70% of the women were college graduates.
taking feelings (F(6, 464) Å 4.3, P õ 0.001). As can beseen in Figs. 1A and 1B, for most of the motheringfactors the nonpregnant group gave the lowest ratingsand the postpartum groups gave the highest ratings Procedures
with the pregnant groups between the two (differencesbetween nonpregnant and 12 weeks postpartum sig- Participants took the questionnaires home with them nificant for maternal adequacy, caretaking, and preg- from class or from a physician’s office or they received nancy id). When one compares the pregnant groups, them by mail. They were asked to complete them at an interesting pattern emerges. For the factor, fetus, home and return them by mail within 2 weeks.
there occurred a precipitous increase in positive feel- The wording of items for the different childbearing ings between the first and the second trimesters (P õ groups or stages varied slightly in order to be appro- 0.05) and a slight decline once again at the very end of priate to each phase (e.g., ‘‘I expect to enjoy looking pregnancy (P õ 0.05). For the factors, feelings about pregnant’’; ‘‘I enjoy looking pregnant’’; and ‘‘I enjoyed other children and caretaking, ratings across the three looking pregnant’’). Since not all women filled out the trimesters did not change appreciably, but there also entire questionnaire at each time point and only a sub- occurred a slight (albeit nonsignificant) decline in rat- set of women gave blood for the radioimmunoassays ings in the 2 to 4 weeks prior to the infant’s birth. In different numbers of women are involved in different contrast, preparations for the infant (preparation) in- creased linearly across all the time points (comparisonsbetween each successive period, P õ 0.05) and maternaladequacy feelings (maternal adequacy) increased onlywith the birth of the infant (12 weeks postpartum ú In contrast to the mothering factors, no systematic pattern of change was seen in mothers’ relation with The first question we addressed was the extent to their mothers. However, there was a significant change which feelings of nurturance and related feelings about in mothers’ feelings about their spouses (relationship infants and self as mother change across pregnancy. A with partner; F(6, 464) Å 3.45, P õ 0.002) with the most related issue was how specific are these changes? Do positive feelings occurring during the second and third they apply only to infants or do they apply to all social trimesters and relatively more negative feelings prior to pregnancy and at 12 weeks postpartum (different Separate one-way ANOVAs (with seven stages, from midpregnancy, P õ 0.05).
groups) were undertaken on attitudes that reflect ma-ternal feelings. The first set of ANOVAs compared allseven groups in their feelings about infants, pregnancy, Discussion
caretaking activities, and their own maternal adequacy.
The second compared only the four pregnant groups These results indicate large differences between non- in their feelings toward the fetus and in preparations pregnant and postpartum women in attitudes toward being made for the baby. In addition, other social atti- caretaking and own maternal adequacy, with little tudes were assessed by comparing the seven groups in change across pregnancy. These differences probably their attitudes toward their spouses and their mothers.
reflect the differences between the groups in their actual There were significant main effects of stage on all experiences interacting with their infants and hence in factors that related to own infants, infant care, or moth- their comptence caring for infants (Ruble et al., 1990).
ering. In contrast, attitudes toward one’s own spouse, Where there seemed to occur a real growth in respon- toward mothers, or toward other (nonown) infants siveness across pregnancy itself was in the attitude fac- tors relating to mothers’ feelings about the fetus and FIG. 1. Changes in maternal attitudes across pregnancy and the postpartum period in seven cross-sectional samples of first-time mothers.
Scores reflect rating on a 7-point likert scales, ranging from negative (1) to positive (7).
preparation for the baby. Positive feelings about the income, in social classes III, IV, and V (Hollingshead, fetus increased primarily between the first and the sec- 1965) and 55% were college graduates.
ond trimesters, whereas preparation for the infant con- There were no significant differences (by t test) on tinued to intensity across the entire pregnancy. The any of the dependent measures (e.g., age, education, slight decline in positive feelings about the fetus as well income, attitudes) between women who completed all as that in feelings about other infants immediately prior the questionnaires and those completing only some.
to the birth is interesting in light of a similar prepartum There were also no significant demographic differences dip in responsiveness in monkeys (Maestriepieri and between those who gave blood and those who did not; Wallen, 1995). Again, these effects seem not to be due to in general, time constraints determined whether women inhibitory effects of hormones. More likely, they reflect did or did not give blood at any particular time point.
physical discomfort associated with the impending Moreover, a number of assays were eliminated because birth or with a heightened anxiety toward pregnancy of breakdown of a refrigeration unit in which the plas- These attitude changes are explored in greater detail in the next longitudinal experiment, where the contri- Questionnaires
butions to attitude change of both hormones and expe-rience are also considered.
The same questionnaires were given to the longitudi- nal sample as to the cross-sectional sample. In addition,the longitudinal sample also completed a childbirthquestionnaire on the first postpartum day. This child- STUDY 2: RELATION BETWEEN
birth questionnaire contained questions relating to de- HORMONAL AND ATTITUDINAL
mographics and background factors, birth events and CHANGES ACROSS PREGNANCY
experiences (e.g., length of labor, use of anesthetic), la- AND THE PUERPERIUM:
tency to first nurse or hold the infant, time spent nurs- A LONGITUDINAL ANALYSIS
ing, time spent in the presence of the infant during thefirst postpartum hours and days, as well as the maternal factor, feelings of attachment to the infant. Other atti-tude factors were not included in the childbirth ques-tionnaire.
Subjects
The longitudinal sample came from Toronto only and Hormonal Assays
participants were recruited by general practitioners orobstetricians who monitored them throughout the Blood was drawn at each stage of pregnancy and pregnancy and postpartum period. As with the cross- postpartum by the general practitioner/obstetrician in sectional sample, the two basic requirements for eligi- the doctor’s office (during pregnancy) or in the hospital bility were that the woman was married (or cohabitat- (Day 1 postpartum). Blood was drawn from the antecu- ing) and living with her husband (partner) at the time bital vein using a 5-ml syringe. Bloods were centri- of the study and that the child was the woman’s first.
fuged, aliquoted into cryovials, and stored at 080ЊC, Women were recruited into, and remained through- until assayed for cortisol, estradiol, testosterone, pro- out, the longitudinal study, which involved question- gesterone, and sex-hormone-binding globulin (SHBG).
naires (n Å 38) and hormonal assessments (n Å 29) at All hormones were assayed by radioimmunoassay. Es- five to six time points from gestational weeks 0 – 16 (first tradiol, testosterone, and SHBG kits were supplied by trimester), 20 – 27 (second trimester), 28 – 35 (third tri- DPC Corp. (Los Angeles, CA); progesterone kit was mester), and (in some women) 36 – 42 (end) as well as supplied by Johnson/Johnson (Toronto, Canada); and on Day 4 postpartum and Week 6 postpartum. Since cortisol was assayed by an in-house method (P. Y.
not all women filled out the entire questionnaire at each Wong, A. V. Mee, and F. Ho, Clini. Chem. 25, 914 – 917,
time point and only a subset of women gave blood 1979). The sensitivities of the hormone assays for corti- for RIA, different numbers of women are involved in sol, estradiol, testosterone, and progesterone were 20 different analyses. The participants in sample had a (nmol/L), 10 (pmol/L), 0.2 (nmol/L) and 0.1 (nmol/ mean age of 29, with a range of 18 to 36, and were L), respectively. Between-assay precisions for cortisol, over 80% Caucasian. They were predominantly middle estradiol, testosterone, and progesterone were as fol- lows: cortisol, 5.9% at 70 nmol/L and 7.1% at 580 nmol/ across the gestational period. Correlations computed L; estradiol, 4.9% at 270 pmol/L and 5.5% at 1600 pmol/ between concentrations of the different hormones at L; testosterone, 9.8% at 3.9 nmol/L and 5.4% at 22.4 two consecutive pregnancy periods were all highly sig- nmol/L; and progesterone, 4.8% at 7.9 nmol/L and nificant (P õ 0.05 to P õ 0.001). In contrast, there were 3.7% at 40 nmol/L. Recovery for cortisol, estradiol, tes- no meaningful correlations between hormone levels at tosterone, and progesterone was 92 – 103, 90 – 110, 93 – points during pregnancy and hormone levels either 4 days or 6 weeks postpartum. Moreover, there were nosignificant relations between levels of any of the hor-mones during pregnancy or postpartum and any of thebirth events (type of delivery, duration of labor, type of delivery, etc.). Analyses of circulating hormones duringpregnancy as a function of the gender of the child Pregnancy
showed a marginal difference between women giving Changes in maternal attitudes across pregnancy.
birth to male and female offspring in the ratio of testos- Within-group ANOVAs (including five or six time terone to sex-hormone binding globulin (T/SHBG) at points; three or four points during pregnancy, (in the end of pregnancy. Mothers of males had marginally some cases, 4 days postpartum) and 6 weeks postpar- higher levels of circulating biologically active testoster- tum) were undertaken on the same set of mothering and social factors as described for the cross-sectional Hormonal correlates: Correlations between hor-
mones and maternal attitudes. In order to determine
For this small sample, there were a number of sig- whether there is a significant relation between the dif- nificant changes across pregnancy that were consistent ferent hormones and maternal feelings, Pearson corre- with the cross-sectional comparisons. A series of one- lation coefficients were computed relating for each time way ANOVAs, with repeated measures, were per- point separately, levels of hormones (both total and formed comparing women’s attitudes across the first free) and the different maternal factors. To determine three trimesters of pregnancy. These comparisons were whether the hormones are related to other characteris- recomputed including the 6-week time point. As shown tics of the mother, correlations were also computed re- in Figs. 2A – 2C, there was a significant growth across lating the hormones to mothers’ anxiety, depression, pregnancy in mother’s feelings of attachment to the and well-being and to both past experience with infants fetus/baby (F(2, 36) Å 6.23, P õ 0.005), feelings about and postpartum experience. Only consistent (two or caretaking activities (F(2, 38) Å 5.72, P õ 0.007), and more time points) relations are reported.
preparations for the baby (F(2, 18) Å 6.89, P õ 0.006).
There were no significant correlations between any Worries about the infant also subsided over this same hormone (either total or free fractions) or hormone ra- period (F(2, 34) Å 4.69, P õ 0.016). In contrast, there tios and any of the maternal attitudes within any preg- were no changes in women’s feelings about their part- nancy period. Of particular interest was the observation ners, their mothers, or other children.
that none of the hormones correlated with feelings of Changes in hormones across pregnancy. Values of
attachment to the infant or fetus within the same preg- the hormone concentrations were expressed both as to- nancy time point or between hormones at one time and tal concentrations of the different sex steroids (estradiol, attachment feelings at the next time point. Moreover, progesterone, and testosterone) and cortisol and as free when change scores were computed for the different (e.g., biologically active) values (the ratio of the total sex hormones between successive pregnancy time points, steroids to the sex hormone-binding globulin (SHBG)).
they showed no correlation with maternal attitudes at Since, in other animals (Rosenblatt, 1990; Insel, 1990; the later (i.e., the second of two consecutive) pregnancy Bridges, 1990; Poindron and Levy, 1990), the steroids estradiol and progesterone, and, especially, the ratio of Reinforcing these results, when women were divided these two hormones at the end of pregnancy, are known by median split on different maternal attitudes at each to be important for maternal behavior, in the present of the pregnancy time points, indicating low versus paper analyses were also done on the estradiol/proges- high feelings of attachment to the fetus, etc. there were no significant group differences in their different hor- As shown in Figs. 3A and 3B there were very large mone concentrations either at the same or at previous and highly significant changes in all the hormones FIG. 2. Changes in maternal attitudes across pregnancy in a longitudinal sample of first-time mothers. Scores reflect ratings on a 7-point likert
scales, ranging from 1 (negative) to 7 (positive).
Hormonal correlates: Correlations between hor-
õ 0.05). However, prior experience with infants did mones and mood. In contrast, hormones were related
not relate to their relationship with their partners or to mothers’ mood state. During the second trimester and at the end of pregnancy, mothers with either highfree estradiol or high estradiol to progesterone ratios Postpartum
experienced more negative mood (depression and anxi-ety) (second: r Å 0.72, n Å 10, P õ 0.001; r Å 0.55, n Å Hormonal correlates of attachment. When moth-
19, P õ 0.014; third: r Å 0.40, n Å 17, P Å 0.05, 1 T and ers’ immediate postpartum, as opposed to pregnancy, r Å 0.52, n Å 16, P õ 0.039) and reduced well-being feelings of attachment to the infant were considered, a (second: r Å 00.55, n Å 19, P õ 0.01).
different pattern of results was found. As indicated un- Prior experience correlates. In contrast, when
der Methods, the attachment factor is based on re- mother’s prior experience with infants (assessed at 5 sponses to questionnaires and includes such items as months of pregnancy) was correlated with maternal at- ‘‘I am amazed and thrilled when I look (think of) at my titudes during pregnancy, a number of interesting rela- baby,’’ and ‘‘I do not feel as close to my baby as I tions were found. Mothers who had had more prior expected.’’ Although there were no significant relations childcare experience (with siblings or infants of friends between postpartum feelings of attachment to the in- or within the babysitting context) felt more positively fant and concentrations of postpartum hormones, post- during the second or third trimester of the pregnancy partum attachment feelings were related to pregnancy periods about caretaking (r Å 0.43, P õ 0.04), about hormones. Mothers with high feelings of attachment children, in general (r Å 0.75, n Å 23, P õ 0.001) and during the early postpartum period had a low ratio of about their own fetus (r Å 0.41, P õ 0.05). Already at estrogen to progesterone concentrations at the first two the beginning of pregnancy (first trimester), they also pregnancy time points, generating correlations between felt more adequate in the mothering role (r Å 0.39, P postpartum attachment and E/P hormone ratios at 5 FIG. 2—Continued
FIG. 3. Changes in plasma concentration of steroids across pregnancy and the postpartum period in a longitudinal sample of first-time mothers.
and 7 months of r Å 00.45 (n Å 21, P õ 0.04) and r Å postpartum feelings of attachment to their infants (5 00.50 (n Å 20, P õ 0.02), respectively. These relations months, 7 months, and 9 months, to 4 days postpartum, seem to be due primarily to concentrations of estradiol r Å 00.60, n Å 17, P õ 0.01; r Å 00.35, n Å 16, P Å during pregnancy (both total and unbound), which 0.18; and r Å 00.35, n Å 17, P Å 0.16, respectively). Put showed essentially the same inverse relation with post- another way, mothers who maintained a higher E/P partum attachment (for each of the pregnancy time ratio over this time period (from pregnancy to postpar- points hormones correlated with postpartum attach- tum) tended to have higher attachment feelings post- ment; e.g., total: rs Å 00.48 to 00.52, n Å 18 – 20, P õ partum. Again, these relations to the E/P ratio were 0.018; unbound, r Å 00.34 to 00.56, n Å 18 – 20, P õ due primarily to changes in estrogen levels. Mothers who maintained relatively higher levels of estradiol Moreover, consistent with the literature, mothers over the parturitional period (less of a decline) experi- who experienced a smaller decrease in the E/P ratio enced marginally higher attachment feelings postpar- from early pregnancy to day 4 postpartum had higher tum (5 months: r Å 00.44, n Å 17, P Å 0.075; 7 months: accounted for by hormones. Hierarchical regressionanalyses were undertaken on attachment postpartumin which first postpartum mood and then the changein the E/P ratio or E from pregnancy to postpartumwere entered. Separate analyses were done in whichwe entered, as predictors, change in hormones from 5,7, and 9 months of pregnancy to the postpartum period.
Considering the change from 9 months of pregnancyto the postpartum, analyses showed that well-beingpostpartum accounted for 32% of the variance in attach-ment (F(1, 16) Å 7.8, P õ 0.01; R2 Å 0.327) but thatchange in the E/P ratio or in E levels made no signifi- FIG. 4. Plasma estradiol/progesterone ratios during different
cant additional contributions (E/P: F(2, 14) Å 3.3, P Å phases of pregnancy in women expressing high and low levels of 0.06; R2 Å 0.37; E: F(2, 15) Å 4.4, P õ 0.03, R2 Å 0.37).
attachment (by median split) to their infants during the first few Using hormonal change from earlier pregnancy time- points in the equation produced somewhat differentresults. For both the 5- and 7-month time periodschanges in the E/P ratio or in E alone made a significant r Å 00.45, P Å 0.075; 9 months: r Å 00.32, n Å 18, P additional contribution to attachment postpartum, after Å 0.11). Progesterone concentrations per se appeared the entry of postpartum well-being. Thus, the change in E levels from 5 or 7 months of pregnancy to postpartum The relation between postpartum attachment and accounted for an additional 12 and 28% of the variance changes in the E/P ratio between pregnancy time points, in postpartum attachment, respectively. The overall as opposed to across the parturitional period, were ana- variance accounted for by both well-being and estradiol lyzed. Mother’s postpartum attachment scores were di- change from the 5- and 7-month time periods is 43% vided by median split into high versus low postpartum (F(2, 14) Å 5.2, P õ 0.02) and 48% (F(2, 13) Å 6.2, P õ attachment scores, and ANOVAs were performed on the change in the E/P ratio from the fifth to the ninth Prior experience effects. The relation of prior expe-
months of pregnancy. This analysis showed a signifi- rience to postpartum attachment was also evaluated.
cant interaction between postpartum attachment status Surprisingly, prior experience caring for infants did not (high vs low) 1 E/P ratio change (F(1, 17) Å 6.1, P õ .04.
correlate with attachment to own infant postpartum (r As shown in Fig. 4 mothers expressing high attachment Å 0.30, n Å 19, P Å 0.20). Other attitudes during preg- underwent an increase from the early to late pregnancy nancy were, however, predictive of postpartum feelings in the E/P ratio whereas those with low attachment of attachment. Interestingly, attitudes during the early experienced a decrease in the E/P ratio over this same pregnancy period were better predictors of Day 3 post- partum attachment than attitudes expressed at 7 or 9 Mood correlates of attachment. Attachment post-
months of pregnancy. For instance, mothers who felt partum was also related to mood postpartum. Mothers more attached to their infants postpartum felt more who felt more elated and happier postpartum also felt attached to the fetus, had more positive feelings toward more attached to their infants (r Å 0.57, n Å 17, P õ other infants and liked caretaking activities more dur- 0.01). Finally, the E/P ratios and E from pregnancy (5, ing the first trimester of the pregnancy (rs Å 0.47 – 0.51, 7, and 9 months) also correlated with mood postpartum n Å 18–20, P õ 0.03–0.05). Moreover, these mothers (E/P: 5 months, r Å 00.37, n Å 20, P Å 0.05, 1 t; 7 tended to have a better overall sense of well-being early months, r Å 00.44, n Å 19, P Å 0.03, 1 t; 9 months: r in the pregnancy (r Å 0.46, n Å 18, P õ 0.05). Finally, Å 00.43, n Å 20, P õ 0.05; E: 5 months: r Å 00.28, n there was no relation between either pre- or postpartum Å 20, ns; 7 months: r Å 00.41, n Å 19, P Å 0.04, 1 t; 9 hormones and the amount of reported prior experience months: r Å 00.44, n Å 21, P õ 0.05).
Multiple regression analyses. Given these relations
Hormone 1 experience correlates. Although prior
between hormones and postpartum mood we tested experience with infants did not correlate with postpar- the hypothesis that mother’s mood state postpartum in tum attachment feelings, the possibility still exists that fact accounts for much of the variance in attachment hormonal effects would enhance the effects of experi- ence on mothering. To test this, we undertook a series more positive feelings during pregnancy and relatively of multiple regression analyses in which we first en- less positive feelings pre- and postpregnancy.
tered prior childcare experience, then hormones (either The results of these studies suggest that the pattern pregnancy E/P ratios based on total concentrations and of change in maternal attitude factors across the preg- total or unbound E levels during the different preg- nancy period can best be explained, not by hormonal nancy stages or change in hormone levels from the dif- changes, but rather by changes in the women’s cogni- ferent pregnancy time points to postpartum). The last tions and expectations. For instance although in other term entered into the equations was the hormone 1 species, the growth in responsiveness across pregnancy experience interaction term. Experience per se did not seems to be dependent on hormones (see Fleming and predict attachment on Postpartum Day 4, whereas for Corter, 1988; Corter and Fleming, 1995), in women we a number of the analyses the hormones accounted for find no relation between any of the hormones at any a significant proportion of the variance; for example, pregnancy time point and any of the maternal attitudes.
prior experience with infants accounted for 13% of the We interpret the pregnancy-related elevation in posi- variance in postpartum attachment feelings (not a sig- tive feelings on a number of the maternal factors to be nificant effect) and the E/P ratio change from early due instead to the cognitive changes associated with pregnancy to postpartum accounted for an additional the knowledge of being pregnant and the associated 26% of the variance (a significant change) bringing the demand characteristics (Fleming and Corter, 1988; total variance accounted for to 39% (F(2, 13) Å 4.1, P õ 0.03). However, in none of the analyses did the inter- Also, the change in mothers’ positive feelings toward action term make an additional contribution.
the fetus during the second trimester in both the cross-sectional and the longitudinal samples is consistentwith descriptive results reported by Leifer (1977, 1980) DISCUSSION
that positive feelings at or around 20 weeks of gestationcoincides with the ‘‘quickening’’ or time of first percep- The primary findings of these studies are (1) feelings tion by mothers of the movement of the fetus. Leifer of nurturance grow during pregnancy and from preg- describes this as a time when the fetus is first perceived nancy to postpartum and (2) although hormones play as a distinct and unique individual, separate from the no role in the growth of attachment feelings across mother. Unfortunately, since we did not specifically ask pregnancy, they may well contribute to postpartum mothers when they first felt fetal movement, we cannot nurturant feelings; mothers showing less of a decline correlate time of onset of fetal movement and elevated in the ratio of estradiol to progesterone or of estradiol levels from pregnancy to the early postpartum period Although mother’s postpartum feelings about the reported the highest attachment feelings. Moreover, newborn were not related to the amount of time she mothers with higher postpartum attachment feelings had spent with infants prior to the pregnancy, they showed an increase in the E/P ratio from the fifth and were positively related to her earlier feelings about in- from the seventh month, to the ninth month of preg- fants and infant-related issues and overall well-being nancy, whereas those with lower postpartum attach- during earlier stages of pregnancy. Whether these early ment scores showed no change or a decrease in the pregnancy attitudes actually reflect prepregnancy atti- ratio over this same time period. Finally, (3) hormonal tudes, we cannot tell from our longitudinal sample.
effects may act on feelings of nurturance directly and However, the fact that there tends to be a high positive by influencing mothers’ feelings of well-being.
correlation between attitude factors across different Although there occurs across pregnancy a change in stages of pregnancy suggests that they may in part. In a number of maternal factors, different factors show contrast, our cross-sectional samples clearly show that different patterns of change. In the cross-sectional first trimester mothers express more positive maternal study, for most of the factors relating to infants or moth- attitudes than do non pregnant mothers, suggesting ering, pregnancy and postpartum responses were more that something does indeed happen attitudinally with positive than prepregnancy responses; in the longitudi- pregnancy onset. Consistent with the decline in positive nal study, many of these factors also showed elevations maternal feelings toward the end of pregnancy, the re- across the pregnancy itself as well as further elevations lation between pregnancy attitudes and postpartum at- with the birth of the infant. For the factor related to the tachment drops out when attitudes later in pregnancy partner or spouse, the opposite pattern was found, with are correlated with postpartum feelings.
responsiveness in humans: Usefulness of an animal model. Psycho- In contrast to the absence of a strong hormonal corre- neuroendocrinology 13, 189 – 212.
late of maternal feelings during pregnancy, the early Fleming, A. S., Morgan, H. D., and Walsh, C. (1996). Experiential fac- postpartum feelings of nurturance or attachment to the tors in postpartum regulation of maternal care. Adv. Study Behav. infant seem to be related to the change in estrogen or Fleming, A. S., Ruble, D. N., Flett, G. L., and Shaul, D. (1988). Postpar- the estrogen/progesterone ratio from early pregnancy tum adjustment in first-time mothers: Relations between mood, to late pregnancy and to the early postpartum period.
maternal attitudes and mother – infant interactions. Dev. Psychol. 24,
77 – 81.
Mothers who undergo less of a decline in the E/P ratio Fleming, A. S., Ruble, D. N., Flett, G. L., and Van Wagner, V. (1990).
or in E from pregnancy to postpartum or who in fact Postpartum adjustment in first-time mothers: Changes in mood and experienced an increase in the E/P ratio across preg- mood content during the early postpartum months. Dev. Psychol. nancy were more positively disposed toward their in- 26, 137 – 143.
fants immediately after the birth. In contrast, the abso- Fleming, A. S., Steiner, M., and Anderson, V. (1987). Hormonal and attitudinal correlates of maternal behavior during the early postpar- lute levels or ratios of these hormones either during tum period. J. Reprod. Infant Psychol. 5, 193 – 205.
pregnancy or during the postpartum period were not Gibber, J. R. (1986). Infant-directed behavior of rhesus monkeys dur- related to postpartum responsiveness. There are, unfor- ing their first pregnancy and parturition. Folia Primatol. 47, 118 –
tunately, no animal studies that explore the relation between hormonal profiles during the early stages of Hollingshead, A. (1965). Two-factor Index of Social Position. Yale Station, pregnancy and postpartum maternal behavior with Insel, T. R. (1990). Oxytocin and maternal behavior. In N. A. Kras- which to compare these results. In fact, there are re- negor and R. S. Bridges (Eds.), Mammalian Parenting: Biochemical, markably few studies that undertake correlational anal- Neurobiological and Behavioral Determinants, pp. 260 – 280. Oxford Although we cannot comment on the temporal pa- Keverne, E. B., and Kendrick, K. M. (1990). Neurochemical changes rameters of the relevant hormonal change, the present accompanying parturition and their significance for maternal be-havior. In N. A. Krasnegor and R. S. Bridges (Eds.), Mammalian Par- results are consistent with findings for other mamma- enting: Biochemical, Neurobiological and Behavioral Determinants, pp.
lian mothers in showing a relation between the change 281 – 304. Oxford Univ. Press, New York.
or dynamics of the steroids estradiol and progesterone Keverne, E. B. (1995). Neurochemical changes accompanying the re- during pregnancy and over parturition and maternal productive process: Their significance for maternal care in primates responsiveness in human mothers (Rosenblatt, 1990; and other mammals. In C. R. Pryce, R. D. Martin, and D. Skuse Keverne, 1990; Bridges, 1990; Insel, 1990).
(Eds.), Motherhood in Human and Nonhuman Primates — Biosocial De-terminants, pp. 69 – 77. Karger, Basel.
Kraemer, G. W. (1992). A psychobiological theory of attachment. Be- hav. Brain Sci. 15, 493 – 541.
REFERENCES
Leifer, M. (1977). Psychological changes accompanying pregnancy and motherhood. Genet. Psychol. Monogr. 95, 55 – 96.
Leifer, M. (1980). Psychological Effects of Motherhood: A Study of First Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., and Erbaugh, J.
Pregnancy. Praeger, New York.
(1961). An inventory for measuring depression. Arch. Gen. Psych. Maestriepieri, D., and Wallen, K. (1995). Interest in infants varies with 4, 561 – 571.
reproductive condition in group-living female pigtail macaques Bridges, R. S. (1990). Endocrine regulation of parental behavior in (Macaca nemestrina). Physiol. Behav. 57(2), 353 – 358.
rodents. In N. A. Krasnegor and R. S. Bridges (Eds.), Mammalian Main, M., Kaplan, N., and Cassidy, J. (1985). Security in infant child- Parenting: Biochemical, Neurobiological and Behavioral Determinants, hood and adulthood — A move to the level of representation. In I.
pp. 93 – 117. Oxford Univ. Press, New York.
Bretherton and E. Waters (Eds.), Growing Points of Attachment Theory Corter, C., and Fleming, A. S. (1991). Maternal responsiveness in hu- and Research. Monographs of the Society for Research in Child Develop- mans: Emotional, cognitive and biological factors. Adv. Study Behav. ment, Vol. 50, 1-2, Serial No. 209.
19, 83 – 136.
Poindron, P., and Levy, F. (1990). Physiological, sensory and experien- Corter, C., and Fleming, A. S. (1995). Fathers’ and mothers’ respon- tial determinants of maternal behavior in sheep. In N. A. Krasnegor siveness to newborns: The roles of attitudes, experience and infant and R. S. Bridges (Eds.), Mammalian Parenting: Biochemical, Neurobio- odors. Poster presented at the biennial meetings of the Society for logical and Behavioral Determinants, pp. 133 – 156. Oxford Univ. Press, Research in Child Development, Indianapolis, March.
Corter, C., and Fleming, A. S. (1995). Psychobiology of maternal be- Pryce, C. R. (1995). Determinants of motherhood in human and non- havior in human beings. In M. Bornstein (Ed.), Handbook of Parent-ing, Vol. 2, Biology and Ecology of Parenting, pp. 87 – 116. Erlbaum, human primates: A biosocial model. In C. R. Pryce, R. D. Martin, and D. Skuse (Eds.), Motherhood in Human and Nonhuman Primates — Fairbanks, L. A. (1989). Early experience and cross-generational conti- Biosocial Determinants, pp. 1 – 15, Karger, Basel.
nuity of mother – infant contact in vervet monkeys. Dev. Psychobiol. Pryce, C. R. (1993). The regulation of maternal behavior in marmosets 22, 669 – 681.
and tamarins. Behav. Processes 30, 201 – 224.
Fleming, A. S., and Corter, C. (1988). Factors influencing maternal Pryce, C. R., Abbott, D. H., Hodges, J. K., and Martin, R. D. (1988).
Maternal behavior is related to prepartum urinary estradiol levels Measurement, stability and change. J. Personality Soc. Psychol. 58(3),
in red-bellied tamarin monkeys. Physiol. Behav. 44, 717 – 726.
Pryce, C. R., Dobeli, M., and Martin, R. D. (1993). Effects of sex ste- Ruble, D., Fleming, A. S., Hackel, L., and Stangor, C. (1988). Changes in the marital relationship during the transition to first-time moth- roids on maternal motivation in the common marmoset (Callithrix erhood: Effects of violated expectations concerning division of jacchus): Development and application of an operant system with household labor. J. Personality Soc. Psychol. 55, 78 – 87.
maternal reinforcement. J. Comp. Psychol. 107(1), 99 – 115.
Ruppenthal, G. C., Arling, G. L., Harlow, H. F., Sackett, G. P., and Pryce, C. R., Martin, R. D., and Skuse, D. (1995). Motherhood in Human Suomi, S. J. (1976). A 10-year perspective of motherless-mother be- and Nonhuman Primates: Biosocial Determinants. Karger, Basel.
havior. J. Abnormal Psychol. 85, 341 – 349.
Rosenblatt, J. S. (1990). Landmarks in the physiological study of ma- Suomi, S. J. (1990). The role of tactile contact in rhesus monkey social ternal behavior with special reference to the rat. In N. A. Krasnegor development. In K. E. Barnard and T. B. Brazellton (Eds.), Touch: and R. S. Bridges (Eds.), Mammalian Parenting: Biochemical, Neurobio- The Foundation of Experience, pp. 129 – 164. Internal Univ. Press, Mad- logical and Behavioral Determinants, pp. 40 – 60. Oxford Univ. Press, Zuckerman, M., and Lubin, B. (1965). Manual for the Multiple Affect Ruble, D. N., Brooks-Gunn, J., Fleming, A. S., Fitzmaurice, G., Stangor, Adjective Checklist. Educational and Industrial Testing Service, San C., and Deutsch, F. (1990). Transition to motherhood and the self:

Source: http://www.flyfishingdevon.co.uk/salmon/year2/hormones/hormonal%20and%20experiential%20correlates%20of%20maternal%20responsiven.pdf