More frequent partner hugs and higher oxytocin
levels are linked to lower blood pressure and
heart rate in premenopausal women
Kathleen C. Lighta,*, Karen M. Grewena, Janet A. Amicob
aDepartment of Psychiatry, University of North Carolina, CB 7175 Medical Building A,
Chapel Hill, NC 27599-7175, USA
bDepartments of Pharmaceutical Science and Medicine, University of Pittsburgh, Pittsburgh, PA, USA
Accepted 25 August 2004
Available online 29 December 2004
In animals, ventral stroking for >5 days increases oxytocin (OT) activity and decreases blood
of partner hugs, plasma OT and BP levels were examined in 59 premenopausal women before and
after warm contact with their husbands/partners ending with hugs. Higher baseline OT before partner
contact was associated with lower BP and heart rate, and met criteria to be a partial mediator of the
lower resting BP shown by women reporting more frequent hugs (P < 0.05). OT levels during post-
contact stress were unrelated to hugs or BP. Menstrual cycle phase did not influence anyOT measure.
Thus, frequent hugs between spouses/partners are associated with lower BP and higher OT levels in
premenopausal women; OT-mediated reduction in central adrenergic activity and peripheral effects
of OT on the heart and vasculature are pathways to examine in future research.
# 2004 Elsevier B.V. All rights reserved.
Keywords: Oxytocin; Blood pressure; Partner support; Warm touch; Hugs
Emotional support from a spouse or long-term partner is related to lower risk of
cardiovascular and all-cause mortality (Berkman, 1995; Knox and Uvnas-Moberg, 1998;
Kiecolt-Glaser and Newton, 2001; Tower et al., 2002). Blood pressure (BP) is especially
sensitive to supportive and non-supportive interactions between partners (Ewart et al.,
Biological Psychology 69 (2005) 5–21
* Corresponding author. Tel.: +1 919 966 2544; fax: +1 919 966 0708.
E-mail address: firstname.lastname@example.org (K.C. Light).
0301-0511/$ – see front matter # 2004 Elsevier B.V. All rights reserved.
1991; Carels et al., 1998; Broadwell and Light, 1999; Gump et al., 2001; Grewen et al.,
2003; Holt-Lunstad et al., 2003). Communication of emotional support is accomplished
though multiple, complex modalities, including facial expression and body language as
convey positive emotion and connectedness while couples interact, all of which may
influence BP (Uchino et al., 1996; Denton et al., 2001; Gottman and Notarius, 2002;
Broadwell and Light, 2004). Emotional support and affection in couples is also expressed
through physical touch, such as hand-holding, hugs, and sitting or lying ‘‘cuddled up’’
(Diamond, 2000; Grewen et al., 2004a, 2004b).
A number of experts (Carter, 1998; Uvnas-Moberg, 1998, 2004; Insel and Young, 2001;
Taylor, 2002; Moyer et al., 2004) have hypothesized that enhanced oxytocin (OT) activity
of emotional support, particularly those linked towarm touch. Although best known for its
role inparturition, breast-feedingand initiationof maternal behavior,OTisahypothalamic
neuropeptide shown in animal models to be critically involved in important social
behaviors including social recognition, partner preference and, in certain species,
monogamous pair-bonding (Williams et al., 1994; Pedersen, 1997; Carter et al., 2001;
Ferguson et al., 2002; Pedersen and Boccia, 2002; Bales and Carter, 2003; Champagne
et al., 2003; Choleris et al., 2003). Furthermore, increases in endogenous OT activity are
elicited by massagelike stroking in both infant and adult mammals (Uvnas-Moberg, 1998,
of OT in rats reflect the cumulative effect of repeated episodes of stroking. In their study,
plasma OT increases were significant after 14 days but not after 3 days of such stroking.
OT has both central and peripheral actions on cardiovascular function (Petersson et al.,
1996; Gutkowska et al., 2000; Petersson, 2002). Although a single dose of OTin rats leads
to BP increases, not decreases, daily OT administration (or massagelike stroking to
enhance endogenous OT activity) for 5–14 days leads to enduring BP reductions that far
outlast the intervention (Petersson et al., 1999; Holst et al., 2002). Enhanced OTactivity of
this kind leads to inhibition of central and peripheral alpha-adrenergic and hypothalamic-
pituitary-adrenal (HPA) activity, while promoting parasympathetic cardiac control (Diaz-
Cabale et al., 2000; Janowski et al., 2000; Mukaddam-Daher et al., 2001). Co-localized
estrogen and OT receptors influence each other, such that OT activity can have greater
cardiovascular effects in cycling vs. ovariectomized females or males (Petersson et al.,
1999; Holst et al., 2002).
Despite the extensive literature on OT in animal models, relatively few published
studies on human OT responses exist to date. This paucity of findings is due in part to the
fact that, unlike animal research that can assess OT mRNA or use central administration of
OT antagonists, in humans, researchers must rely upon less direct measures to index
oxytocinergic activity: plasma levels of OT and OT-precursor peptides. These peptide
levels in humans appear to reflect both general oxytocinergic activity levels of the past
weeks, such as increases in OT intermediate peptide seen in post-menopausal women on
estrogen replacement (Crowley et al., 1995; Amico and Hempel, 1990; Bossmar et al.,
or nursing in post-partum women) that elicits a consistent increase in plasma OT, or a
K.C. Light et al./Biological Psychology 69 (2005) 5–216
personal characteristic (such as maternal attachment for her infant, or spouse/partner
relationship quality) that is linked to higher overall plasma levels of OT. Single episodes of
massage have been shown to elicit increases in plasma OT levels in some individuals but
not consistently enough to yield reliable group effects (Turner et al., 1999; Wikstrom et al.,
2003). However, given the findings of Lund et al. (2002) on the need for many days of
massage repetitions to elicit increases in plasma OTin animals, this inconsistent effect of a
single episode of massage on human plasma OT levels is not surprising.
In an initial study of effects of warm contact with loved ones on OT, Light et al. (2000)
studied OTand BP responses before and during a speech task in 24 mothers of infants on 2
days, once after holding their babies and once after a control rest alone. BP levels were
lower on both test days before, during and after the speech task in mothers whose OTlevels
increased versus decreased over baseline levels in samples obtained 5 min after baby
holding (during the task). The OTincrease group did not show a reliable OTincrease to the
speech task when tested without their babies, so the prior warm contact was critical in
eliciting an OT response. An extension of this work comparing responses of these healthy
mothers with mothers who had been exposed to cocaine during pregnancy (Light et al.,
2004a) confirmed that the cocaine exposed group had lower OT levels and higher BP and
norepinephrine (NE) levels both in the lab on the no baby contact day and during
ambulatory monitoring at home. The cocaine exposed mothers also showed a tendency to
hold their babies less often at home. In rats, cocaine exposure during pregnancy disrupts
normal post-partum OTactivity and maternal behavior (Elliott et al., 2001). These findings
link greater mother–infant warm contact time to higher OTactivity and lower BP in post-
Next, we examined OTresponse and resting BP in 38 couples before, during and after a
10 min period of warm partner contact ending with a 20 s hug (Grewen et al., 2004a,
2004b). In this study, which involved no stressors, both men and women with more
supportive partners showed higher levels of plasma OT before, during and after the warm
contact period. Women (but not men) with greater partner support showed lower baseline
SBP and plasma NE levels, and OT met criteria as significant mediator of the reduced NE
a pilot study involving 11 subjects retested four different times, during both rest and stress
sessions, both with and without partner contact. OT responses were higher during rest vs.
stress sessions with warm partner contact.
Thus, in the present protocol where partner contact preceded a speech stressor, we
elected to focus on plasma OT levels of premenopausal women during baseline prior to
warm partner contact rather than on OT responses during the stressor, and to determine
whether women with higher OT levels had lower BP and HR: (1) during a resting baseline
period just prior to warm partner contact, and (2) during a speech task that followed warm
contact. We further examined whether greater self-reported frequency of one type of warm
touch between partners, Partner Hugs, predicted lower BP and HR during baseline and/or
stress events. Finally we attempted to test whether OTactivity (indexed by plasma levels)
may be a partial mediator of the expected relationships between greater Partner Hugs and
lower cardiovascular responses. As a methodological issue, we also examined whether
plasma OT levels differ by menstrual cycle phase, to determine if cycle phase must be
controlled in this and future research on OT.
K.C. Light et al./Biological Psychology 69 (2005) 5–217
Premenopausal women aged 20–49-years old (n = 59) were recruited using local
newspaper advertisements and fliers. Subjects were required to be living with current
spouse or monogamous partner for at least 6 months; these partners participated with the
women in our study by providing them with 10 min of warm physical and emotional
contact during the testing session. Reasons for exclusion included current use of
prescription medication affecting the cardiovascular and autonomic nervous system,
chronic systemic disease, current clinical depression or other psychiatric disorder,
pregnancy, breast-feeding, post-menopausal status, or being less than 11 months post-
partum. The protocol was approved by the local IRB and all subjects and their partners
signed approved consent forms before participating.
Women were subgrouped into three groups based on the magnitude of their OT levels at
baseline. Those in the top, middle and bottom tertiles made up the high, moderate and low
OT groups, respectively. Table 1 gives demographic information for these groups. The low
OT group showed weak trends toward younger age, higher BMI, and greater percentage on
non-white members compared to the high OT group, but these differences were not
significant. Surprisingly, a significantly greater percentage of this low OT group were
married versus living with long-term partners compared to the other groups (x2(2) = 6.86,
P < 0.033). Although the low OT group included more women who had never been
pregnant or borne a child (Nulliparous: 42% versus 13% and 20%), this difference was not
significant (x2(2) = 4.33, P < 0.20).
2.1. Test session procedures
Subjectswere firstscreenedduringabrief telephoneinterview.Partnersarrivedtogether
but were immediately separated. Each woman was instrumented with the Accutracker II
BP monitor and an intravenous (i.v.) catheter for blood sampling. A 20 min period for
instrumentation/adaptation was followed by solitary resting baseline (10 min). Subjects
then joined their partners in a different room for the warm contact period (10 min), and
were separatedagainforthe stressor events(10 min). Baseline period:As described above,
women were seated aloneincomfortable chairsinaroom separated fromtheirpartners.BP
and HR data sampled at 4, 6 and 8 min was averaged to represent baseline levels. Blood
was drawn starting at 8 min of baseline for OT. Warm contact period: Couples were seated
on a loveseat in a quiet room and instructed to sit close together, holding hands if they felt
comfortable doingso.Theywereasked totalkabout atimetheyhadspenttogetherthathad
made them feel closer as a couple (2 min). Next they watched a 5 min segment of a
K.C. Light et al./Biological Psychology 69 (2005) 5–218
Demographic data for high, moderate and low OT groups
OT groupN AgeBMI Non-white (%) Married (%) Nulliparous (%)
30.5 ? 2.4
29.1 ? 1.3
27.5 ? 1.1
25.3 ? 1.7
27.3 ? 1.4
31.2 ? 2.3
*P < 0.05.
romantic video they had previously seen. They then were instructed to talk again for 2 min
about a time during which they felt especially close as a couple. During this time couples
were left alone for privacy, unmonitored and unobserved except when the experimenter
entered the room to give instructions. At the end of this session partners stood for a 20 s
hug. Post-contact stressor: Women were immediately separated from their partners to
undergo stressor testing modeled after the methods of Light et al. (2000). Briefly, the
stressor first involved 2 min of task instructions, followed by three components: (1) silent
speech preparation (2 min), (2) actively giving a tape-recorded speech about a recent
interpersonal event (one not involving their partners) that made the woman feel angry or
stressed (3 min), and (3) post-speech recovery while listening to a replay of their own tape-
recorded speech (3 min). BP and HR were measured once during preparation, twice during
active speech and twice during recovery, with 1 min intervals separating each reading; the
two readings for each of the latter events were averaged. Additional blood samples for OT
were drawn during speech preparation, active speech and post-speech recovery.
2.2. Cardiovascular assessment
Subjects were instrumented with the Accutracker II ambulatory BP monitor (Suntech,
Raleigh, NC), a device whose prototype has been validated against direct arterial and
standard auscultatory measurements (Light et al., 1988). In order to standardize the
Accutracker readings to clinic BP assessments, a minimum of three seated BP readings
were then taken with the Accutracker monitor, while simultaneous auscultatory BP
readings were assessed by a trained technician. Monitor readings for systolic (SBP) and
diastolic (DBP) BP falling within 5 mmHg of the stethoscopic values were considered
acceptable, providedthe Accutrackerdisplayednoerrorcodes.Heartrate(HR)levelswere
determined from the Accutracker ECG leads and mean arterial pressure (MAP) was
calculated automatically by the formula (SBP ? DBP)/3 + DBP.
2.3. Plasma OT measures
Blood was sampled four times for determination of plasma OT levels: (1) immediately
before the end of the 10 min solitary baseline, just prior to the period of warm partner
contact, (2) at the midpoint of speech preparation, (3) at the midpoint of the active speech,
and (4) during speech recovery. The level of OT in EDTA plasma was determined by
extraction and radioimmunoassay in the Laboratory of Janet Amico (Amico et al., 1981;
Amico and Hempel, 1990). The intra-assay coefficient of variation was 10–12% and the
low limit of sensitivity was 0.5 pg/ml; plasma OT values of these women ranged from 0.5
to 4.8 pg/ml. Preliminary analyses determined that OT levels were not normally
distributed, and thus, in all subsequent analyses, OT measures first underwent square-root
transformation to achieve greater normality of distribution.
For analyses in which OT groups were compared, the transformed baseline OT cut-
points defining the three OT groups were: low OT group < 0.89 (mean = 0.79), high group
OT > 1.09 (mean = 1.23), and moderate OT group = 0.89–1.09 pg/ml (mean = 0.98)
(n = 19,15and25,respectively;groupsizesdifferfromtruetertilesplitofalln = 19and20
owing to ties at cut-points). Although defined based on their OT levels at baseline,
K.C. Light et al./Biological Psychology 69 (2005) 5–219
ANCOVAwith age as a covariate confirmed that the high OTand low OT groups differed
reliably in their square-root transformed OT levels across all sampling periods except
active speech (baseline: 1.23 versus 0.79, P < 0.0001; speech preparation: 1.07 versus
0.88, P < 0.008; active speech: 1.03 versus 0.97, P > 0.50; recovery: 1.11 versus 0.91,
P < 0.006).
2.4. Relationship quality questionnaires
Physical Affection Scale (PAS) (Diamond, 2000) assessing the frequency that they
typically do the following with their partners: hold hands, sit close or lie down close
together, give each other neck or back massages or similar warm touches, hug, or kiss. The
six response choices ranged from ‘‘never or almost never’’ to ‘‘more than once a day’’.
Based on the animal literature that multiple days of ventral stroking elicits plasma OT
increases, we hypothesized that two items, the frequency of massage/warm touch and the
frequency of hugging (ventral contact) would show the strongest relationships to increased
OT activity. Because the partner contact condition of our protocol ended with a 20 s hug
(but did not include any massage component), we elected to use frequency of Partner Hugs
as the most directly relevant index of individual differences in habitual warm physical
contact between partners.
Subjects also completed the spousal version of the five-item Social Relationships Index
1992, 1996; Holt-Lunstad et al., 2003), to assess general emotional support from the
partner. Items ask the subject to rate both the positive (love and closeness) and the negative
(upsetting, mixed or conflicted feelings) emotional components of their partner
2.5. Statistical methods
Preliminary analyses were performed to examine whether plasma OT levels differed by
menstrual phase at time of testing. Women were subgrouped based on days since the onset
of their last menstrual period as follows: (1) follicular: days 1–12 (n = 11), (2) midcycle:
days 13–18 (n = 11), and (3) luteal: days 19–34 (n = 13). Oral contraceptiveusers (n = 17),
tested while actively taking their pills, were grouped separately, while the small group of
women (n = 4) reporting long and/or irregular cycles exceeding 34 days were excluded
from these comparisons. No differences in plasma OT levels were seen among the women
in the follicular, midcycle, luteal phase and oral contraceptive groups at any of the our
sampling times (main effect of menstrual phase F(3,50) = 1.17, P > 0.30 and menstrual
phase ? time period interaction F(12,150) = 0.86, P > 0.50). In fact, mean plasma OT
levels of all four groups were so similar at baseline (1.09, 0.91, 0.93 and 0.91 mg/dl for
follicular, midcycle, luteal and oral contraceptive groups, respectively), they support the
interpretation that menstrual cycle phase variations in OT levels in premenopausal women
are small and might be easily masked by other influences.
Primary analyses took two forms. First, mean SBP, DBP, MAP and HR levels during
four time periods (baseline, speech preparation, active speech and recovery) were
K.C. Light et al./Biological Psychology 69 (2005) 5–21 10
compared among the high, moderate and low OT groups in a repeated-measures
MANCOVA with age as covariate. Independent t-tests using age-adjusted means were
used to clarify which specific groups differed significantly from each other in these
dependent measures. Second, partial regression analyses adjusting for age were used
to examine relationships of Partner Hugs to OT, BP and HR measures during all four
time periods listed above. For those instances where Partner Hugs predicted both age-
adjusted OT and cardiovascular responses, mediational analyses using the approach
recommended by Baron and Kenny (1986) were used to test whether the effect of Partner
Hugs was partially mediated by OT. Unless other specified, alpha level was P < 0.05 two-
3.1. Cardiovascular responses of high, moderate and low OT groups
Cardiovascular responses of women grouped by their baseline OT levels were found to
differ as predicted, but these differences were significant only at baseline. Repeated
measures age-adjusted MANCOVAs yielded a significant interaction of OT group ? time
period for SBP and DBP (F(6,102) = 2.20 and 2.19, respectively, P ? 0.05). To clarify this
interaction, we compared group responses at each time period separately with ANCOVA,
and obtained a significant effect of OT group at baseline only for SBP and MAP
(F(2,55) = 5.35 and 3.57, P < 0.0075 and 0.035, respectively). Subsequent mean
comparisons confirmed that the low OT group had significantly higher SBP, DBP and
MAP at baseline than the high OT group (least-square comparisons among age-adjusted
means, P < 0.0023, 0.04, and 0.01, respectively; see Figs. 1–3). Also, the low OT group
had significantly higher baseline SBP than the moderate OT group (P < 0.032). During the
threestressorperiods (speech preparation, activespeechand post-speechrecovery),the OT
groups no longer differed significantly in any BP or HR measure (all P > 0.10), although
there was a tendency for the High OT group to maintain the lowest mean BP levels across
events(seeFigs.1–3).OTgroupdifferencesinbaselineHRwerealsoseen(F(2,55) = 3.73,
P < 0.033; not depicted). Similar to SBP, the low OT group had significantly higher
baseline HR than the high OT or the moderate OT groups (age-adjusted least-square
means ? S.E. = 80.3 ? 3.1 versus 68.5 ? 3.3 and 71.8 ? 2.7 beats/min, P < 0.014 and
3.2. Frequency of hugs from partner: links to OT and cardiovascular responses
As predicted, partial correlation coefficients (adjusted for age) indicated that greater
frequency of Partner Hugs and of Partner Massages were associated with higher baseline
OT level (r = +0.31 and +0.29, P ? 0.02 and 0.03, respectively); other PAS items (kissing,
hand-holding, sitting/lying close) were not reliably correlated with any OT measure.
Interestingly, greater frequency of Partner Hugs was related to having fewer children at
home (range zero to four children; r = ?0.32, P < 0.02). However, consistent with the link
between OTand maternal behavior, caring for more children at home was associated with
K.C. Light et al./Biological Psychology 69 (2005) 5–2111
higher baseline OT, after partialing out the effect associated with Partner Hugs (partial
r = +0.32, P < 0.04).
Subsequent analyses focused exclusively on greater Partner Hugs as a potential
predictor of increased OTand decreased cardiovascular responses. Frequency of hugs was
correlated moderately but did not show extensive overlap with partner support (defined by
SRI score; r = +0.49, P < 0.0001), and in this sample, unlike the women and men studied
by Grewen et al. (2004a, 2004b), higher SRI score was not significantly associated with
higher baseline OT level. Consistent with a priori hypotheses, women reporting more
frequent Partner Hugs had lower baseline BP levels (r = ?0.29, ?0.34 and ?0.33,
P ? 0.032 for SBP, DBP and MAP, respectively). Greater Partner Hugs also predicted
lower HR levels during speech preparation and active speech (r = ?0.35 and ?0.33,
P ? 0.04, respectively), but not during baseline (r = ?0.20, P > 0.20) or recovery
(r = +0.02, P > 0.90). Partner Hugs showed a marginally significant relationship to DBP
during speech preparation (r = ?0.23, P < 0.10) but did not significantly predict any other
BP measure during the speech preparation, active speech or post-speech recovery
(r = ?0.12 to ?0.17 during preparation, r = ?0.05 to ?0.10 during active speech, and
r = ?0.05 to +0.04 during recovery, respectively).
Higher baseline OT was significantly correlated with lower age-adjusted SBP, DBP,
MAP and HR at baseline (r = ?0.39, ?0.32, ?0.36 and ?0.42, P ? 0.003, 0.016, 0.006
K.C. Light et al./Biological Psychology 69 (2005) 5–2112
Fig. 1. Systolic BP levels in the high OT (open squares), moderate OT (filled triangles) and low OT groups (filled
diamonds) during baseline (Base) prior to partner contact, and the three post-contact periods: speech preparation
(Prep), active speech (Speech) and post-speech replay/recovery (Recover). Low OT group > high OT and
moderate OT groups at Base, P < 0.0023 and 0.032, respectively.
and 0.007, respectively) and at least marginally related to all cardiovascular measures
during speech preparation (r = ?0.23, ?0.27, ?0.28 and ?0.27, P ? 0.09, 0.04, 0.04 and
0.10, respectively) and to all BP measures during recovery (r = ?0.24, ?0.22, ?0.24,
P ? 0.08, 0.10 and 0.08, respectively),butunrelated to the cardiovascular measures during
active speech (r = ?0.14 to ?0.20, P > 0.13). In contrast, OT levels from the other three
time periods surrounding the stressor were consistently unrelated to any cardiovascular
measure (r = ?0.07 to ?0.21 for speech preparation OT, P > 0.10; r = +0.04 to ?0.11 for
active speech OT, P > 0.40, and r = +0.06 to ?0.08 for recovery OT, P > 0.50). Likewise,
Partner Hugs was consistently unrelated to OT levels obtained during speech preparation,
active speech or recovery (r = +0.06, +0.07 and +0.08, P > 0.50). Thus, only baseline OT
was a potential candidate as a mediator of the Partner Hugs link to lower BP.
Subsequently, a series of linear regression analyses was performed to test whether the
effect of Partner Hugs on age-adjusted baseline BP measures and on DBP and HR during
preparation might be mediated by baseline OT, as per the method described by Baron and
Kenny (1986). Evidence of mediation would be provided only if (1) greater frequency of
Partner Hugs was related to lower BP or HR, (2) higher plasma OTat baseline was related
to greater Partner Hugs and also, as previously demonstrated, to lower BP or HR, and (3)
the addition of baseline OT into the model reduced the regression coefficient of Hugs as
predictor of baseline SBP, DBP, MAP or of preparation or speech HR levels. Results of
these regression analyses, shown in Table 2, reveal that all of these statistical conditions
K.C. Light et al./Biological Psychology 69 (2005) 5–2113
Fig. 2. Diastolic BP levels in the high OT, moderate OTand low OT groups, depicted as described in Fig. 1. Low
OT group > high OT group at Base, P < 0.04.
were met for baseline OT serving as a significant partial mediator of the effect of Partner
Hugs onbaseline SBPandMAPinthesepremenopausalwomen (all P < 0.05),aswell asa
marginally significant mediator of the effect of Partner Hugs on baseline DBP and speech
preparation DBP (P < 0.067 and 0.09). During speech preparation (depicted) and during
active speech (latter not depicted in Table 2), although greater Partner Hugs was related to
lower HR levels, baseline OT was too weakly related to HR to meet criteria as a partial
mediator of these relationships.
This novel translational study in premenopausal women confirmed that higher baseline
OT levels were linked to lower BP and HR levels, and that a history of more frequent
Partner Hugs was associated with higher baseline OTand lower cardiovascular responses.
The observed baseline OT and BP differences may reflect the combined effects of the
individual’s general state of oxytocinergic activity, her anticipation of the upcoming
partner contact, and her prior experience in regard to warm contact with her partner. In
contrast, OT levels obtained during speech preparation, active speech, and post-speech
recovery were unrelated to cardiovascular responses, Partner Hugs, or baseline OT. Other
investigations by Altemus et al. (2001a), Turner et al. (2002) and Bonfiglio and Stoney
K.C. Light et al./Biological Psychology 69 (2005) 5–21 14
Fig. 3. Mean arterial BP levels in the high OT, moderate OTand low OT groups, depicted as described in Fig. 1.
Low OT group > high OT group at Base, P < 0.01.
(2004)likewisefailedtoobtainexpectedincreasesinplasma OTduringstress oremotional
stimuli. OT is known to be stress-sensitive, but consistent with findings by Sanders et al.
(1991), we hypothesize that persons who experience a more pronounced sympathetic and/
or HPA response during stress are more likely to show greater OT increases during stress;
we would expect these subjects to show higher, not lower, BP and HR. However, higher
baseline OT was correlated with lower BP during post-contact speech preparation and
correlated marginally with lower BP during speech recovery, though not during active
speech.This supports our earlier interpretation (Light et al., 2000) that women with greater
OT activity show a more efficient, time-limited stress response, not a reduction in peak
K.C. Light et al./Biological Psychology 69 (2005) 5–21 15
Mediational analyses: effect of Partner Hugs on cardiovascular measures is partially mediated by baseline (Base)
plasma oxytocin (OT)
Models 1–6b (CRITERION/Predictor)tb
Full model (R2)
(1) BASE OT
(2a) BASE SBP
(2b) BASE SBP
(3a) BASE DBP
(3b) BASE DBP
(4a) BASE MAP
(4b) BASE MAP
(5a) PREP DBP
(5b) PREP DBP
(6a) PREP HR
(6b) PREP HR
+P < 0.10.
*P < 0.05.
**P < 0.01.
The association between greater self-reported frequency of Partner Hugs and lower BP
during baseline before partner contact, and its potential mediation by higher OT, are
plasma OTincreases and BP decreases (Lund et al., 2002; Holst et al., 2002). Interestingly,
mothers with more children at home also had higher baseline OT levels, after partialing
out effects of Partner Hugs, which was inversely related to number of children at home.
The association between few or no children and high Partner Hugs may be due to the
‘‘honeymoon effect’’ early in a marital or partner relationship that normally precedes
parenthood. The link between more children at home and higher baseline OT is consistent
with the animal literature linking higher OT activity to greater maternal behavior
(Pedersen, 1997; Pedersen and Boccia, 2002).
Baseline OT met all criteria to be a partial mediator of the lower baseline SBP
and MAP associated with greater Partner Hugs, and was marginally significant as a
mediator for baseline and speech preparation DBP as well. The reduction in BP linked to
OT might be due to decreased central alpha-adrenergic activity, or to direct peripheral
effects of circulating OT on the heart and vasculature (Petersson, 2002). Despite meeting
statistical criteria as a mediator, it must be emphasized that all of the relationships
are regression-based (correlational), and thus this type of cause–effect role for OT
remains hypothetical in humans. A plausible alternative explanation is that some other
yet unidentified factor (behavioral or biological) correlated with all of these measures
is the true mediator. Other potential factors that might be associated with greater hugs
and OT and with reduced BP and HR prior to partner contact include personality factors
(e.g., affectionate nature, low hostility), experiential factors (past history of consistently
loving and supportive interactions with partner and other loved ones, or modeling
of positive partner interactions by ones’ own parents), and biological factors (greater
chronic estrogenic or other reproductive hormonal activity, or enhanced dopaminergic
contact followed by solitary rest, baseline OT met criteria as a mediator of the relationship
between greater partner support and lower plasma norepinephrine in women, which would
be explained equally well by either the central or peripheral OT effect. In that protocol,
lower BP in those women (Grewen et al., 2004a, 2004b). One possible explanation for this
difference between studies is that OTactivity may be more readily evoked by warm touch
and physical expressions of affection and support between partners, like hugs, and is only
indirectly linked to other less physical ways of conveying support. It is noteworthy that
another form of physical affection, frequency of neck or back massages between partners,
was also related to higher OT levels. An alternative explanation is that the sample size in
the present study was larger, and therefore the present study had more power to detect a
mediational effect. Other protocol differences, such as knowledge of the upcoming task,
may also be factors.
Greater Partner Hugs also predicted lower HR during speech preparation and active
speech, but the HR reduction did not appear to be due to OT. Our prior research indicated
that OT is a significant mediator of differences in sympathetic nervous system activity,
reflected in lower plasma NE levels (Grewen et al., 2004a, 2004b). This is also consistent
K.C. Light et al./Biological Psychology 69 (2005) 5–2116
with studies documenting that long term increases in OTactivity in rats leads to decreases
in central alpha-adrenergic tone, and this is the source of the sustained BP decreases that
result (Diaz-Cabale et al., 2000; Petersson, 2002). Lower HR levels during speech
preparation and active speech associated with greater Partner Hugs may instead be due to
increases in parasympathetic activity.
Although the present report was restricted to premenopausal women, it is worthwhile
to integrate the present findings with other reports focused on gender and reproductive/
hormonal status differences in OT activity. Men as well as women have OT activity,
known to play a role in male sexual arousal and orgasm (Carter et al., 1995; Barbaris and
Tribollet, 1996); in monogamous species like prairie voles, it also enhances pair-bonding
and mate-guarding (territoriality) (Williams et al., 1994; Bales and Carter, 2003). In our
prior study of couples, although plasma OT levels were higher in both men and women
with high versus low partner support, men did not show the OT increase 7 min after warm
partner contact or the link between higher OT and lower sympathetic nervous system
activity that women showed (Grewen et al., 2004a, 2004b). Because of OTs role in the
initiation of maternal behavior (one of the imperatives of survival for mammalian species),
it is logical in an evolutionary sense that reproductive-agefemales might have the potential
for greater and more broadly influential OTactivity (Insel and Young, 2001; Taylor, 2002;
Champagne et al., 2003). Genetic knockout mouse models have confirmed that estrogen
and OT receptors co-localized in the paraventricular and supraoptic nuclei of the
hypothalamus and in the amgydala modulate each other, and that female mice lacking
either the estrogen or the OT receptors demonstrate similarly impaired social behavior
(Choleris et al., 2003). As infants, these OT knock-out mice vocalize less during
separations from their dams, eliciting less licking and other maternal behavior when
separation ends, while as adults, the females show less maternal behavior and the males
show increased aggression (Winslow et al., 2000; Mantella et al., 2003; Choleris et al.,
Our study obtained no differences in plasma OT measures in normally cycling women
tested during the follicular, mid-cycle or luteal menstrual phases, and no differences
between such women and others tested on oral contraceptives. Although we did not
confirm self-reported cycle phase with estradiol, progesterone or luteinizing hormone
measures, our findings are consistent with prior reports by Steinwall et al. (1998) and
Altemus et al. (2001a, 2001b); Steinwall et al. (1998, p. 983) conclude that, although its
sister nonapeptide, vasopressin, does vary over the menstrual cycle, ‘‘the influence of
ovarian hormones on OT secretion is minimal’’ in non-pregnant, normally cycling women.
Nevertheless, even though plasma OT does not vary consistently across the menstrual
cycle, OTactivity (as assessed by OT mRNA) does vary in both hypothalamic and uterine
tissues (Amico et al., 2000; Steinwall et al., 2004). This serves as a reminder that plasma
OT level does not consistently mirror OT activity within the central nervous system,
especially subtle increases or decreases.
These findingsencouragefurther studyinhumansoftheeffects oflong-termormultiple
repeated episodes of warm touch, with OT and cardiovascular responses included among
and women arewith their partners than with other individuals, and that this effect is greater
ifthepartnerrelationship ismore supportiveandlessambivalent(Gump etal.,2001;Baker
K.C. Light et al./Biological Psychology 69 (2005) 5–21 17
et al., 2003; Holt-Lunstad et al., 2003). It would be fascinating to employ a more direct
intervention approach to assess whether OT may be a potential mediator of these BP
benefits. In addition to couple studies, further research is encouraged on the importance of
warm touch in mother–infant and father–infant bonding, and the possible involvement of
(friendships, homosexual partner relationships, even pets) may be usefully studied with a
similar approach (Uchino et al., 1996; Diamond, 2000; Allen et al., 2002; Odendaal and
Among 59 premenopausal women tested in the context of a 10 min period of
warm physical and emotional contact with their partners, those women showing
higher plasma OT levels just prior to the contact period had lower pre-contact resting BP
and HR. Higher pre-contact OT level was also related to lower BP during preparation
for and recovery after a stressful speech task, though these relationships were in some
cases only marginally significant. Women who reported greater frequency of hugs
with their partners were found to have lower baseline BP as well, and OTwas determined
statistically to meet criteria as a potential mediator of these effects. These findings
encourage further research into the relationships of physical and/or emotional support,
OT, and cardiovascular or other health-relevant measures, especially within the context of
This research was supported by NIH grants RR00046 and HL64927.
Allen, K., Blascovich, J., Mendes, W.B., 2002. Cardiovascular reactivity and the presence of pets, friends, and
spouses: the truth about cats and dogs. Psychosomatic Medicine 64, 727–739.
Altemus,M., Redwine,L., Leong,Y.,Frye,C.,Porges,S.,Carter, S.,2001a.Responsesto laboratorypsychosocial
stress in postpartum women. Psychosomatic Medicine 63, 814–821.
Altemus, M., Roca, C., Galliven, E., Romanos, C., Deuster, P., 2001b. Increased vasopressin and adrenocortico-
tropin responses to stress in the midluteal phase of the menstrual cycle. Journal of Clinical Endocrinology and
Metabolism 86, 2525–2530.
Amico, J.A., Davis, A.M., McCarthy, M.M., 2000. An ovarian steroid regimen that increases hypothalamic
oxytocinergic expression alters3H-muscimol in the hypothalamic supraoptic nucleus of the female rat. Brain
Research 857, 279–282.
Amico, J.A., Hempel, J., 1990. An oxytocin precursor intermediate circulates in the plasma of humans and rhesus
monkeys administered estrogen. Neuroendocrinology 51, 437–443.
Amico, J.A., Seif, S.M., Robinson, A.G., 1981. Oxytocin in human plasma-correlations with neurophysin and
stimulation with estrogen. Journal of Clinical Endocrinology and Metabolism 52, 988–993.
K.C. Light et al./Biological Psychology 69 (2005) 5–2118
Baker, B., Szalai, J., Paquette, M., Tobe, S., 2003. Marital support, spousal contact and the course of mild
hypertension. Journal of Psychosomatic Research 55, 229–233.
Bales, K., Carter, C.S., 2003. Developmental exposure to oxytocin facilitates partner preference in male prairie
voles (microtus ochrogaster). Behavioral Neuroscience 117, 854–859.
of Neurobiology 10, 119–154.
Baron, R., Kenny, D., 1986. The moderator–mediator variable distinction in social-psychological research:
conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51, 1173–
Berkman, L., 1995. The role of social relations in health promotion. Psychosomatic Medicine 57, 245–254.
Bonfiglio, D., Stoney, C., 2004. Social support influences BP, but not oxytocin, responses to stress. Presented at
American Psychosomatic Society. Orlando, FL, March 2004.
Bossmar, T., Forsling, M., Akerlund, M., 1995. Circulating oxytocin and vasopressin is influenced by ovarian
steroid replacement in women. Acta Obstetrica Gynecologia Scandinavica 74, 544–548.
Broadwell, S.D., Light, K.C., 1999. Family support and cardiovascular responses in married couples during
conflict and other interactions. International Journal of Behavioral Medicine 6, 40–63.
Broadwell, S., Light, K.C., 2004. Hostility, conflict and cardiovascular responses in married couples: a focus on
the dyad. International Journal of Behavioral Medicine.
Carels, R., Szczepanski, R., Blumenthal, J., Sherwood, A., 1998. BP reactivity and marital distress in employed
women. Psychosomatic Medicine 160, 639–643.
Carter, C.S., 1998. Neuroendocrine perspectives on social attachment and love. Psychoneuroendocrinology 23,
Carter, C.S., Altemus, M., Chrousos, G., 2001. Neuroendocrine and emotional changes in the post-partum period.
Progressive Brain Research 133, 2411–2449.
Carter, C.S., DeVries, A., Getz, I., 1995. Oxytocin and sexual behavior. Neuroscience Biobehavior Review 19,
Champagne, F.A., Weaver, I.C.G., Diorio, J., Sharma, S., Meaney, M.J., 2003. Natural variations in maternal care
are associated with estrogen receptor-alpha expression and estrogen sensitivity in the medial preoptic area.
Endocrinology 144, 4720–4724.
Choleris, E., Gustafsson, J-A., Korach, K.S., Muglia, L.J., Pfaff, D.W., Ogawa, S., 2003. An estrogen-dependent
four-gene micronet regulating social recognition: a study with oxytocin and estrogen receptor-alpha and -beta
knockout mice. PNAS 100, 6192–6197.
Crowley, R.S., Insel, T.R., O’Keefe, J.A., Kim, N.B., Amico, J.A., 1995. Increased accumulation of oxytocin
mRNA in the hypothalamus of the female rat: induction by long term estradiol and progesterone admin-
istration and subsequent progesterone withdrawal. Endocrinology 136, 224–231.
Denton, W.H., Burleson, B.R., Hobbs, B.V., Von Stein, M., Rodriguez, C.P., 2001. Cardiovascular reactivity and
initiate/avoid patterns of marital communication: a test of Gottman’s psychophysiologic model of marital
interaction. Journal of Behavioral Medicine 24, 401–421.
Diamond, L.M., 2000. Passionate friendships among adolescent sexual-minority women. Journal of Research in
Adolescence 10, 191–209.
Diaz-Cabale, Z., Petersson, M., Narvaez, J.A., Uvnas-Moberg, K., Fuxe, K., 2000. Systemic oxytocin treatment
modulates alpha-2 adrenoceptors in the telencephalic and diencephalic regions of the rat. Brain Research 887,
Elliott, J., Lubin, D.A., Johns, J.M., 2001. Acute cocaine alters oxytocin levels in the medial preoptic area and
amygdala of lactating rat dams: Implications for cocaine-induced changes in maternal behavior and maternal
aggression. Neuropeptides 35, 1–8.
Ewart, C., Taylor, C., Kraemer, H., Agras, W., 1991. High BP and marital discord: not being nasty matters more
than being nice. Health Psychology 10, 155–163.
Ferguson, J., Young, L., Insel, T., 2002. The neuroendocrine basis of social recognition. Frontiers in Neuroen-
docrinology 23, 200–224.
Gottman, J.M., Notarius, C.I., 2002. Martial research in the 20th century and a research agenda for the 21st
century. Family Processes 41, 159–197.
K.C. Light et al./Biological Psychology 69 (2005) 5–2119