Psychophysiology, 41 (2004), 510–520. Blackwell Publishing Inc. Printed in the USA.
Copyright r 2004 Society for Psychophysiological Research
DOI: 10.1111/j.1469-8986.2004.00187.x
The psychophysiology of the maternal–fetal relationship
JANET A. DIPIETRO,a RAFAEL A. IRIZARRY,b KATHLEEN A. COSTIGAN,c and EDITH D.
GUREWITSCHc
a
Department of Population and Family Health Sciences, Johns Hopkins University, Baltimore, Maryland, USA
Department of Biostatistics, Johns Hopkins University, Baltimore, Maryland, USA
c
Division of Maternal–fetal Medicine, Johns Hopkins University, Baltimore, Maryland, USA
b
Abstract
The enigmatic quality of the maternal–fetal relationship has been extolled throughout history with little empirical
support. We apply time series analysis to data for 137 maternal–fetal pairs collected at 20, 24, 28, 32, 36, and 38 weeks
gestation. Maternal heart rate and skin conductance data were digitized in tandem with fetal heart rate and motor
activity. No temporal relations between fetal heart rate and either maternal variable were found, although averaged
maternal and fetal heart rates were correlated from 32 weeks. Consistent temporal associations between fetal movement and maternal heart rate and skin conductance were detected. Fetal movement stimulated rises in each parameter,
peaking at 2 and 3 s, respectively. Associations did not change over gestation, were unaffected by a maternal stressor,
and showed within-pair stability. The bidirectional nature of the maternal–fetal relationship is considered.
Descriptors: Fetal heart rate, Fetal movement, Pregnancy, Maternal stress
subcutaneous electrodes resulted in a strong association when
mean values were averaged over the course of the day (r 5 .78) or
hour (r 5 .71; Patrick, Campbell, Carmichael, & Probert, 1982).
Although fetal heart rate has a circadian rhythm with lowest
levels occurring in the middle of the night (Patrick et al., 1982), it
does not appear to be affected by maternal sleep states per se
(Hoppenbrouwers et al., 1981).
Cursory understanding of the psychophysiological association between mother and fetus has been generated by studies that
monitor the fetus during periods of maternal arousal. Physical
exertion is associated with elevation in fetal heart rate (Artal et
al., 1986) and declines in both variability (Macphail, Davies,
Victory, & Wolfe, 2000; Manders, Sonder, Mulder, & Visser,
1997) and motor activity (Manders et al., 1997) measured shortly
after the termination of exercise. Induced psychological stress, as
effected through the Stroop Color–Word test, has been associated with suppressed motor activity and increased fetal heart rate
variability (DiPietro, Costigan, & Gurewitsch, 2003). Another
study reported increased fetal heart rate in women with high, but
not low, anxiety (Monk et al., 2000).
There are a handful of studies that have evaluated the relations among maternal dispositional attributes, including anxiety
and stress perception, and fetal neurobehavior. The most consistent, although not universal, findings indicate that persistent
emotional arousal is associated with greater levels of fetal motor
activity (DiPietro, Hilton, Hawkins, Costigan, & Pressman,
2002; Ianniruberto & Tajani, 1981; Sontag, 1941; Van den Bergh
et al., 1989). Examination of fetal heart rate measures, including
variability, has yielded less consistent results: negative associations with maternal stress over the second half of gestation in one
cohort (DiPietro, Hodgson, Costigan, Hilton, & Johnson, 1996),
There is perhaps no relationship more profound yet enigmatic
than the first one. The bond between mother and fetus has been
extolled throughout history and literature, with supposition on
its nature ranging from the material to the metaphysical. Yet
there has been surprisingly little academic examination of the
maternal–fetal physiologic connection beyond that providing
direct obstetric relevance.
The most common indicators of fetal functioning are heart
rate and heart rate variability. There is a substantial body of
knowledge concerning the neural genesis of both over the course
of gestation (Dalton, Dawes, & Patrick, 1983; Martin, 1978;
Parer, 1999; Yoshizato et al., 1994) but there are few reports of
associations with maternal parameters under baseline conditions.
Early reports using relatively primitive methods of maternal–
fetal monitoring failed to detect associations between maternal
and fetal heart rate (Sontag & Richards, 1938). A brief (15 min)
recording using more sophisticated monitoring techniques also
failed to detect significant correlations (Lewis, Wilson, Ban, &
Baumel, 1970). However, subsequent implementation of 24-hr
recording of both maternal and fetal electrocardiogram using
This research was supported by grant R01 HD27592, National Institute of Child Health and Human Development, awarded to the first
author. Preliminary results were presented at the International Conference on Infant Studies, Brighton, England, July 2000. We are grateful for
the diligent and generous participation of our study families, without
which this research would not have been possible.
Address reprint requests to: Janet A. DiPietro, Ph.D., Johns Hopkins
Blomberg School of Public Health, Department of Population and
Family Health Sciences, 280 Hampton House, 624 N. Broadway, Baltimore, MD 21205, USA. E-mail: [email protected].
510
Maternal–fetal psychophysiology
but positive associations with pregnancy-specific stress at term
only in another (DiPietro et al., 2002).
The goal of the current study was to examine, for the first
time, the temporal nature of the maternal–fetal relationship by
applying time series analysis to fundamental features of maternal
and fetal functioning. This is an initial step toward ascertaining
whether the maternal–fetal associations detailed above are mediated through immediate and direct stimulation of the fetus
through maternal physiologic processes. Our method of antepartum fetal monitoring generates two streams of Doppler-based
fetal data: fetal heart rate and fetal motor activity. Maternal data
collection included measures of heart rate and electrodermal activity. Given the small but growing literature indicating that induced and chronic maternal sympathetic activation is associated
with fetal heart rate and/or motor activity, we expected that
changes in maternal physiological measures would be followed
by fluctuations in fetal heart rate and motor activity. Prediction
of the direction of effects was difficult given the limited and inconsistent body of knowledge in this area. We have previously
proposed a biphasic relation between maternal arousal and motor activity based on other data collected from these participants
(DiPietro et al., 2003), such that transient suppression in response to acute stress is followed by rebound in motor activity.
Although the unit of analysis in this study is small (i.e., minutes),
the data are aggregated over a longer period. Thus we expected
that evidence of maternal sympathetic activation would be associated with increased fetal motor activity and faster fetal heart
rate. Because maternal psychological factors have been shown to
yield both immediate and prolonged effects on fetal behavior, we
evaluated the degree to which both induced maternal arousal and
chronic perception of stress, anxiety, and depression may modify
any observed temporal relation.
Methods
Participants
Eligibility was restricted to nonsmoking women with uncomplicated pregnancies carrying singleton fetuses. Accurate dating of
the pregnancy was required and based on early first trimester
pregnancy testing or examination and/or confirmed by ultrasound. A total of 185 self-referred pregnant women were enrolled; 48 were either prospectively or retrospectively excluded as
follows: preterm labor, preterm delivery, or both (21; 11%); gestational diabetes (6; 3%); congenital malformation (2; 1%);
fetal death in utero or nonviable delivery (2; 1%); growth
retardation or other condition of antepartum origin detected in
the newborn (6; 3%); and lack of completion of protocol due to
scheduling difficulties, moving, and so forth (12; 6%). The final
sample comprised 137 maternal–fetal pairs representing a population of healthy, relatively well-educated women (M years education 5 16.7, SD 5 2.1, range 12–20; M age 5 31.3, SD 5 4.1,
range 21–39). Fifty percent of the fetuses were female and this
was the first child for 56% of the sample.
Design and Procedure
Maternal–fetal monitoring commenced at 20 weeks gestation
and was repeated at 24, 28, 32, 36, and 38 weeks. Testing occurred at the same time of day during each visit (either 1:00 or
3:00 p.m.). Women were instructed to eat 1.5 hr prior to testing,
but not thereafter. A brief real-time ultrasound scan was conducted to determine fetal position as a consideration in locating
the Doppler transducer, measure amniotic fluid using a common
511
index, and provide photographs to parents. Fetal monitoring
proceeded for 50 min, with the mother resting comfortably in a
semirecumbent, left-lateral position. At 32 weeks, monitoring
duration was shortened to 30 min to accommodate a maternal
stress manipulation. The manipulation consisted of a 29-min labor and delivery video in which women recounted their children’s
births, interspersed with explicit delivery scenes in black and
white (Birth Stories, The Cinema Guild, NY). Women were
monitored in the same semirecumbent position and viewed the
video wearing headphones, without the presence of companions.
Individual response patterns of women and fetuses to this manipulation are presented elsewhere (DiPietro, Ghera, Costigan,
& Gurewitsch, 2004).
Fetal data. Fetal data were collected using a Toitu (MT320)
fetal actocardiograph. This monitor detects fetal movement and
fetal heart rate through the use of a single wide array transabdominal Doppler transducer and processes this signal through a
series of filtering techniques. The actograph detects fetal movements by preserving the remaining signal after bandpassing frequency components of the Doppler signal that are associated
with fetal heart rate and maternal somatic activity. Reliability
studies comparing actograph-based versus ultrasound-visualized
fetal movements have found this monitor to be highly accurate in
detecting both fetal motor activity and quiescence (Besinger &
Johnson, 1989; DiPietro, Costigan, & Pressman, 1999; Maeda,
Tatsumura, & Utsu, 1999).
Fetal data were collected from the output port of the monitor
and digitized at 1000 Hz through an internal analog-to-digital
board using streaming software. Data were analyzed off–line
using software developed in our laboratory. Digitized heart rate
data underwent error rejection procedures based on moving averages of acceptable values as needed; fetal movement data represent raw voltage values generated from the actograph, scaled
from 0 to 100 in arbitrary units. Mean values for fetal neurobehavioral measures are presented elsewhere (DiPietro, Caulfield, et al., in press).
Maternal physiological data. Maternal physiological signals
were amplified using a multichannel, electrically isolated bioamplifier. Electrocardiogram was recorded from three carbon fiber
disposable electrodes in triangulated placement (right mid subclavicle, left mid axillary thorax, and upper left thigh for ground
lead). Electrodermal activity (skin conductance) was monitored
from two silver–silver chloride electrodes with a gelled skin contact area placed on the distal phalanxes of the first and index
fingers of the nondominant hand affixed with adhesive collars to
limit gel contact to a 1 cm diameter circle and velcro. Maternal
data were time synchronized and analyzed in conjunction with
fetal data. ECG data underwent R-wave detection, manual editing for artifact, and interbeat interval computation. To maintain consistency with fetal measures, interbeat interval values
were prorated to second-by-second heart rate in beats per minute
(bpm). Skin conductance was measured by administering a constant 0.5-V root-mean-square 30-Hz AC excitation signal and
detecting the current flow. Skin conductance level was scaled
from 0 to 25 mS and detrended to remove the mean, thereby
amplifying the signal-to-noise ratio ( 2.5 to 12.5 mS).
Maternal psychosocial data. Five validated self-report scales
were administered at either one or two gestational ages. The
schedule of administration was selected to minimize participant
burden. Perceived daily stress was recorded at 24 and 36 weeks
512
(Brantley, Waggoner, Jones, & Rappaport, 1987), state and trait
anxiety at 28 and 38 weeks (Spielberger, 1983), and depressive
symptoms at 32 weeks (Radloff, 1977). Pregnancy-specific stress
was measured at 32 and 38 weeks using a scale validated, in part,
on this sample of participants (DiPietro, Ghera, Costigan, &
Hawkins, in press).
Data Analysis
The data were reduced by resampling at 1 Hz producing time
series measurements (X1, Y1), (X2, Y2), y, (XT, YT ) for each
pair of measures. Fifty minutes of data, sampled once per second,
generates a time series of 3000 data points (1800 at 32 weeks).
Each pair of values can be considered outcomes of a two-component stationary time series. A standard way of measuring the
association between two components in a time series is through
the cross-correlation function (Diggle, 1990) obtained by computing the correlation coefficient between each series at various
time lags. The mathematical formula is included in the Appendix.
Cross-correlations were conducted for each of four maternal–
fetal variable pairs at each lag from 100 s: maternal heart rate
with both fetal heart rate and movement, and maternal skin conductance with fetal heart rate and movement. Two descriptive
statistics were extracted. The first is the individual lag, in seconds,
at which the strongest association occurs. The second is the magnitude of the cross-correlation function at the overall peak lag.
Because this is an unfamiliar analysis to many readers, a few
comments on the interpretation of the ensuing values are in order. Peak magnitude associations should not be interpreted in the
same manner as Pearson correlations between mean values of
two variables within a cohort of individuals. Fetal movement
data are comprised of a continuous signal generated by the actograph; thus both small values representing minor movements
(including some noise) are not distinguished from large values
denoting gross motor activity. These lower amplitude signals
were not filtered out of the analysis to avoid a priori decisions
relating to movement size that were inconsistent with the exploratory nature of these analyses. Numerous sources of variation, including stochastic variability in the true lags, shape of
responses, and measurement error can make a relationship that is
perfectly paired have peak cross-correlation coefficients of values
much less than one. Although some of these issues are also germane to Pearson correlations computed between variables, their
influence is magnified in time series analyses that rely on multiple
associations between streams of data.
Weighted least squares analysis was used to model the developmental trends in peak magnitudes of each variable pair from
20 to 38 weeks. This method estimates the correlation structure
generated by the repeated measurements on the same fetus and
uses the estimate to appropriately weight the observations in the
regression analysis and produces appropriate estimates of regression parameters and their variances (Diggle, Liang, & Zeger,
1994). Although there were few instances of missing data due to
technical problems or noncompliance at any visit prior to the
final one (two instances of missing data at 28 weeks, one at 36
weeks), a substantial proportion (33%) of women who delivered
at term did so prior to their scheduled 38 week recording. However, cases with missing values are not excluded from weighted
least squares analysis. Fetal sex and parity (nulliparous vs. multiparous) effects were examined by comparing both peak magnitudes and lags by t tests at each gestational age. Pearson
correlations were used to evaluate temporally independent asso-
J.A. DiPietro et al.
ciations between each maternal and fetal measure and withinfetal stability of the time series pairs over gestation.
Changes in response to the stress manipulation for maternal
heart rate, skin conductance, and peak magnitudes of the maternal–fetal relations were evaluated using repeated-measures
analysis of variance comparing baseline values to those generated
during the video. Two series of Pearson correlation analyses were
conducted to examine associations between self-report psychosocial measures and time series peak magnitudes. In the first,
correlations were computed within specific gestational ages (e.g.,
24 week stress with 24 week maternal heart rate/fetal heart rate
peak magnitude). The second set of analyses was based on mean
values of each time series pair averaged across gestation and
mean values of each psychosocial measure. Given the large
number of individual correlations, a Bonferroni correction was
applied and only correlations that achieved significance of o.01
were considered significant.
Results
Figures 1 and 2 illustrate results from the time series analysis for
each fetal measure with maternal heart rate at the six gestational
periods evaluated. Plots reflect mean peak magnitudes (darkest
shaded areas) at each lag ranging from 40 to 140s for display
purposes although analyses included values through 100 s.
Lighter shadings indicate interquartile ranges. Examination of
Figure 1 indicates no systematic relations between maternal and
fetal heart rate, but Figure 2 depicts consistent associations between maternal heart rate and fetal movement beginning at 20
weeks gestation. The peak lag is stable at 2 s throughout gestation. Positive lag values generated by the time series analyses
indicate that the fetal variable precedes the maternal variable;
thus the strongest association is found 2 s after a signal indicating
fetal movement. Because it is difficult to ascertain the mean values of the peaks from visual inspection of Figures 1 and 2, Figure 3
provides means and 99% confidence intervals for each over time.
Peak magnitudes are significantly different from 0 for maternal
heart rate/fetal movement at each gestational age, but it is clear
that the maternal heart rate/fetal heart rate values hover around 0.
No significant linear trends over gestation were detected.
Similar results were found between fetal measures and maternal skin conductance; these are depicted in Figures 4 and 5. There
are no systematic relations between maternal skin conductance
and fetal heart rate, but consistent associations were detected with
fetal movement beginning at 20 weeks gestation. Peak lags were
stable at 3 s; again, a positive lag indicates that the fetal movement
preceded changes in maternal skin conductance. Peak magnitude
data presented in Figure 6 include maternal skin conductance/
fetal movement values that are significantly different from 0 but
maternal skin conductance/fetal heart rate values are near 0. No
significant linear trends over gestation were detected. Analyses
involving skin conductance were conducted with and without
African-American participants (n 5 17) because race has a wellknown suppressive effect on skin conductance reactivity (Vrana &
Rollock, 2002). Exclusion of this subsample made little difference
to results. African-American women had significantly lower peak
values for skin conductance/fetal movement, F(1,119) 5 5.04,
po.05, but not skin conductance/fetal heart rate.
Maternal Parity and Fetal Sex
Peak magnitude or lags did not differ by fetal sex or maternal
parity at any gestational age.
513
Maternal–fetal psychophysiology
Week 20
Week 24
0.10
Cross−correlation
Cross−correlation
0.10
0.05
0.00
−0.05
−20
0
20
−40
40
−20
0
Lag (in seconds)
Lag (in seconds)
Week 28
Week 32
0.10
20
40
20
40
20
40
0.10
Cross−correlation
Cross−correlation
0.00
−0.05
−40
0.05
0.00
−0.05
0.05
0.00
−0.05
−40
−20
0
20
−40
40
−20
0
Lag (in seconds)
Lag (in seconds)
Week 36
Week 38
0.10
Cross−correlation
0.10
Cross−correlation
0.05
0.05
0.00
−0.05
0.05
0.00
−0.05
−40
−20
0
20
40
Lag (in seconds)
−40
−20
0
Lag (in seconds)
Figure 1. Times series analysis results showing no association between maternal heart rate and fetal heart rate at six gestational ages.
Intrafetal Stability
Evident in Figures 3 and 6 is the degree of variability among
maternal–fetal pairs in peak magnitude of maternal heart rate/
fetal movement and maternal skin conductance/fetal movement.
Peak values ranged from .11 to .34 for the former and .10
to .31 for the latter. In general, individuals with higher associ-
ations on one variable had higher associations on the other, rs at
each gestational age 5 .62, .66, .52, .62, .59, and .43 at each
gestational age from 20 to 38 weeks, respectively (pso.001).
Correlations computed between peak magnitudes for each of
these maternal–fetal variable pairs over gestation are presented in
Table 1.
514
J.A. DiPietro et al.
Week 20
Week 24
0.10
Cross−correlation
Cross−correlation
0.10
0.05
0.00
−0.05
−20
2
0
20
−40
40
−20
2
0
Lag (in seconds)
Lag (in seconds)
Week 28
Week 32
0.10
20
40
20
40
20
40
0.10
Cross−correlation
Cross−correlation
0.00
−0.05
−40
0.05
0.00
−0.05
0.05
0.00
−0.05
−40
−20
2
0
20
−40
40
−20
2
0
Lag (in seconds)
Lag (in seconds)
Week 36
Week 38
0.10
Cross−correlation
0.10
Cross−correlation
0.05
0.05
0.00
0.05
0.00
−0.05
−0.05
−40
−20
2
0
20
40
Lag (in seconds)
−40
−20
2
0
Lag (in seconds)
Figure 2. Time series analysis results showing a consistent relation between maternal heart rate and fetal movement from 20 to 38
weeks gestation. Note peak lag time of 2 s.
Post Hoc Analyses
Two correlational analyses were conducted to determine whether
local uterine factors (i.e., amniotic fluid and fetal size) were associated with peak magnitudes for either maternal heart rate/
fetal movement or maternal skin conductance/fetal movement.
Fetal size, approximated by using weight at birth, was unrelated
to these values. Amniotic fluid volume, measured at each gestational age, was also unrelated.
Maternal–fetal psychophysiology
Figure 3. Cross-correlation (peak magnitude) function values for
maternal heart rate with fetal heart rate and fetal movement. Vertical
lines indicate 99% confidence intervals.
Bivariate Maternal–Fetal Associations
Table 2 presents the correlations between maternal and fetal
variables averaged over the entire recording at each gestational
age. Significant associations emerged between maternal and fetal
heart rate beginning at 28 weeks and maternal heart rate and fetal
movement beginning at 32 weeks. No significant associations
were found between maternal skin conductance and fetal heart
rate. Positive associations were detected with fetal movement,
although these were inconsistent over gestation.
Response to Stress Manipulation
Women responded to the video with a decrease in heart rate (M
baseline 5 88.2 bpm; M video 5 86.2 bpm; F(1,135) 5 65.06,
po.0001) and increase in skin conductance (M baseline 7.3 mS; M
video 5 8.36 mS; F(1,131) 5 64.22, po.0001), indicating both attentional and sympathetic activation responses. Maternal parity
was unrelated to responsiveness. None of the four maternal–fetal
synchrony measures changed in response to the manipulation.
Psychosocial Measures
Of all the associations evaluated, only one significant, modest correlation emerged between maternal anxiety at 38 weeks and the
degree of association between skin conductance and fetal movement at that time, r(87) 5 .30, po.01. Of the remaining correlations, the majority (79%) were less than r .10 in magnitude.
Discussion
These results illustrate how little is known about psychophysiology of the earliest relationship. No time-dependent associations
515
between fetal heart rate and either maternal heart rate or skin
conductance were detected; thus we conclude that fast-acting
aspects of maternal sympathetic activation do not directly affect
fetal heart rate. These data are consistent with findings of relative
noncorrespondence of fetal heart rate to induced maternal heart
rate and blood pressure changes (Monk et al., 2000). However,
we temper this conclusion with a caveat. Fetal heart rate was
measured using Doppler technology, which quantifies heart rate
by detecting motions of the fetal heart. Commercially available
fetal electrocardiography devices function only with the use of
scalp electrodes applied after rupture of membranes; existing experimental systems that extract fetal from maternal ECG are
challenged by a period of increased electrical isolation of the fetus
during the middle period of the third trimester (Oostendorp, van
Oosterom, & Jongsma, 1989). However, although the sampling
rate of fetal monitors is necessarily lower than that used to
quantify interbeat intervals, the autocorrelation techniques underlying modern Doppler detection of heart rate are quite accurate (Dawes, 1993).
In contrast to the lack of significant time-based associations,
beginning at 32 weeks gestation, women with higher overall heart
rates had fetuses with higher heart rates and greater motor activity. Although there is inconsistency in the literature regarding
whether maternal and fetal heart rates are correlated, the magnitude of the correlations we detected were below those reported
following 24-hr recordings (Patrick et al., 1982) but significant
despite reports of no relations by others (Lewis et al., 1970).
Despite its apparent sensibility, this remains a phenomenon that
is difficult to explain (Patrick et al., 1982). Because we were
unable to document time-dependent associations, such correlations must be explained through characteristics of the uterine
milieu that were not measured in this study, including genetic,
metabolic, endocrine, or other unidentified environmental factors. For example, cortisol levels are known to be correlated
between mother and fetus (Gitau, Cameron, Fisk, & Glover,
1998), and thus to the extent that activation of the maternal
sympathetic pathways would involve changes in cortisol level,
there may be coactivation of the fetal sympathetic pathway by
the same mechanism. An alternative explanation may be that the
fetal heart rate entrains to the maternal heart rate through
acoustic sensory channels. This proposal is consistent with the
developmental change in the size of this association, because fetal
auditory capacities continues to develop over the third trimester
(Querleu, Renard, Boutteville, & Crepin, 1989).
Unlike fetal heart rate, fetal movement was temporally associated with both maternal heart rate and skin conductance. The
direction of this effect was in direct contrast to expectations, in
that the fetus precipitated a maternal response. Although maternal heart rate is subject to dual autonomic innervation, electrodermal activity is singly innervated by the sympathetic branch
(Venables, 1991). The detected lag times indicate that maternal
heart rate responds maximally after 2 s and skin conductance
responds maximally after 3 s. Both latencies are consistent with
expectations based on observations of elicited physiological responses (Dawson, Schell, & Filion, 2000; Stern, Ray, & Quigley,
2001; Turpin, 1983; Venables, 1991). The magnitudes of the
cross-correlations between maternal and fetal measures were
consistent over gestation, but relatively small in magnitude.
As previously discussed, the time series analysis employed is
more likely to underestimate associations. To put these findings
in some context, a report of the well-known, robust relation between fetal movement and fetal heart rate using similar time
516
J.A. DiPietro et al.
Week 20
Week 24
0.10
Cross−correlation
Cross−correlation
0.10
0.05
0.00
−40
−20
0
20
−40
40
−20
0
Lag (in seconds)
Lag (in seconds)
Week 28
Week 32
20
40
20
40
20
40
0.10
Cross−correlation
0.10
Cross−correlation
0.00
−0.05
−0.05
0.05
0.00
0.05
0.00
−0.05
−0.05
−40
−20
0
20
−40
40
−20
0
Lag (in seconds)
Lag (in seconds)
Week 36
Week 38
0.10
Cross−correlation
0.10
Cross−correlation
0.05
0.05
0.00
−0.05
0.05
0.00
−0.05
−40
−20
0
20
40
Lag (in seconds)
−40
−20
0
Lag (in seconds)
Figure 4. Time series analysis results showing no association between maternal skin conductance and fetal heart rate at six
gestational ages.
series techniques established the strength of that relation to
range from approximately .10 to .25 over the same gestational period (DiPietro, Irizarry, Hawkins, Costigan, &
Pressman, 2001).
Because of this unexpected result, a number of confirmatory
post hoc manipulations were conducted, including reversing the
order of the sequencing of variables entered into the time series
and randomly examining individual movements in relation to
517
Maternal–fetal psychophysiology
Week 20
Week 24
0.10
Cross−correlation
Cross−correlation
0.10
0.05
0.00
−0.05
0.05
0.00
−0.05
3
−40
−20
0
3
20
−40
40
0
Lag (in seconds)
Lag (in seconds)
Week 28
Week 32
0.05
0.00
−0.05
40
20
40
20
40
0.05
0.00
−0.05
3
−40
−20
0
3
20
−40
40
−20
0
Lag (in seconds)
Lag (in seconds)
Week 36
Week 38
0.10
Cross−correlation
0.10
Cross−correlation
20
0.10
Cross−correlation
0.10
Cross−correlation
−20
0.05
0.00
−0.05
0.05
0.00
−0.05
3
−40
−20
0
3
20
40
Lag (in seconds)
−40
−20
0
Lag (in seconds)
Figure 5. Time series analysis results showing a consistent relation between maternal skin conductance and fetal movement from 20
to 38 weeks gestation. Note peak lag time of 3 s.
maternal measures. All supported the temporal linkage of the
findings. The magnitude of the cross-correlations did not change
from mid- to late gestation, although this does not rule out the
possibility that a developmental trend might have been detected
had monitoring commenced earlier. However, the emergence of
fetal behavioral states begins within the period of study and maturity of state patterns is evident at term (Nijhuis & van de Pas,
1992). The lack of developmental trends in maternal–fetal
518
J.A. DiPietro et al.
Figure 6. Cross-correlation (peak magnitude) function values for
maternal skin conductance with fetal heart rate and fetal movement.
Vertical lines indicate 99% confidence intervals.
synchrony suggests that these processes are not dependent on
fetal neuromaturation in general, and unrelated to state development in particular. It should be noted that there are multiple
influences on both fetal behavior and maternal sympathetic
arousal and the fact that shared variance between the two was
detected does not preclude the possibility of other mediating influences on either aspect.
How does fetal motor activity affect maternal functioning?
Although women perceive most large amplitude or prolonged
movements, they are not good detectors of smaller spontaneous
or evoked fetal movements (Kisilevsky, Killen, Muir, & Low,
1991) and detect as few as 16% of all movements at term (JohnTable 1. Intrafetal Consistency in Maternal Heart Rate/Fetal
Movement and Maternal Skin Conductance/Fetal Movement Time
Series Peak Magnitude over Gestation
Gestational age (weeks)
20
24
Gestational age (weeks)
20
.50nn
24
.29nn
28
.40nn
.47nn
32
.15
.40nn
36
.34nn
.39nn
38
.24n
.35nn
28
32
36
38
.55nn
.50nn
.45nn
.41nn
.45nn
.40nn
.41nn
.42nn
.38nn
.36nn
.30nn
.45nn
.41nn
.39nn
.33nn
.45nn
.34nn
.41nn
.28nn
.44nn
Note: Values above the diagonal are for maternal heart rate/fetal movement; values below the diagonal are for maternal skin conductance/fetal
movement.
n
po.05; nnpo.01.
son, Jordan, & Paine, 1990). This suggests that the maternal
sympathetic response is evoked in the absence of perception of
movement. On average, the fetus moves approximately once per
minute during the second half of gestation (DiPeitro, Caulfield,
et al., in press; DiPietro, Costigan, Shupe, Pressman, & Johnson,
1998; Manning, Platt, & Sipos, 1979; Nasello-Paterson, Natale,
& Connors, 1988; Roberts, Griffin, Mooney, Cooper, & Campbell, 1980; Roodenburg, Wladimiroff, van Es, & Prechtl, 1991).
Thus, despite the ubiquity of the stimulus, the lack of a change in
the degree of maternal responsiveness over gestation suggests
that women neither habituate nor become sensitized to fetal
movements.
The question must be answered, at least in part, by mechanisms that do not require conscious maternal mediation. This is
further supported by our failure to detect any influence of maternal physiological arousal or psychosocial stress characteristics
on these associations. Other internal stimuli unrelated to the fetus, such as tachygastria, generate autonomic response in pregnant and nonpregnant individuals (Koch, Stern, Vasey, &
Dwyer, 1990). Perhaps the most obvious mechanism through
which fetal movement may generate an autonomic response involves the perturbations to the uterine wall. The normal response
of the uterus to distension is contraction. Contractions are
known to be stimulated by beta adrenergic pathways and modulated by noradrenergic pathways (Andersson, Ingemarsson, &
Persson, 1973). It is possible that distension of the uterine wall
caused by fetal movement may stimulate neural fibers that would
effect a contraction, but may also stimulate a noradrenergic response in the mother that would modify this to prevent excessive
contractions. Such modulation of the normal physiologic response is not a unique maternal physiologic adaptation to pregnancy. For example, the renin–angiotension–aldosterone system
is markedly enhanced throughout pregnancy, yet the effect of
angiotensin is blunted in normal pregnancies as a protective
mechanism against hypertension (Parisi & Creasey, 1992).
Although physiologic feedback systems may provide local
mechanisms that explain these associations, these findings also
have broader implications of the role of the fetus in pregnancy.
There is precedent for considering the fetus as an active agent in
arenas ranging from ontogeny (Smotherman & Robinson, 1987)
to stimulation of labor (Challis et al., 2001). We suggest that the
observed phenomenon of maternal sympathetic activation by the
fetus may serve a signal function to the pregnant woman in
preparation for the consuming early demands of child rearing
and redirecting maternal resources directed at competing but less
relevant environmental demands. Pregnancy has been associated
with a dampening of physiological activation on a variety of
levels to laboratory challenges including cognitive stressors
(Matthews & Rodin, 1992), postural changes and isometric exercise (Barron, Mujais, Zinaman, Bravo, & Lindheimer, 1986),
and cold pressor (Kammerer, Adams, von Castelberg, & Glover,
2002). This pattern of findings can be interpreted as reflecting
diminished responsivity to external stressors that may jeopardize
pregnancy. Transient deficits in cognitive performance have also
been documented (Buckwalter et al., 1999; deGroot, Adam, &
Hornstra, 2003). Although the well-known variations in sex steroids or glucocorticoids that occur during pregnancy are putative
culprits, no such linkage in human pregnancies has been established (Buckwalter et al., 1999; Keenan, Yaldoo, Stress, Fuerst,
& Ginsburg, 1998). We suggest that sympathetic activation of the
mother by the fetus is an unidentified mechanism that may
contribute to observed decrements in cognitive performance in
519
Maternal–fetal psychophysiology
Table 2. Time-Independent Maternal–Fetal Correlations at Each
Gestational Age
Gestational age (weeks)
Variable pair
Maternal heart rate/
fetal heart rate
Maternal heart rate/
fetal movement
Maternal skin conductance/
fetal heart rate
Maternal skin conductance/
fetal movement
20
.13
.03
24
.14
28
32
.17n
.26nn .21nn
.33nn
.24nn .25nn
.24nn
.05 .02
.00
.02
.00
.41nn
.13
.37nn
.05
36
38
.04
.07
.38nn .16
.03
po.05. nnpo.01.
n
pregnancy by interfering with parasympathetic processes that are
required for maintenance of attention. Moreover, periodic sympathetic surges generated by the fetus may serve to entrain maternal arousal patterns to the behavior of the fetus with implications
for the impending needs of newborn care. Among these is the
potential that the periodic declaration through movements of the
fetal presence and resultant sympathetic surge provides the rudimentary basis of the strong maternal protective response.
A seminal paper by Bell (1968) challenged the notion that
child rearing was a unidirectional phenomenon and introduced
the now well-accepted construct that the relationship between
mothers and children is bidirectional. More recently, a similar
model has been applied to understanding the effects that rodent
offspring have on the development of the maternal nervous system (Kinsley et al., 1999). The complex nature of the most fundamental human relationship raises questions regarding the
longer term implications of this phenomenon. The degree of stability within individuals in the magnitude of the fetal movement
associations with skin conductance and heart over the course of
gestation is large, particularly when considered in relation to
most other developmental phenomenon. These correlations are
comparable to or exceed those for fetal heart rate and motor
activity alone (DiPietro, Hodgson, Costigan, & Johnson, 1996),
suggesting that synchrony is a stable characteristic of individual
maternal–fetal pairs.
A distal, but intriguing question is whether maternal–fetal
synchrony sets the stage for postnatal synchrony in maternal–
child interaction. Are women who are more physiologically responsive to fetal movements more responsive to infant behavior?
Does success at fetal stimulation of a maternal response translate
to an infant’s success at serving an elicitor of care giving? Efforts
to address these questions are currently underway in our laboratory as we follow these maternal–child pairs through the first
years of life and examine the continuing intricacies of the earliest
relationship.
REFERENCES
Andersson, K., Ingemarsson, I., & Persson, C. (1973). Relaxing effects of
beta-receptor stimulators in isolated, gravid human myometrium.
Life Sciences, 13, 335–344.
Artal, R., Rutherford, S., Romem, Y., Kammula, R., Dorey, F.,
& Wiswell, R. (1986). Fetal heart rate responses to maternal
exercise. American Journal of Obstetrics and Gynecology, 155,
729–733.
Barron, W., Mujais, S., Zinaman, M., Bravo, E., & Lindheimer, M.
(1986). Plasma catecholamine responses to physiologic stimuli in
normal human pregnancy. American Journal of Obstetrics and Gynecology, 154, 80–84.
Bell, R. Q. (1968). A reinterpretation of the direction of effects in studies
socialization. Psychological Review, 75, 81–95.
Besinger, R. E., & Johnson, T. R. B. (1989). Doppler recordings of fetal
movement: Clinical correlation with real-time ultrasound. Obstetrics
and Gynecology, 74, 277–280.
Brantley, P. J., Waggoner, C. D., Jones, G. N., & Rappaport, N. B.
(1987). A daily stress inventory: Development, reliability, and validity. Journal of Behavioral Medicine, 10, 61–73.
Buckwalter, J., Stanczyk, F., McCleary, C., Bluestein, B., Buckwalter,
D., & Randin, K. (1999). Pregnancy, the postpartum, and steroid
hormones: Effects on cognition and mood. Psychoneuroendocrinology, 24, 69–84.
Challis, J., Lye, S., Gibb, W., Whittle, W., Patel, F., & Alfaidy, N.
(2001). Understanding preterm labor. Annals of the New York Academy of Sciences, 943, 225–234.
Dalton, K., Dawes, G. S., & Patrick, J. E. (1983). The autonomic nervous system and fetal heart rate variability. American Journal of Obstetrics & Gynecology, 146, 456–462.
Dawes, G. S. (1993). The fetal ECG: Accuracy of measurements. British
Journal of Obstetrics and Gynaecology, 100(Supp. 9), 15–17.
Dawson, M., Schell, A., & Filion, D. (2000). The electrodermal system.
In J. Cacioppo, L. Tassinary, & G. Berntson (Eds.), Handbook of
psychophysiology (2nd ed., pp. 200–223). Cambridge, UK: Cambridge University Press.
deGroot, R. H., Adam, J. J., & Hornstra, G. (2003). Selective attention
deficits during human pregnancy. Neuroscience Letters, 340, 21–24.
Diggle, P. (1990). Time series: A biostatistical introduction. Oxford, UK:
Oxford University Press.
Diggle, P., Liang, K., & Zeger, S. (1994). Analysis of longitudinal data.
New York: Oxford University Press.
DiPietro, J. A., Caulfield, L. E., Costigan, K., Merialdi, M., Nguyen, R.,
Zavaleta, N., & Gurewitsch, E. (in press). Fetal neurobehavioral development: A tale of two cities. Developmental Psychology.
DiPietro, J. A., Ghera, M., Costigan, K. A., & Gurewitsch, E. D. (2004).
Maternal psychological arousal generates a fetal response. Paper presented at the International Conference on Infant Studies, Chicago, IL.
DiPietro, J. A., Costigan, K. A., & Gurewitsch, E. (2003). Fetal response
to induced maternal stress. Early Human Development, 74, 125–138.
DiPietro, J. A., Costigan, K. A., & Pressman, E. K. (1999). Fetal movement detection: Comparison of the Toitu actograph with ultrasound
from 20 weeks gestation. Journal of Maternal–fetal Medicine, 8, 237–
242.
DiPietro, J. A., Costigan, K. A., Shupe, A. K., Pressman, E. K., &
Johnson, T. R. B. (1998). Fetal neurobehavioral development: Associations with socioeconomic class and fetal sex. Developmental
Psychobiology, 33, 79–91.
DiPietro, J. A., Ghera, M., Costigan, K., & Hawkins, M. (in press).
Measuring the ups and downs of pregnancy stress. Journal of Psychosomatic Obstetrics and Gynaecology.
DiPietro, J. A., Hilton, S., Hawkins, M., Costigan, K., & Pressman, E.
(2002). Maternal stress and affect influence fetal neurobehavioral development. Developmental Psychology, 38, 659–668.
DiPietro, J. A., Hodgson, D. M., Costigan, K. A., Hilton, S. C., &
Johnson, T. R. B. (1996). Fetal neurobehavioral development. Child
Development, 67, 2553–2567.
DiPietro, J. A., Hodgson, D. M., Costigan, K. A., & Johnson, T. R. B.
(1996). Fetal antecedents of infant temperament. Child Development,
67, 2568–2583.
DiPietro, J. A., Irizarry, R., Hawkins, M., Costigan, K., & Pressman, E.
(2001). Cross-correlation of fetal cardiac and somatic activity as an
indicator of antenatal neural development. American Journal of Obstetrics and Gynecology, 185, 1421–1428.
Gitau, R., Cameron, A., Fisk, N., & Glover, V. (1998). Fetal exposure to
maternal cortisol. Lancet, 352, 707–708.
Hoppenbrouwers, T., Combs, D., Ugartechea, J. C., Hodgman, J.,
Sterman, M. B., & Harper, R. M. (1981). Fetal heart rates during maternal wakefulness and sleep. Obstetrics and Gynecology, 57, 301–309.
520
J.A. DiPietro et al.
Ianniruberto, A., & Tajani, E. (1981). Ultrasonographic study of fetal
movement. Seminars in Perinatology, 5, 175–181.
Johnson, T. R. B., Jordan, E. T., & Paine, L. L. (1990). Doppler recordings of fetal movement: II. Comparison with maternal perception. Obstetrics & Gynecology, 76, 42–43.
Kammerer, M., Adams, D., von Castelberg, B., & Glover, V. (2002).
Pregnant women become insensitive to cold stress. BMC Pregnancy
and Childbirth, 2, 8.
Keenan, P., Yaldoo, D., Stress, M., Fuerst, D., & Ginsburg, K. (1998).
Explicit memory in pregnant women. American Journal of Obstetrics
and Gynecology, 179, 731–737.
Kinsley, C., Madonia, L., Gifford, G., Tureski, K., Griffin, G., & Lowry,
C., et al. (1999). Motherhood improves learning and memory. Nature,
402, 137–138.
Kisilevsky, B. S., Killen, H., Muir, D. W., & Low, J. A. (1991). Maternal
and ultrasound measurements of elicited fetal movements: A methodologic consideration. Obstetrics and Gynecology, 77, 889–892.
Koch, K., Stern, R., Vasey, M., & Dwyer, A. (1990). Gastric
dysrhythmias and nausea of pregnancy. Digestive Diseases and Sciences, 35, 961–968.
Lewis, M., Wilson, C., Ban, P., & Baumel, M. (1970). An exploratory
study of resting cardiac rate and variability from the last trimester of
prenatal life through the first year of postnatal life. Child Development, 41, 799–811.
Macphail, A., Davies, G., Victory, R., & Wolfe, L. (2000). Maximal
exercise testing in late gestation. Obstetrics & Gynecology, 96, 565–570.
Maeda, K., Tatsumura, M., & Utsu, M. (1999). Analysis of fetal movements by Doppler actocardiogram and fetal B-mode imaging. Clinics
in Perinatology, 26, 829–851.
Manders, M., Sonder, G., Mulder, E., & Visser, G. (1997). The effects of
maternal exercise on fetal heart rate and movement patterns. Early
Human Development, 48, 237–247.
Manning, F. A., Platt, L. D., & Sipos, L. (1979). Fetal movements in
human pregnancies in the third trimester. Obstetrics and Gynecology,
54, 699–702.
Martin, C. (1978). Regulation of the fetal heart rate and genesis of FHR
patterns. Seminars in Perinatology, 2, 131–146.
Matthews, K. A., & Rodin, J. (1992). Pregnancy alters blood pressure
responses to psychological and physical challenge. Psychophysiology,
29, 232–240.
Monk, C., Fifer, W., Myers, M., Sloan, R., Trien, L., & Hurtado, A.
(2000). Maternal stress responses and anxiety during pregnancy: Effects on fetal heart rate. Developmental Psychobiology, 36, 67–77.
Nasello-Paterson, C., Natale, R., & Connors, G. (1988). Ultrasonic
evaluation of fetal body movements over twenty-four hours in the
human fetus at twenty-four to twenty-eight weeks; gestation. American Journal of Obstetrics and Gynecology, 158, 312–316.
Nijhuis, J. G., & van de Pas, M. (1992). Behavioral states and their
ontogeny: Human studies. Seminars in Perinatology, 16, 206–210.
Oostendorp, T., van Oosterom, A., & Jongsma, H. (1989). The effect of
changes in the conductive medium on the fetal ECG throughout gestation. Clinical Physics and Physiological Measurement, 10, 11–20.
Parer, J. (1999). Fetal heart rate. In R. Creasy & R. Resnik (Eds.),
Maternal–fetal medicine (4th ed, pp. 270–299). New York: W.B.
Saunders Company.
Parisi, V., & Creasey, R. (1992). Maternal biological adaptations to
pregnancy. In E. Reece, J. Hobbins, M. Mahoney, & R. Petrie (Eds.),
Medicine of the fetus and mother (pp. 831–848). Philadelphia: JB
Lippincott.
Patrick, J., Campbell, K., Carmichael, L., & Probert, C. (1982). Influence of maternal heart rate and gross fetal body movements on the
daily pattern of fetal heart rate near term. American Journal of
Obstertics and Gynecology, 144, 533–538.
Querleu, D., Renard, X., Boutteville, C., & Crepin, G. (1989). Hearing
by the human fetus? Seminars in Perinatology, 13, 409–420.
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for
research in the general population. Applied Psychological Measurement, 1, 385–401.
Roberts, A. B., Griffin, D., Mooney, R., Cooper, D. J., & Campbell, S.
(1980). Fetal activity in 100 normal third trimester pregnancies. British Journal of Obstetrics and Gynaecology, 87, 480–484.
Roodenburg, P. J., Wladimiroff, J. W., van Es, A., & Prechtl, H. F. R.
(1991). Classification and quantitative aspects of fetal movements
during the second half of normal pregnancy. Early Human Development, 25, 19–35.
Smotherman, W. P., & Robinson, S. R. (1987). Prenatal influences on
development: Behavior is not a trivial aspect of fetal life. Development
and Behavioral Pediatrics, 8, 171–175.
Sontag, L. W. (1941). The significance of fetal environmental differences.
American Journal of Obstetrics and Gynecology, 42, 996–1003.
Sontag, L. W., & Richards, T. W. (1938). Studies in fetal behavior: I.
Fetal heart rate as a behavioral indicator. Monographs of the Society
for Research in Child Development, 3(4, Serial No. 17), 1–67.
Spielberger, C. (1983). Manual for the State-Trait Anxiety Inventory
(Form Y). Palo Alto, CA: Mind Garden, Inc.
Stern, R., Ray, W., & Quigley, K. (2001). Psychophysiological recording
((2nd ed)). Oxford, UK: Oxford University Press.
Turpin, G. (1983). Unconditioned reflexes and the autonomic nervous
system. In D. Siddle (Ed.), Orienting and habituation: Perspectives in
human research (pp. 1–70). New York: John Wiley & Sons.
Van den Bergh, B. R. H., Mulder, E. J. H., Visser, G. H. A., PoelmannWeesjes, G., Bekedam, D. J., & Prechtl, H. F. R. (1989). The effect of
(induced) maternal emotions on fetal behaviour: A controlled study.
Early Human Development, 19, 9–19.
Venables, P. (1991). Autonomic activity. Annals of the New York Academy of Sciences, 620, 191–207.
Vrana, S., & Rollock, D. (2002). The role of ethnicity, gender, emotional
content, and contextual differences in physiological, expressive, and
self-reported emotional responses to imagery. Cognition and Emotion,
16, 165–192.
Yoshizato, T., Koyanagi, T., Takashima, T., Satoh, S., Akazawa, K., &
Nakano, H. (1994). The relationship between age-related heart rate
changes and developing brain function: A model of anencephalic human fetuses in utero. Early Human Development, 36, 101–112.
(Received April 9, 2003; Accepted November 5, 2003)
APPENDIX
Assuming that the measurements are the outcome of a stationary
stochastic process, we define the cross-correlation function with
rðhÞ ¼
E fðXt mX ÞðYtþh mY Þg
; h ¼ 0; 1; 2; . . .
sX sY
with mX 5 E(Xt) and mY 5 E(Yt) the expected value of
X1,y,XT and Y1,y,YT, respectively and s2X and s2Y the variance of X1,y,XT and Y1,y,YT, respectively. With the assumption of stationarity, none of these quantities depend on time
t. If fetal motor activity tends to increase whenever there is a
change in maternal heart rate, we expect r(0) (the common definition of correlation between X1,y,XT and Y1,y,YT is the
cross-correlation at lag (0) to be relatively high. However, if fetal
motor activity tends to increase some time h after maternal heart
rate change occurs, then it is r(h) that we expect to be high.
Notice that r(h) has an intuitive definition as the correlation
between X1,y,XT h and Yh11,y,YT. The sample cross-correlation function is
^ ðhÞ ¼
r
Th
1 P ðX X ÞðY Y Þ
tþh
T h t¼1 t
^Y
^X s
s
with X and Y the sample averages of X1,y,XT and Y1,y,YT,
^2X and s
^2Y the sample variances of X1,y,XT,
respectively, and s
and Y1,y,YT, respectively. The sample cross-correlation is the
standard estimate of the cross-correlation function.