ArticlePDF Available

Romantic Involvement Often Reduces Men's Testosterone Levels—but Not Always: The Moderating Role of Extrapair Sexual Interest

Authors:

Abstract

Testosterone (T) appears to facilitate what biologists refer to as mating effort--the investment of time and energy into same-sex competition and mate-seeking behavior. Multiple studies show that men who are romantically involved (i.e., are paired) have lower T than single men, which may be due to a facultative adjustment by men of T levels in response to lower demands for mating effort. The authors proceeded on the basis of the idea that men who retain interests in sexual opportunities with women other than a primary partner continue to dedicate more time and energy to mating effort when romantically paired, and so they predicted that the association between relationship status and T depends on men's extrapair sexual interests. Study 1 used the Sociosexual Orientation Inventory to measure extrapair sexual interests, whereas Study 2 used a broader measure to examine this interaction. Both studies found support for it. These results have implications for an understanding of the biosocial regulation of men's behavior in romantic relationships.
INTERPERSONAL RELATIONS AND GROUP PROCESSES
Romantic Involvement Often Reduces Men’s Testosterone Levels—But
Not Always: The Moderating Role of Extrapair Sexual Interest
Matthew McIntyre
Harvard University
Steven W. Gangestad
University of New Mexico
Peter B. Gray
University of Nevada, Las Vegas
Judith Flynn Chapman
Harvard University
Terence C. Burnham
Harvard Business School
Mary T. O’Rourke
Harvard University
Randy Thornhill
University of New Mexico
Testosterone (T) appears to facilitate what biologists refer to as mating effort—the investment of time and
energy into same-sex competition and mate-seeking behavior. Multiple studies show that men who are
romantically involved (i.e., are paired) have lower T than single men, which may be due to a facultative
adjustment by men of T levels in response to lower demands for mating effort. The authors proceeded
on the basis of the idea that men who retain interests in sexual opportunities with women other than a
primary partner continue to dedicate more time and energy to mating effort when romantically paired,
and so they predicted that the association between relationship status and T depends on men’s extrapair
sexual interests. Study 1 used the Sociosexual Orientation Inventory to measure extrapair sexual interests,
whereas Study 2 used a broader measure to examine this interaction. Both studies found support for it.
These results have implications for an understanding of the biosocial regulation of men’s behavior in
romantic relationships.
Keywords: testosterone, relationship status, sociosexuality, sexual selection
The steroid hormone testosterone (T) plays a role in facilitating
what biologists refer to as mating effort—the investment of time
and energy into same-sex competition and mate-seeking behavior
(Ellison, 2001, pp. 274–280). In both men and women, T promotes
the maintenance and growth of skeletal muscle, particularly the
sexually dimorphic mass in the chest, upper arms, and shoulders
(Basaria et al., 2002; Bhasin, 2003; Schroeder et al., 2003). Male
upper body musculature is thought to have facilitated success in
competition between men among human ancestors and, hence, the
sexual dimorphism in muscle mass is thought to be partly the result
of sexual selection on men’s abilities to compete for mates
(through direct competition with each other and display of intra-
sexual competitive abilities to women; Ellison, 2001, pp. 273–274;
Gaulin & Sailer, 1984; Martin, 1980).
1
Furthermore, T supports
psychological and behavioral outcomes that appear to encourage
success in male–male competition and sexual behavior.
A recent review of the literature on the association between T
and sexual motivation suggests a threshold effect of T on libido
(Bancroft, 2002). T levels below a certain level appear to be
positively related to libido. Above this level, however, the mar-
1
A competing theory states that the sexual dimorphism in muscular
strength in humans has been maintained by a sexual division of labor rather
than sexual selection. See Ellison (2001, pp. 272–273) for a discussion.
Matthew McIntyre, Department of Epidemiology, Harvard School of
Public Health, Harvard University; Steven W. Gangestad, Department of
Psychology, University of New Mexico; Peter B. Gray, Department of
Anthropology and Ethnic Studies, University of Nevada, Las Vegas; Judith
Flynn Chapman and Mary T. O’Rourke, Department of Anthropology,
Harvard University; Terence C. Burnham, Harvard Business School;
Randy Thornhill, Department of Biology, University of New Mexico.
Correspondence concerning this article should be addressed to Matthew
McIntyre, Department of Epidemiology, Channing Laboratory, Harvard
University, 181 Longwood Avenue, 3rd Floor, Boston, MA 02115. E-mail:
mmcintyr@hsph.harvard.edu
Journal of Personality and Social Psychology, 2006, Vol. 91, No. 4, 642– 651
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.4.642
642
ginal increase in libido associated with T is minimal. In another
review of both human and rhesus macaque findings, Wallen (2001)
discussed the difficulty of identifying changes in sexual motivation
by using behavioral measures. Extrinsic social conditions constrain
the ability of an individual to act on sexual feelings and may do so
differently for different individuals. Relationships between sexual
behaviors and T are stronger when social conditions are experi-
mentally controlled.
Even when not involved directly in increasing libido, T may
function to facilitate male pursuit of female interest (i.e., mating
effort). Wild male chimpanzees have increased T when parous
females exhibit maximally tumescent sexual swellings but not in
response to maximal swelling in nulliparous females, whom males
do not compete as intensively over. As males copulate with nul-
liparous and parous females at approximately the same rates, these
findings suggest that these changes in T facilitate male pursuit of
females rather than sexual performance per se (Muller & Wrang-
ham, 2004). A recent study found that men’s T increased after they
interacted with an attractive woman and particularly so when the
woman thought they were trying to impress her (Roney, Mahler, &
Maestripieri, 2003).
In related findings, T also appears to be associated with aspects
of social assertiveness (Ellison, 2001, p. 265) or dominance seek-
ing (Mazur & Booth, 1998). In chimpanzees, males become more
aggressive when parous, but not nulliparous, females exhibit sex-
ual swellings, which parallels changes in T (Muller & Wrangham,
2004). Associations between T and overt intrasexual aggressive-
ness in humans are weak, though meta-analysis reveals that they
are reliable (Archer, Birring, & Wu, 1998; Book, Starzyk, &
Quinsey, 2001). Social conditions constrain aggressive behavior
just as they constrain sexual behavior. Dominance seeking may or
may not entail aggressive behavior depending on social conditions
and other moderating factors. For example, once an informal
dominance hierarchy in a primate group is established, aggressive
behavior may be minimal (and instead dominance relations may be
established through stare downs). Furthermore, one should not
expect physical aggressivity between two individuals who are able
to accurately assess each other’s abilities to win a contest (de
Waal, 1986; de Waal & Hoekstra, 1980). (In the human case,
legal/moral strictures [i.e., the threat of institutional retaliation] are
apropos.) Dominance seeking may be expressed, among other
behaviors, in greater selective attention to angry faces (van Honk
et al., 2000; van Honk et al., 1999), in less pronounced smiling
(Dabbs, 1997), or in more visual attention toward interaction
partners (Dabbs, Bernieri, Strong, Campo, & Milun, 2001).
T and Romantic Bonding
If T is associated with mating effort in men, then one might
expect men who are in committed romantic relationships to have
lower levels of T. In fact, much evidence supports this prediction.
Married men have lower T than unmarried men (Booth & Dabbs,
1993; Gray, Campbell, Marlowe, Lipson, & Ellison, 2004; Gray,
Kahlenberg, Barrett, Lipson, & Ellison, 2002; Mazur & Michalek,
1998). Unmarried men in committed, romantic relationships have
lower T than unpaired men (Burnham et al., 2003). In an important
longitudinal study, Mazur and Michalek (1998) found that di-
vorced men’s T dropped if they remarried. The T of men who were
married and then divorced was particularly high right around the
period of the divorce. Together, these findings suggest that T
increases when men (presumably) are searching for romantic part-
ners and decreases when men (presumably) are not searching for
romantic partners.
In an evolutionary framework, responsiveness of T production
to mating effort implies that there is a cost to men of maintaining
high T when they are mated. In fact, T has such costs. It channels
energy that could be used otherwise into, for instance, maintaining
muscle mass. Perhaps for this reason, T may impair immune
functioning (Campbell, Lukas, & Campbell, 2001; Klein, 2000;
but see Granger, Booth, & Johnson, 2000). T may also encourage
risk-taking behavior that, although potentially beneficial when
men are competing for mates, entails costs not worth undertaking
when men are in committed relationships. Finally, T may interfere
with affiliative and nurturing behaviors that are important in spe-
cies characterized by long-term social bonds between relationship
partners and parental care by both sexes. In one such species,
marmosets, T is reduced when males become fathers (Nunes, Fite,
& French, 2000; Nunes, Fite, Patera, & French, 2001). Three
studies have shown that T is similarly low in men who have
recently become fathers (Berg & Wynne-Edwards, 2001; Fleming,
Corter, Stallings, & Steiner, 2002; Storey, Walsh, Quinton, &
Wynne-Edwards, 2000).
Other examples of behaviors shown to affect T production in
men may also be indirectly related to mating effort. The most
important paradigm revealing responsive fluctuation in T has been
the effects, in men, of winning or losing competitions. In antici-
pation of competition, T level rises. However, after the competi-
tion, T levels fall in losers but remain high in winners. Following
the original findings in tennis matches and among recent recipients
of the MD degree (Mazur & Lamb, 1980), and then among
wrestlers (Elias, 1981), a similar difference has been reported
among chess players (Mazur, Booth, & Dabbs, 1992), video game
players (Mazur, Susman, & Edelbrock, 1997), and in fans (Bern-
hardt, Dabbs, Fielden, & Lutter, 1998). (A study of competitors in
judo matches failed to replicate the effect; Suay et al., 1999.) T has
also been found to increase following sexual intercourse (Dabbs &
Mohammed, 1992). As noted above, T increases after men interact
with an attractive woman (Roney et al., 2003). Competition, sexual
intercourse, and association with women could all be seen as
important signals of opportunity for, or devotion to, finding new
romantic partners as opposed to nonsexual pursuits, parenting, or
maintaining existing romantic relationships that have important
nonsexual value.
Although it remains unclear whether short-term fluctuations in
T that are affected by social experiences could also have short-
term influences on psychological states or behavior, it seems likely
that social experiences might serve as a cue about conditions for
mate seeking over a longer term, during which T could affect both
relevant physical states, such as metabolism and physical strength,
and also psychological states, such as mood, libido, and compet-
itiveness. In addition to the evidence for the effects of T already
discussed, clinical trials of T administration have shown clear
effects on mood and aggressiveness (Pope, Kouri, & Hudson,
2000). Whereas sexual and romantic relationships in any par-
ticular contemporary society involve particular norms, physical
and psychological states associated with power, elevated mood,
and elevated libido (to the extent that they can be generalized
across cultures and species) are closely associated with male
643
MATING, TESTOSTERONE, AND EXTRAPAIR INTEREST
mating activities in mammals as a whole and in humans taken
as a species.
The Potential Moderating Role of Extrapair Sexual
Interests
Although longitudinal data indicate that T levels change as a
result of changing relationship status (Mazur & Michalek, 1998),
from a conceptual standpoint, it is not clear that all men’s T levels
should change in the same way. Whereas the levels of many men
may drop when they become involved in a committed relationship,
the levels of others may remain relatively high. The current studies
investigated potential moderators of the effects of relationship
status on men’s T.
Paired men may be faithful or unfaithful. Faithful men are
relatively committed to the sexually exclusive nature of their
relationships. Although they may find women other than their
partners attractive, they do not engage in effort to attract these
women—that is, they do not engage in extrapair mating effort.
Unfaithful men remain relatively interested in and open to oppor-
tunities to engage in sex outside of a relationship. They hence do
engage in extrapair mating effort, at least when opportunities arise.
If men’s changing interests in seeking and attracting new mates
account for the association between their relationship status and T,
men’s faithfulness may moderate this effect. The T levels of men
who tend not to be interested in seeking new mates when in a
committed relationship should hence change as a function of their
relationship status. By contrast, the T levels of men relatively open
to pursuing short-term sexual relationships with women other than
a primary long-term partner may remain relatively high and
change less markedly as a function of their relationship status.
Hence, their mating effort may remain relatively high even when
in a committed relationship. It follows, from this reasoning, that
men’s extrapair sexual interest should statistically interact with
their relationship status to predict T.
Study 1: Sociosexual Orientation, Relationship Status, and T
Men’s degree of interest in pursuing or being open to sex
outside of a relationship may vary along a dimension. Sociosexual
orientation refers to individual differences in the willingness to
engage in sex outside of a committed, emotionally involved ro-
mantic relationship (Simpson & Gangestad, 1991). Individuals
with a restricted sociosexual orientation claim to be uncomfort-
able having sex outside of a committed relationship in which they
are emotionally involved and have little history of doing so. When
in a committed relationship, they tend not to fantasize about sex
with individuals other than their partners. By contrast, individuals
with an unrestricted sociosexual orientation claim to be relatively
comfortable engaging in “casual,” uncommitted sex and tend to
have a history of doing so. When in committed relationships, they
tend to fantasize about sex with individuals other than their part-
ners. Sociosexual orientation can be measured with the Sociosex-
ual Orientation Inventory (SOI), a short, seven-item questionnaire
assessing past sexual history, sexual fantasies, and attitudes toward
uncommitted, casual sex (Simpson & Gangestad, 1991). High
scores reflect relatively unrestricted sociosexual orientation; low
scores reflect a relatively restricted one. Research has provided
evidence for convergent validation of the SOI and discriminant
validity evidence that sociosexual orientation does not merely
reflect generalized sex drive. In one study, the SOI did not predict
the frequency of sex occurring within committed relationships,
though it did predict how soon after the beginning of a relationship
sex occurred (Simpson & Gangestad, 1991).
Conceptually, sociosexual orientation should relate to men’s
willingness to engage in sex outside of a relationship. In Study 1,
then, we tested the hypothesis that male SOI moderates the asso-
ciation between relationship status and T.
Method
Participants. Participants were undergraduate students at Harvard
University between the ages of 17 and 26 years (M 20 years, SD 1.51).
Participants were recruited from an introductory lecture course in psychol-
ogy and anthropology that satisfies general matriculation requirements
(n 67) and at a table outside the dining room of an undergraduate
residence hall (n 40). Of 107 participants, 5 were excluded from the
analyses presented in this article because they completed fewer than 4 of 5
components of the SOI questionnaire.
Procedure. Lecture course students who chose to participate provided
saliva samples and completed questionnaires at the end of required dis-
cussion sessions (in groups of 12–17), at various days and times, over the
course of 1 week. Residence hall students were asked to provide saliva
samples and complete the questionnaire either (a) at the table or (b) if
already late in the day, on waking the next day. The latter participants
returned their materials to researchers the following day. Consent was
obtained verbally, and all materials were confidential and anonymous.
The questionnaires included the following:
1. A brief introductory questionnaire containing items requesting
demographic information (e.g., age), the time of day at which the
participant awoke, the time at which the saliva sample was
collected, relationship status (specifically, whether the partici-
pant was in a committed, romantic relationship), the duration of
his relationship (if applicable), and the percentage of time the
participant had been in a relationship during the past 3 years.
2. SOI (Simpson & Gangestad, 1991), an eight-item questionnaire
asking (a) the number of partners in the past year, (b) the number
of partners estimated for the next 5 years, (c) the number of
partners with whom the respondent has had sex once and only
once, (d) the frequency of fantasy about sex with someone other
than a current romantic partner, and (e) three items, forming one
component, concerning attitudes toward uncommitted, casual sex
(e.g., “I can imagine myself being comfortable and enjoying
“casual” sex with different partners”). These five components
were first z scored and then averaged to produce total scores
(␣⫽.81).
Before completing the questionnaire, participants expectorated at least 1
ml of saliva into a test tube treated with the preservative sodium azide.
Trident Sugar-Free Gum, which negligibly interferes with the T assay, was
used to stimulate saliva production.
After completion of the session, participants were fully debriefed, given
the opportunity to ask questions, and thanked for their participation. The
study was approved by the Harvard University review board for human
subjects research.
Salivary T assays. Analysis of saliva samples followed published
protocols (Granger, Schwartz, Booth, & Arentz, 1999). The time at which
saliva samples were collected ranged between 0700 and 0110. Participants
had been awake at the time of collection for between 0001 and 1630. T
concentration declines over the course of the waking day (Nelson, 2000).
Therefore, all analyses include time awake as a covariate. Saliva samples
644
MCINTYRE ET AL.
were assayed for T in the Reproductive Ecology Laboratory at Harvard
University. The assay method used is a modified version of Granger et al.
(1999) that was based on an application of the
125
I double antibody kit
produced by Diagnostic Systems Laboratories (Webster, TX). Sample and
standard reactions were run in duplicate. Substrate (400 l) was pipetted
into borosilicate tubes, 200 l of sample and 200 l of buffered saline or,
for the standard reactions, a 400 pg/ml standard concentration was added
in volumes of 2, 5, 15, 50, 125, and 375 l, with volumes of buffered saline
adjusted to yield 400 l total volume. Undiluted antiserum (20 l) and
tracer (50 l) were added to sample and standard tubes. Reactions incu-
bated overnight for at least 18 hr, after which precipitating reagent (500 l)
was added, and the tubes were centrifuged. After centrifugation, the su-
pernatant was aspirated before the tubes were placed in a gamma counter.
The assays were sensitive to 14 pmol/L T.
Participants were allocated randomly into three lots. Interassay coeffi-
cients of variation were 12.0% for low pools and 15.3% for high pools. The
intraassay coefficient of variation was 9.5%. T concentrations reported in
this article are the natural-logged averages of duplicates. For these data,
natural-logging T concentration (a) normalized the right-skewed concen-
tration variable and (b) maximized the variance explained by time awake.
Identification of relationship status groups. Differences in T levels
between groups of individuals identified by dating status in this sample are
reported elsewhere by Gray, Chapman, et al. (2004). As noted by Gray,
Chapman, et al., the two groups of individuals who were not in a relation-
ship could be distinguished on the basis of T. Of the 102 participants, 37
reported being in a romantic relationship. Of the 65 participants not
currently in a relationship, 21 also reported not having been in a relation-
ship in the past 3 years. The less experienced men were somewhat younger
than their 44 more experienced counterparts not currently in a relationship
(19.5 years old relative to 20.1 years old; Cohen’s d 0.39, p .13), had
somewhat lower (though insignificantly so) SOI scores (Cohen’s d 0.40,
p .12), and, most notably, had lower salivary T (Cohen’s d 0.51, p
.05). In fact, men who reported no experience with being in a relationship
had mean T levels slightly (though nonsignificantly) lower than men in a
relationship. The low T levels of these men suggest that, despite their not
currently being in a relationship, they were investing relatively little in
overall mating effort. One possibility is that, because of their youth or
developmental immaturity, they had not developed the same level of
interest in attracting romantic partners as other men. A second possibility
is that they were relatively unattractive as mates and hence invested less in
mating effort. In any case, in light of their low T levels, our prediction that
SOI moderates the effect of relationship status on T is not clearly tested
with this group. Also, Roney et al. (2003) did not find behavioral effects on
T among romantically less experienced men, supporting their treatment as
a distinct group. We do not include this group in our main analyses, though
we also report results with them included. In our main analyses to test our
hypothesis, we contrasted the paired group with the single group that
included only men with some experience (in the last 3 years) with rela-
tionships. We also tested the relationship between SOI and T in paired men
only, among whom a positive relationship was expected.
Treatment of SOI and relationship length. To be consistent with our
treatment of T concentration, we examined the distribution of SOI scores.
It too was strongly right skewed. Hence, we log-transformed it as well. We
added two points to the original SOI composite before logging these values
to ensure positive outputs. The resulting skew was near zero. SOI was zero
centered prior to being entered into analyses to ensure orthogonality of
main effects and its interaction with relationship status.
In some analyses on paired men, we include the variable relationship
length (in months). This variable was also right skewed and hence we
log-transformed it as well.
Results
Logged T levels were analyzed with SPSS 12.0 (GLM Univar-
iate) in a two-level factor design based on relationship status
(paired vs. single; excluding less experienced men) and three
continuous predictors. Logged SOI score was entered as a contin-
uous predictor of interest, along with its interaction with relation-
ship status (except when only paired men were included). Time
awake was also entered as a control for the diurnal pattern of T
production and was, in our overall analysis, a significant predictor,
F(1, 75) 17.73, p .001, but is not discussed further. Age was
also entered as a continuous predictor because of its association
with both T and relationship history. Predicted effects were as-
sessed with directed tests that allocate a probability of .04 to a
predicted “tail’s” region of rejection and .01 to a nonpredicted tail
(Rice & Gaines, 1994). (This procedure enhances power to detect
a predicted effect relative to two-tailed tests without the problem
of excluding any possibility of a nonpredicted effect entailed by
one-tailed tests.)
Results revealed the predicted interaction between relationship
status and SOI, F(1, 75) 5.38,p .02. The main effects for
relationship status, F(1, 75) 1.17, and SOI, F(1, 75) 0.18, and
age, F(1, 75) 2.63, were not significant. Table 1 reports all effects.
When less experienced men were included (with relationship
experience included as a second factor), the interaction between
relationship status and SOI score remained significant, F(1, 95)
3.58, p .04. Relationship experience also had a significant effect,
with more experienced men having higher T, F(1, 95) 6.39, p
.01, as did age, with older men having lower T, F(1, 95) 3.43,
p .04. Again, relationship status, F(1, 95) 1.29, and SOI, F(1,
95) 1.22, did not have a significant effect.
Are the observed interactions between relationship status and
SOI in predicting T levels explained by an association between T
and SOI among single men, paired men, or both? To answer this
question, we performed similar GLM analyses on paired men only
and on single men only. In paired men, SOI approached significant
prediction of T levels, F(1, 33) 2.68, p .069. Higher SOI
scores tended to be associated with higher T levels. In paired men,
however, there exists one additional factor that can be controlled—
relationship length, which has been discussed previously as a
negative predictor of T, at least over the first several months of a
relationship (Gray, Chapman, et al., 2004). In an analysis including
relationship length as a predictor, both SOI, F(1, 32) 3.69, p
.04, and relationship length, F(1, 32) 3.43, p .04, were indeed
significant predictors (see Table 2). Consistent with our prediction,
then, men’s SOI is positively associated with T in paired men with
relationship length controlled.
Table 1
Study 1 General Linear Model Analysis of Testosterone (T)
Levels
Variable F(1, 75) Partial r
Time awake 17.73** .44
Age 2.63 .18
Relationship status 1.58 .14
SOI 0.18 .04
SOI Relationship Status 5.38* .26
Note. N 81. T and Sociosexual Orientation Inventory (SOI) are logged
values. Relationship status: 0 single, 1 paired. For all variables,
directed tests were used to obtain p values.
* p .05. ** p .01.
645
MATING, TESTOSTERONE, AND EXTRAPAIR INTEREST
In more experienced single men, SOI did not significantly
predict T, F(1, 40) 2.35, p .133 (see Table 2). When less
experienced men were included and experience added as an addi-
tional predictor, SOI similarly was not significantly associated
with T, F(1, 60) 0.50. In this analysis, we also included
percentage of time men had spent in a relationship in the past 3
years as a predictor. It had negligible effect, F(1, 39) 0.07, ns,
and its inclusion did not alter other effects. Its lack of effect is
consistent with the proposal that relationship status affects T and
not consistent with the proposal that T affects men’s tendency to
be in relationships.
Discussion
On the basis of the idea that romantically paired men with
unrestricted sociosexual orientation continue to dedicate more time
and energy to mating effort, the allocation of which is partly
modulated by T, whereas men with relatively restricted sociosex-
ual orientation engage in substantially less mating effort when
romantically paired, we predicted that the association between
relationship status and T would depend on men’s sociosexual
orientation. In a sample of undergraduates more experienced with
being in relationships, this prediction was supported by two lines
of evidence. First, we detected a significant interaction between
relationship status and SOI. Second, we found that among paired
men, but not among single men, SOI predicts T levels. A predic-
tion following from an alternative conceptualization, that SOI
would be associated with T independent of relationship status, was
not supported.
To replicate and extend this finding, we examined these asso-
ciations in a second study. The second study differed from the first
in one important respect and in two more minor respects.
The first study included only the SOI as a measure of interest in
mating effort when in a relationship. Conceptually, it makes sense
that the SOI could effectively serve as such a measure, but other
measures are also possible. The SOI was designed to assess indi-
vidual differences in willingness to have sex outside of a commit-
ted relationship characterized by emotional closeness. In the
present context, we are interested in individual differences in
willingness and desire to have sex with a partner other than a
primary partner with whom an individual is currently paired. We
added two simple measures to tap these individual differences: a
question asking whether an individual could imagine themselves
having sex outside of a relationship (an “affair”) and a question
asking whether an individual actually had had sex with someone
other than a primary partner while being involved in a committed
relationship. To test the prediction that willingness to engage and
to have interest in sex with someone other than a primary partner
would moderate the effect of relationship status on T, we examined
T levels as a function of these measures as well as the SOI. (In fact,
as we later describe, we created a composite measure, as all three
covaried in our sample.)
In the first study, we controlled for relationship length. We did
not ask about relationship length in Study 2. (Data were collected
initially for other purposes, for which relationship length was not
required.) We did, however, inquire about the nature of a relation-
ship (e.g., whether individuals were married, engaged, living with
their partner, or dating exclusively), and we controlled for these
variations. Whereas participants of Study 1 were Harvard under-
graduates, Study 2 was conducted at the University of New Mex-
ico, an institution with a student body presumably more diverse
along sociocultural and intellectual dimensions.
Study 2
Method
Participants. Participants were 74 men who took part in a larger study
on scents and attraction (see Thornhill & Gangestad, 1999). Individuals
were recruited from Introductory Psychology or other classes and given
either course credit for a research requirement or extra credit toward their
grade for participating. All participants were asked their sexual orientation
(heterosexual, homosexual, or bisexual). As our interest in this study was
on heterosexual relationships, only the 69 men who reported to be hetero-
sexual were retained for analysis. Of these men, the mean age was 20.3
years (range 17–33, SD 2.40). (One individual did not report his age
and hence could not be included in analyses controlling for age.) Fifty-five
percent reported themselves to be Caucasian, 27% Hispanic, 7% African
American, 4% Native American, 3% Asian American, and 3% another
ethnicity.
Procedure. Participation involved two sessions. In a first session, after
providing informed consent, participants were asked to fill out a series of
questionnaires. These questionnaires included the following:
1. A personal data sheet. This questionnaire asked individuals basic
demographic information such as age, ethnic background, and
sexual orientation. In addition, relationship status was assessed
with a series of seven items. Each individual was asked to
indicate with a check mark whether each of the following char-
acteristics applied to him: married, married but separated, di-
vorced, engaged to be married, not currently married but cohab-
iting with a partner, dating one person exclusively, dating
multiple persons, and not currently dating. An individual was
classified as paired (n 26) if he was dating one person
exclusively (n 19), married (n 4), cohabiting with a partner
(n 2), or engaged to be married (n 1). He was classified as
single (n 43) if he was dating multiple persons (n 15) or not
dating (n 29). (One individual claimed to be dating multiple
persons and not dating.)
2. SOI (Simpson & Gangestad, 1991). As in Study 1, the five
components of this measure were z scored and averaged to create
a composite (␣⫽.75).
3. Willingness to engage in extrapair sex. A single item asked
Table 2
Study 1 General Linear Model Analysis of Testosterone (T)
Levels in Paired and Single Men
Variable
Paired men Single men
F(1, 32) Partial rF(1, 40) Partial r
Time awake 18.09** .60 1.59 .19
Age 0.59 .13 1.92 .21
Relationship length 3.43* .31
SOI 3.69* .32 2.35 .24
Note. n 37 and 44, respectively. T, relationship length, and Sociosexual
Orientation Inventory (SOI) are logged values. For all variables, directed
tests were used to obtain p values.
* p .05. ** p .01.
646
MCINTYRE ET AL.
whether men would ever consider having extrapair sexual rela-
tions. Specifically, they were asked
Would you ever consider having an “affair” (sex with a person
other than a main, current relationship partner) behind the back
of your relationship partner? (Here, consider not only your
present partner [if you have one], but any partner you might have
in the future.) Check one: A. No, I would never have sex outside
of a relationship under any circumstances. B. I can imagine that
I could possibly have sex outside of a relationship under certain
circumstances.
Of all men, 52% claimed that they would not; 48% said that they
could imagine such circumstances. Within groups, 23% and 63%
of paired and single men, respectively, claimed that they could,
2
(1, N 69) 10.24, p .001.
4. History of extrapair sex. Men were asked whether they had ever
engaged in sex with a partner other than a current partner while
involved in a romantic relationship. Of all men, 29% claimed that
they had, including 23% of paired men and 33% of single men,
2
(1, N 69) 0.81, ns.
At the first session, men were provided a test tube, a stick of sugarless
gum, and instructions for how to collect saliva. Each tube had a small
amount of a preservative, sodium azide. Each individual brought a tube
with approximately 1 ml of saliva to a second session. They were asked to
collect the saliva on waking the morning of their session to control for
diurnal variation in T. (For reasons unrelated to the current study but
related to the study’s focus on scent and attraction, individuals were asked
to refrain from eating pungent foods and from drinking alcohol, smoking,
having sex, or sleeping with someone the two days and nights prior to
collection of the saliva.) Following completion of the study, individuals
were fully debriefed and given an opportunity to ask questions.
Salivary T assay. T concentrations were estimated with a modification
of a commercially available fluoroimmunoassay kit (DELFIA testosterone,
Wallac, Turku, Finland; Wallac is now part of Perkin-Elmer, Boston, MA)
in the Reproductive Ecology Laboratory at Harvard University. One mil-
liliter of sample was extracted in diethyl ether, dried under nitrogen, and
reconstituted in 100 ul of assay buffer; that is, the volume increased by a
factor of 10. Duplicate 25 ul aliquots were assayed for each sample. High
T and low T quality control pools were run with each assay (coefficient of
variation 4.6% high pool, 18.1% low pool), and in all cases where there
were two samples from an individual both samples were run in the same
assay and averaged. Concentrations for the standard curve were from 0.5 to
50.0 nmol/L. Values for all samples were above the lower limit of the
curve. As in Study 1, T concentrations were log-transformed.
Results
Associations between sociosexual orientation, willingness to
have extrapair sex, and history of extrapair sex. We first exam-
ined the correlations between our three measures tapping individ-
ual differences in interest in extrapair or casual sex. Not surpris-
ingly, all three measures correlated significantly: SOI and
willingness to have extrapair sex, r .47, p .001; SOI and
history of extrapair sex, r .37, p .001; and willingness to have
extrapair sex and history of extrapair sex, r .30, p .02. At the
same time, the measures were not so highly correlated so as to be
completely redundant with one another. A first principal compo-
nent running through these variables accounted for 60% of their
total variance (loadings range from .73 to .83). To tap this com-
ponent with a single measure, we z scored all three variables and
averaged these z scores for each individual. (One individual did not
report on his history of affairs and hence his score was the average
of the other two components.) We interpret and refer to this
composite as extrapair sexual interest.
Group comparison on T: Paired versus single men. Before
examining the moderating role of extrapair sexual interest, we
examined whether T levels of paired and single men differed.
Consistent with other work, T did vary as a function of relationship
status, t(67) 2.61, p .01. Single men (mean logged T 6.16,
SD 0.30) had higher levels of T than paired men (M 5.94,
SD 0.39).
T as a function of relationship status and extrapair sexual
interest. T levels were analyzed with SPSS 12.0 (GLM Univar-
iate) in a two group (relationship status: paired vs. single) design
with two continuous predictors, extrapair sexual interest and age.
Directed tests (Rice & Gaines, 1994) were used to assess the
predicted effects of relationship status and the extrapair sexual
interest by relationship status interaction.
Results revealed that extrapair sexual interest significantly in-
teracted with relationship status, F(1, 63) 4.32, p .03 (see
Table 3). The effect of relationship status fell short of statistical
significance, F(1, 63) 2.55, p .072, as did the effect of age,
F(1, 63) 1.80, ns. To interpret the interaction effect, we ran
separate analyses on paired and single men. In paired men, the
effect of extrapair sexual interest approached significance, F(1,
23) 2.44, p .083, as in Study 1. In Study 1, we also controlled
for relationship length. In this study, we did not measure relation-
ship length. We did, however, ask men whether they were married
or engaged, living with their partner, or simply dating one person.
These categories undoubtedly reflect differences in level of com-
mitment to a relationship as well as, to some extent, relationship
length. To control for overall level of commitment (or type) of
relationships, we categorized men into three groups (married/
engaged, living with, and dating only) and included this variable in
the analysis. As predicted, extrapair sexual interest significantly
predicted T, F(1, 21) 5.49, p .02 (see Table 4). This result
conceptually replicates our finding for paired men in Study 1. The
effects of age, F(1, 21) 2.42, p .084, and relationship type,
F(1, 21) 2.21, p .13, fell short of statistical significance.
In single men, neither extrapair sexual interest, F(1, 39) 1.30,
ns, nor age, F(1, 39) 0.80, ns, predicted T (see Table 4). In a
follow-up analysis, we also included whether men said they were
dating multiple women (vs. not currently dating someone). The
effects of extrapair sexual interest, F(1, 38) 0.71, and age, F(1,
Table 3
Study 2 General Linear Model Analysis of Testosterone (T)
Levels
Variable F(1, 63) Partial r
Age 1.80 .17
Relationship status 2.55 .20
EPSI 0.81 .11
EPSI Relationship Status 4.32* .25
Note. N 68. T is logged. Relationship status: 0 single, 1 paired.
For all variables, directed tests were used to obtain p values. EPSI
extrapair sexual interests.
* p .05.
647
MATING, TESTOSTERONE, AND EXTRAPAIR INTEREST
38) 1.43, remained nonsignificant. The effect of dating multiple
women fell just short of significance, F(1, 37) 3.82, p .058.
Perhaps interestingly, men who reported dating multiple women
tended to have lower T levels than men who reported not currently
dating.
2
General Discussion
We found in both Study 1 and Study 2 that involvement in a
romantic relationship and interest in extrapair sexual interest in-
teract to predict T level among male college students. We interpret
these results as suggesting that the relationship between romantic
relationship status and T production is psychologically moderated
by commitment to the sexually exclusive nature of the relationship
and/or sexual interests outside the relationship. Men in relation-
ships but who nonetheless are interested in new sexual encounters
maintain high T levels despite being paired.
These effects persist even when relationship length (Study 1) or
depth of long-term commitment to the relationship (marital and
living status; Study 2) are controlled. These additional results
further bolster the interpretation that it is sexual interest in outside
partners, not long-term commitment to the primary relationship,
that accounts for the effects we observed. Were it the case that
relationship duration or long-term commitment mediated the ef-
fects, the associations between measures of extrapair sexual inter-
ests and T would have weakened when duration and relationship
type were controlled (Baron & Kenny, 1986). They showed no hint
of doing so; indeed, the associations tended to strengthen in our
samples. This pattern of results suggests that factors that contribute
to relationship duration and commitment and those that contribute
to extrapair sexual interests have at least partially independent
effects on the T of men in relationships.
Of course, these results do not imply that there are no differ-
ences in the qualities of the relationships of men who are relatively
open to extrapair sexual relationships and those who are not, even
with relationship duration controlled, or that correlates do not also
predict T in paired men. We suspect there may be differences (e.g.,
with respect to men’s desire for intimacy, willingness to sacrifice
self-interests for partners, honesty with their partners, or desire to
spend time with their partners). Possibly, these characteristics
relate to T in paired men in a fashion similar to men’s interests in
extrapair sex. Future research may explore the more general set of
correlates of T in paired men. The framework we have used
suggests that characteristics that reflect continued mating effort
(e.g., diminished honesty, unwillingness to sacrifice self-interests
for partners) may indeed be associated with higher T amongst men
in relationships.
The Social Modulation of T
These findings are consistent with a growing literature that
indicates the importance of the understanding that men’s T levels
are modulated by social circumstances and the motivations that
they induce. As noted earlier, a variety of studies indicate that T
levels are affected by outcomes of social competition (e.g., Bern-
hardt et al., 1998; Mazur et al., 1992, 1997). In addition, a recent
study showed that men who interacted with an attractive woman
experienced increased T production (Roney et al., 2003). T varies
as a function of relationship status (e.g., Booth & Dabbs, 1993;
Burnham et al., 2003; Gray et al., 2002; Mazur & Michalek, 1998).
The current study suggests that these latter effects are at least
partly due to the modulation of T by social circumstances: Al-
though the T levels of men may, on average, drop when they enter
relationships, the T levels of men who retain extrapair sexual
interests even when in a relationship show no evidence of such a
drop. Were this association due solely to the effects of T on men’s
tendency to be in relationships, there would have been little reason
to expect an interaction between men’s extrapair sexual interests
and their T levels. Men who were less likely to be in committed
relationships (e.g., perhaps men with higher SOI scores) would be
expected to have higher T independent of whether they were
currently in a committed relationship.
Possibly, though men with higher T are less likely to be in
relationships, when they do get into relationships, their higher T
predisposes them to be interested in extrapair sex. Perhaps when
single men are asked about their willingness to have extrapair sex,
their answers are relatively meaningless because they have no
specific partner on whom they would cheat, leading T not to be
associated with their responses (and to the interaction we ob-
served). This view implies no differential effect of relationship
status on T across different sorts of men. Inconsistent with the
view, however, are findings on the SOI. The content of the SOI
largely concerns sexual attitudes and sexual history. Were it sim-
ply the case that men who have high T are more likely to cheat
when in relationships, their T should covary with the SOI when
they are single, but we did not observe that pattern of results. This
view also suggests that more experienced, single men with low T
should be more likely to have been in relationships than those with
high T, but we did not find that the percentage of time more
experienced, single men have been in a relationship the past 3
years predicts their T levels. Further support for the moderating
effect of men’s sociosexual interests on the influence of relation-
2
We followed up our main analyses with parallel analyses by using the
individual components of extrapair sexual interest. The two-way interac-
tion between extrapair sexual interest and relationship status was most
strongly carried by willingness to have extrapair sex, F(1, 63) 5.95, p
.02; this interaction fell short of significance for both the SOI and history
of extrapair sex. The variation across measures could possibly merely be
sampling variability, however, and hence we focus on the main analysis by
using the composite measure and do not interpret differences in results
across measures.
Table 4
Study 2 General Linear Model Analysis of Testosterone (T)
Levels in Paired and Single Men
Variable
Paired men Single men
F(1, 21) Partial rF(1, 39) Partial r
Age 2.42 .32 0.80 .14
Relationship type 2.21
a
.42
a
Extrapair interests 5.49* .45 1.30 .18
Note. n 37 and 44, respectively. T is logged. For all variables except
relationship type, directed tests were used to obtain p values.
a
The F testing relationship type has degrees of freedom 2, 21. The effect
size for relationship type is a partial multiple r.
* p .05.
648
MCINTYRE ET AL.
ship status on T could come from a longitudinal study of men’s T
as they enter and depart from relationships, of the sort performed
by Mazur and Michalek (1998), but which also includes a measure
of sociosexual interests.
If T is modulated by social circumstances, there may well be a
functional reason for it to be modulated. T production presumably
facilitates performance demanded by the circumstances that pro-
mote it (or, conversely, hinders performance demanded by the
circumstances that lead to reduced T levels). As noted at the outset,
T is widely thought to modulate allocation of mating effort, with
increased T levels promoting male mating effort (social competi-
tion between men, particularly in sociosexual circumstances) and
diminished T levels promoting other forms of effort (e.g., parental
effort) at the expense of mating effort. The long-term effects of T
on muscle growth illustrate these effects. Over shorter intervals, T
probably affects performance in more subtle ways through psy-
chological pathways (e.g., motivation, attention, allocation of cog-
nitive effort). A recent study showing that men who had lower T
levels responded with greater sympathy and alertness to infant
cries (Fleming et al., 2002) may illustrate these effects. Similarly,
lower T levels may promote investment in and attention to rela-
tionships and satisfying partner needs, and higher T levels may
promote greater attention to alternative mates and thereby lower
relationship investment.
Implications for an Understanding of Human Mating
More Broadly
These results may have broader implications for an understand-
ing of the evolution of human mating. In recent years, anthropol-
ogists and psychologists have debated the nature of the mating
system and selection pressures that forged human mating adapta-
tions. Much debate concerns the role of male parental effort and its
implications for sexual selection processes. Hawkes and her col-
leagues have proposed that men have not evolved to invest in
offspring for the sake of parental investment (i.e., because of
benefits to offspring). Rather, apparent paternal care (e.g., hunting
and provisioning of women and offspring) evolved through its
benefits in mate acquisition, not offspring quality per se, and hence
qualifies as mating effort, not parental effort (Hawkes & Bird,
2002; Hawkes, O’Connell, & Jones, 2001; O’Connell, Hawkes,
Lupo, & Jones, 2002). According to this view, then, sexual selec-
tion processes (selection on individuals’ ability to acquire mates)
are fundamental to understanding any aspect of human male mat-
ing adaptation, including ones in which T is involved. Others have
argued that substantial paternal effort is key to understanding
adaptations involved in human reproduction (e.g., Kaplan, Hill,
Lancaster, & Hurtado, 2000; see also Ellison, 2001). In this view,
sexual selection may have played some role in forging male
sex-specific adaptations, but many sex-specific adaptations may be
due to the nature of sex-specific parental investments. For in-
stance, men’s greater musculature (as facilitated by T) may have
been maintained at least partly because men have evolved to
specialize in provisioning through hunting, given the constraints
women experience because of childbearing and child care. Wood
and Eagly (2002) proposed a model “that does not assume that any
sexual selection pressures that contributed to physical dimorphism
between the sexes are major influences on sex-typed psychological
attributes” (p. 702). Instead, they propose, behavioral sex differ-
ences are the result of sexual divisions of labor imposed by
physical differences (e.g., women’s role in childbearing).
3
Relat
-
edly, Miller and Fishkin (1997) argued that
although short-term mating strategies may be a fall-out from a failure
of humans to interface with their adapted-for environments, seeking a
long-term mate for a close and enduring relationship is based on
universal design features (i.e., part of our evolutionary heritage). (pp.
228 –229)
Yet others have argued that, though male parental care has been
important in human history, both men and women exhibit “mixed”
evolved strategies that involve seeking long-term mateships as
well as opportunistically seeking short-term (e.g., extrapair) rela-
tionships contingent on circumstances (e.g., Buss & Schmitt, 1993;
Gangestad & Simpson, 2000; Trivers, 1972).
A primary source of information about ancestral selection pro-
cesses is contained within the nature of the organism they have
shaped: Ancestral selection pressures leave telltale “footprints” in
the very nature of organismic adaptive design they shape (e.g.,
Andrews, Gangestad, & Matthews, 2002; Thornhill, 1997). The
current findings, in conjunction with previous research, may con-
tribute to our understanding of adaptations that regulate T. As
shown previously, men’s T levels are reduced when they are mated
and even more so when they have children. In conjunction with
evidence that lower T facilitates parental care (Fleming et al.,
2002), these findings suggest that human men, like some birds, do
indeed have special biological adaptations that promote parental
investment by diverting reproductive efforts away from mate seek-
ing (Gray, Chapman, et al., 2004). The current research adds a
potentially important caveat: Men who are paired do not always
have lower T. When men continue to be interested in pursuing
mates even when mated, they do not have lower T levels. These
findings thereby suggest that men also have adaptations that pro-
mote contingent, opportunistic short-term mating. Overall, then,
the pattern of findings points to roles for both paternal care and
mixed mating strategies in the shaping of male adaptations affect-
ing T.
Though current findings are suggestive, no doubt more research
is needed to fully assess the nature and design of the adaptations
that regulate male T levels. Interesting research questions include
the following: (a) How are T levels of mated men affected by the
sex ratio in the population, which affects men’s mating opportu-
nities? (b) Does extrapair sexual interest affect the T levels even of
men who become fathers, or does it only affect T levels in less
committed relationships? (c) Do T levels in mated men change as
a function of their perceived mating opportunities and success? For
example, as mated men find they have fewer opportunities because
of altered circumstances or attractiveness, do their T levels accord-
ingly drop?
3
It should be noted that Wood and Eagly (2002) did not argue this point
on the basis of a view that human men have evolved to provision offspring
but rather from a view that men and women have different societal roles in
a division of labor. Indeed, they cite Hawkes’ work showing that men have
little control over what they provide their own families—which is some-
what perplexing given the highly contrasting views these authors have
concerning the role of sexual selection in shaping male mating adaptations.
649
MATING, TESTOSTERONE, AND EXTRAPAIR INTEREST
Summary
That relationship status is associated with male T, at least in
North American populations, is now fairly well established. This
association may be due to the fact that men reduce T and thereby
allocate effort away from mate seeking when they are mated. On
the basis of the idea that some men may engage in mate seeking
even when mated, the current research explored whether this
association is moderated by male interest in pursuing mates, even
when paired. The fact that this appears to be the case suggests
complexity in the biosocial regulation of male T. The adaptive
nature of this complexity may be fundamental to an understanding
of human romantic and sexual relationships at both proximate and
ultimate levels of explanation.
References
Andrews, P. A., Gangestad, S. W., & Matthews, D. (2002). Adaptation-
ism—How to carry out an exaptationist program. Behavioral and Brain
Sciences, 25, 489 –504.
Archer, J., Birring, S. S., & Wu, F. C. W. (1998). The association between
testosterone and aggression among young men: Empirical findings and
a meta-analysis. Aggressive Behavior, 24, 411–420.
Bancroft, J. (2002). Biological factors in human sexuality. Journal of Sex
Research, 39, 15–21.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
51, 1173–1182.
Basaria, S., Leib, J., Tang, A. M., DeWeese, T. Carducci, M., Eisenberger,
M., & Dobs, A. S. (2002). Long-term effects of androgen deprivation
therapy in prostate cancer patients. Clinical Endocrinology, 56, 779
786.
Berg, S. J., & Wynne-Edwards, K. E. (2001). Changes in testosterone,
cortisol, and estradiol levels in men becoming fathers. Mayo Clinic
Proceedings, 76, 582–592.
Bernhardt, P. C., Dabbs, J. M., Fielden, J. A., & Lutter, C. D. (1998).
Testosterone changes during vicarious experiences of winning and los-
ing among fans at sporting events. Physiology and Behavior, 65, 5962.
Bhasin, S. (2003). Regulation of body composition by androgens. Journal
of Endocrinological Investigation, 26, 814822.
Book, A. S., Starzyk, K. B., & Quinsey, V. L. (2001). The relationship
between testosterone and aggression: A meta-analysis. Aggression and
Violent Behavior, 6, 579 –599.
Booth, A., & Dabbs, J. M. (1993). Testosterone and mens’ marriages.
Social Forces, 72, 463– 477.
Burnham, T. C., Chapman, J. F., Gray, P. B., McIntyre, M. H., Lipson,
S. F., & Ellison, P. T. (2003). Men in committed, romantic relationships
have lower testosterone. Hormones and Behavior, 44, 119 –122.
Buss, D. M., & Schmitt, D. P. (1993). Sexual Strategies Theory: A
contextual evolutionary analysis of human mating. Psychological Re-
view, 100, 204 –232.
Campbell, B. C., Lukas, W. D., & Campbell, K. L. (2001). The reproduc-
tive ecology of male immune function. In P. T. Ellison (Ed.), Repro-
ductive ecology and human evolution (pp. 159–178). New York: Aldine
de Gruyter.
Dabbs, J. M. (1997). Testosterone, smiling, and facial appearance. Journal
of Nonverbal Behavior, 21, 45–55.
Dabbs, J. M., Bernieri, F. J., Strong, R. K., Campo, R., & Milun, R. (2001).
Going on stage: Testosterone in greetings and meetings. Journal of
Research in Personality, 35, 27–40.
Dabbs, J. M., & Mohammed, S. (1992). Male and female salivary testos-
terone concentrations before and after sexual activity. Physiology and
Behavior, 52, 195–197.
de Waal, F. B. M. (1986). The integration of dominance and social bonding
in primates. Quarterly Review of Biology, 61, 459 479.
de Waal, F. B. M., & Hoekstra, J. A. (1980). Contexts and predictability of
aggression in chimpanzees. Animal Behaviour, 28, 929 –937.
Elias, M. (1981). Serum cortisol, testosterone, and testosterone-binding
globulin responses to competitive fighting in human males. Aggressive
Behavior, 7, 215–224.
Ellison, P. T. (2001). On fertile ground: A natural history of reproduction.
Cambridge, MA: Harvard University Press.
Fleming, A. S., Corter, C., Stallings, J., & Steiner, M. (2002). Testosterone
and prolactin are associated with emotional responses to infant cries in
new fathers. Hormones and Behavior, 42, 399 413.
Gangestad, S. W., & Simpson, J. A. (2000). The evolution of human
mating: Trade-offs and strategic pluralism. Behavioral and Brain Sci-
ences, 23, 675– 687.
Gaulin, S. J., & Sailer, L. D. (1984). Sexual dimorphism in weight among
the primates: The relative impact of allometry and sexual selection.
International Journal of Primatology, 5, 515–535.
Granger, D. A., Booth, A., & Johnson, D. R. (2000). Human aggression
and enumerative measures of immunity. Psychosomatic Medicine, 62,
583–590.
Granger, D. A., Schwartz, E. B., Booth, A., & Arentz, M. (1999). Salivary
testosterone determination in studies of child health and development.
Hormones and Behavior, 35, 18–27.
Gray, P. B., Campbell, B. C., Marlowe, F. W., Lipson, S. F., & Ellison,
P. T. (2004). Social variables predict between-subject but not day-to-day
variation in the testosterone of US men. Psychoneuroendocrinology, 29,
1153–1162.
Gray, P. B., Chapman, J. F., Burnham, T. C., McIntyre, M. H., Lipson,
S. F., & Ellison, P. T. (2004). Human male pair bonding and testoster-
one. Human Nature—An Interdisciplinary Biosocial Perspective, 15,
119 –131.
Gray, P. B., Kahlenberg, S. M., Barrett, E. S., Lipson, S. F., & Ellison, P. T.
(2002). Marriage and fatherhood are associated with lower testosterone
in males. Evolution and Human Behavior, 23, 193–201.
Hawkes, K., & Bird, R. B. (2002). Showing off, handicap signaling, and
the evolution of men’s work. Evolutionary Anthropology, 11, 58 67.
Hawkes, K., O’Connell, J. F., & Jones, N. G. B. (2001). Hadza meat
sharing. Evolution and Human Behavior, 22, 113–142.
Kaplan, H., Hill, K., Lancaster, J., & Hurtado, A. M. (2000). A theory of
human life history evolution: Diet, intelligence, and longevity. Evolu-
tionary Anthropology, 9, 156 –185.
Klein, S. L. (2000). The effects of hormones on sex differences in infec-
tion: From genes to behavior. Neuroscience and Biobehavioral Reviews,
24, 627– 638.
Martin, R. D. (1980). Sexual dimorphism and the evolution of higher
primates. Nature, 287, 273–275.
Mazur, A., & Booth, A. (1998). Testosterone and dominance in men.
Behavioral and Brain Sciences, 21, 353–397.
Mazur, A., Booth, A., & Dabbs, J. M. (1992). Testosterone and chess
competition. Social Psychology Quarterly, 55, 70–77.
Mazur, A., & Lamb, T. A. (1980). Testosterone, status, and mood in human
males. Hormones and Behavior, 14, 236–246.
Mazur, A., & Michalek, J. (1998). Marriage, divorce, and male testoster-
one. Social Forces, 77, 315–330.
Mazur, A., Susman, E. J., & Edelbrock, S. (1997). Sex difference in
testosterone response to a video game contest. Evolution and Human
Behavior, 18, 317–326.
Miller, L. C., & Fishkin, S. A. (1997). On the dynamics of human bonding
and reproductive success: Seeking windows on the adapted-for human-
environmental interface. In J. A. Simpson & D. T. Kenrick (Eds.),
Evolutionary social psychology (pp. 197–235). Mahwah, NJ: Erlbaum.
Muller, M. N., & Wrangham, R. W. (2004). Dominance, aggression and
650
MCINTYRE ET AL.
testosterone in wild chimpanzees: A test of the “challenge hypothesis.”
Animal Behaviour, 67, 113–123.
Nelson, R. J. (2000). An introduction to behavioral endocrinology. Sun-
derland, MA: Sinauer.
Nunes, S., Fite, J. E., & French, J. A. (2000). Variation in steroid hormones
associated with infant care behaviour and experience in male marmosets
(Callithrix kuhlii). Animal Behaviour, 60, 1–9.
Nunes, S., Fite, J. E., Patera, K. J., & French, J. A. (2001). Interactions
among paternal behavior, steroid hormones, and parental experience in
male marmosets (Callithrix kuhlii). Hormones and Behavior, 39, 70 82.
O’Connell, J. F., Hawkes, K., Lupo, K. D., & Jones, N. G. B. (2002). Male
strategies and Plio-Pleistocene archaeology. Journal of Human Evolu-
tion, 43, 831– 872.
Pope, H. G., Kouri, E. M., & Hudson, J. I. (2000). Effects of supraphysi-
ologic doses of testosterone on mood and aggression in normal men.
Archives of General Psychiatry, 57, 133–140.
Rice, W. R., & Gaines, S. D. (1994). ‘Heads I win, tails you lose’: Testing
directional alternative hypotheses in ecological and evolutionary re-
search. Trends in Ecology and Evolution, 9, 235–237.
Roney, J. R., Mahler, S. V., & Maestripieri, D. (2003). Behavioral and
hormonal responses of men to brief interactions with women. Evolution
and Human Behavior, 24, 365–375.
Schroeder, E. T., Singh, A., Bhasin, S., Storer, T. W., Azen, C., Davidson,
T., et al. (2003). Effects of an oral androgen on muscle and metabolism
in older, community-dwelling men. American Journal of Physiology—
Endocrinology and Metabolism, 284, E120–E128.
Simpson, J. A., & Gangestad, S. W. (1991). Individual differences in
sociosexuality: Evidence for convergent and discriminant validity. Jour-
nal of Personality and Social Psychology, 60, 870 883.
Storey, A. E., Walsh, C. J., Quinton, R. L., & Wynne-Edwards, K. E.
(2000). Hormonal correlates of paternal responsiveness in new and
expectant fathers. Evolution and Human Behavior, 21, 79 –95.
Suay, F., Salvador, A., Gonzalez-Bono, E., Sanchis, C., Martinez, M.,
Martinez-Sanchis, S., et al. (1999). Effects of competition and its out-
come on serum testosterone, cortisol and prolactin. Psychoneuroendo-
crinology, 24, 551–566.
Thornhill, R. (1997). The concept of an evolved adaptation. In M. Daly
(Ed.), Characterizing human psychological adaptations (pp. 4 –13).
London: Wiley.
Thornhill, R., & Gangestad, S. W. (1999). The scent of symmetry: A
human sex pheromone that signals fitness? Evolution and Human Be-
havior, 20, 175–201.
Trivers, R. L. (1972). Parental investment and sexual selection. In B.
Campbell (Ed.), Sexual selection and the descent of man (pp. 136–179).
Hawthorne, NY: Aldine de Gruyter.
van Honk, J., Tuiten, A., van den Hout, M., Koppeschaar, H., Thijssen, J.,
de Haan, E., & Verbaten, R. (2000). Conscious and preconscious selec-
tive attention to social threat: Different neuroendocrine response pat-
terns. Psychoneuroendocrinology, 25, 577–591.
van Honk, J., Tuiten, A., Verbaten, R., van den Hout, M., Koppeschaar, H.,
Thijssen, J., & de Haan, E. (1999). Correlations among salivary testos-
terone, mood, and selective attention to threat in humans. Hormones and
Behavior, 36, 17–24.
Wallen, K. (2001). Sex and context: Hormones and primate sexual moti-
vation. Hormones and Behavior, 40, 339–357.
Wood, W., & Eagly, A. H. (2002). A cross-cultural analysis of the behavior
of men and women: Implications of the origins of sex differences.
Psychological Bulletin, 128, 699 –727.
Received September 10, 2004
Revision received December 8, 2005
Accepted December 17, 2005
651
MATING, TESTOSTERONE, AND EXTRAPAIR INTEREST
... These authors point to the role of glucocorticoids in the physiological stress response, in which energetic resources are diverted toward increased cardiovascular and muscular performance at the expense of processes that can safely be disrupted, including reproduction-directed behaviors (Moore et al., 2016). Only two studies have investigated whether previously identified positive associations between T and both sexual desire (van Anders, 2012; Davis, 2000;Shirazi et al., 2019) and sociosexuality (interest in uncommitted sex; Edelstein et al., 2011;Gettler et al., 2019;Mcintyre et al., 2006;Puts et al., 2015;Shirazi et Content courtesy of Springer Nature, terms of use apply. Rights reserved. ...
Article
Full-text available
Objective The dual-hormone hypothesis (DHH) posits that some effects of testosterone on human behavior and psychology related to status-seeking are moderated by cortisol, such that they are stronger when cortisol levels are low. In support of the DHH, studies have found that cortisol negatively moderated the relationship between testosterone and such traits as status-seeking and interest in uncommitted sex. Others indicate a positive moderating influence of cortisol in some cases. Here, we test whether two psychosexual indices—sexual desire and sociosexuality—meet the expectations of the DHH in a large sample of men and women. Method 646 women and 185 men attended lab sessions during which they provided saliva samples for hormonal analysis and responded to the Sociosexual Orientation Inventory-Revised and the Sexual Desire Inventory (180 women and 43 men returned for a second session approximately two months later). We quantified salivary hormone concentrations using ELISA and assessed within- and between-participant effects of hormones on psychosexual measures with mixed-effects models. Results We observed a positive interaction between within-subjects cortisol and testosterone in models of sexual desire in both men and women. For women, these effects emerged in models of general sexual desire and in models of the dyadic desire subscale and were robust to many analytical configurations. For men, the effects were limited to models of solitary desire, but were also robust to alternative analyses. We present data to quantify our risks of both type I and type II error. Conclusions Some of our results contrast with usual dual-hormone hypothesis predictions of negative interactions between testosterone and cortisol. We suggest several potential explanations for these results, including a positive feedback loop whereby elevated testosterone prompts increases in sexual desire and behavior, necessitating cortisol-induced mobilization of energy stores.
... Therefore, baseline measures of testosterone are considered fairly trait-like (Sellers, Mehl, & Josephs, 2007), making it possible to examine associations between testosterone and other constructs such as individual differences in personality and behavior. For example, people with lower levels of testosterone tend to report lower interest in extra-dyadic sex and have a lower likelihood of divorce compared to people with higher levels of testosterone (Edelstein et al., 2011;McIntyre et al., 2006). ...
Thesis
Testosterone is a steroid hormone that is important for close relationship processes (Edelstein & Chin, 2018). For instance, people who are single tend to have higher levels of testosterone compared to people who are in committed relationships (Mazur & Michalek, 1998), suggesting that testosterone lowers once one is in a relationship. Furthermore, lower testosterone might be functional for maintaining relationships: Both men and women who have lower testosterone report higher relationship quality (e.g., Edelstein, van Anders, Chopik, Goldey, & Wardecker, 2014). However, only a few studies have examined associations between testosterone and relationship quality in individuals, let alone in a sample of couples, and studies that include couples tend to have relatively small, homogeneous samples. Thus it is not yet clear whether similar testosterone-relationship quality links and any dyadic associations between partners would be found in other samples. The first goal of this study was to examine whether people reported higher relationship quality when they or their partners have lower testosterone. I also examined how people with lower testosterone behaved towards their partners. Given that lower testosterone is thought to be associated with nurturance and caregiving (van Anders, Goldey, & Kuo, 2011), the second goal of this study was to examine whether people with lower testosterone would be more likely to behave in more nurturant (i.e., prosocial) ways towards their partner. Finally, the third goal was to test prosocial behavior as a potential mechanism underlying testosterone-relationship quality links. To address these questions, I analyzed data from 595 heterosexual couples drawn from three samples (college-aged couples, couples with children, and newlywed couples) that included baseline measures of salivary testosterone, self-reports of relationship quality, and partner interactions that were coded for prosocial behavior. I found that, in the college-aged couples and couples with children, women who had lower testosterone indeed reported higher relationship quality and showed more prosocial behavior. In contrast to expectations, men in the newlywed sample who had lower testosterone reported lower relationship quality and engaged in less prosocial behavior. I also found dyadic associations: In the college-aged couples and couples with children, women who had lower testosterone had partners who reported higher relationship quality; in the newlywed sample, women who had lower testosterone had partners who reported lower relationship quality. I did not find that prosocial behavior accounted for any testosterone-relationship quality links, suggesting that people with lower testosterone felt better about their relationships, but not necessarily because they or their partners were more behaving in more prosocial ways. I discuss potential explanations for discrepant findings across samples: The newlywed couples knew prior to their lab session that they would be discussing a disagreement, which could have caused anticipatory increases in testosterone. This study advances social neuroendocrinology work by assessing the extent to which previous testosterone-relationship quality findings replicate in larger samples of couples and contributes important new information about the associations between testosterone and prosocial behavior.
... In total, 96 studies were selected (Lassek and Gaulin, 2009;Hughes and Gallup, 2003;Weeden and Sabini, 2007;Frederick and Haselton, 2007;Lukaszewski et al., 2014;Hill et al., 2013;Kordsmeyer et al., 2018;Peters et al., 2008;Boothroyd et al., 2017;Pawlowski et al., 2008;Arnocky et al., 2018;Rhodes et al., 2005;Van Dongen and Sprengers, 2012;Alvergne et al., 2009;Apicella, 2014;Apicella et al., 2007;Aronoff, 2017;Atkinson, 2012;Atkinson et al., 2012;Bogaert and Fisher, 1995;Booth et al., 1999;Boothroyd et al., 2011;Boothroyd et al., 2008;Charles and Alexander, 2011;Chaudhary et al., 2015;Edelstein et al., 2011;Falcon, 2016;Farrelly et al., 2015;Frederick, 2010;Frederick and Jenkins, 2015;Gallup et al., 2007;Genovese, 2008;Gettler et al., 2019;Gildner, 2018;(Gómez-Valdés et al., 2013) Hartl et al., 1982;Hoppler et al., 2018;Honekopp et al., 2007;Kirchengast, 2000;Kirchengast and Winkler, 1995;Klimas et al., 2019;Klimek et al., 2014;Kordsmeyer and Penke, 2017;Krzyżanowska et al., 2015;Kurzban and Weeden, 2005;Little et al., 1989;Loehr and O'Hara, 2013;Longman et al., 2018;Luevano et al., 2018;Maestripieri et al., 2014;Manning and Fink, 2008;Manning et al., 2003;Marczak et al., 2018;McIntyre et al., 2006;Međedović and Bulut, 2019;Mosing et al., 2015;Muller and Mazur, 1997;Nagelkerke et al., 2006;Nettle, 2002;Pawlowski et al., 2000;Pollet et al., 2011;Polo et al., 2019;Price et al., 2013;Prokop and Fedor, 2011;Prokop and Fedor, 2013;Puts et al., 2006;Puts et al., 2015;Putz et al., 2004;Rahman et al., 2005;Rosenfield et al., 2020;Schwarz et al., 2011;Scott and Bajema, 1982;Shoup and Gallup, 2008;Sim and Chun, 2016;Simmons and Roney, 2011;Smith et al., 2017;Sneade and Furnham, 2016;Sorokowski et al., 2013;Steiner, 2011;Stern et al., 2020;Strong, 2014;Strong and Luevano, 2014;Subramanian et al., 2009;Suire et al., 2018;Tao and Yin, 2016;van Anders et al., 2007;Varella et al., 2014;von Rueden et al., 2011;Voracek et al., 2010;Walther et al., 2016;Walther et al., 2017c;Walther et al., 2017a;Walther et al., 2017b;Waynforth, 1998;Winkler and Kirchengast, 1994;Honekopp et al., 2006), comprising 474 effect sizes from 99 samples and 177,044 unique participants (Table 1). This exceeds the number of studies for each of the meta-analyses published previously (Grebe et al., 2019;Van Dongen and Sprengers, 2012;von Rueden and Jaeggi, 2016;Xu et al., 2018). ...
Article
Full-text available
Humans are sexually dimorphic: men and women differ in body build and composition, craniofacial structure, and voice pitch, likely mediated in part by developmental testosterone. Sexual selection hypotheses posit that, ancestrally, more 'masculine' men may have acquired more mates and/or sired more viable offspring. Thus far, however, evidence for either association is unclear. Here, we meta-analyze the relationships between six masculine traits and mating/reproductive outcomes (96 studies, 474 effects, N = 177,044). Voice pitch, height, and testosterone all predicted mating; however, strength/muscularity was the strongest and only consistent predictor of both mating and reproduction. Facial masculinity and digit ratios did not significantly predict either. There was no clear evidence for any effects of masculinity on offspring viability. Our findings support arguments that strength/muscularity may be sexually selected in humans, but cast doubt regarding selection for other forms of masculinity and highlight the need to increase tests of evolutionary hypotheses outside of industrialized populations.
... In North American samples, men in long-term committed relationships have lower testosterone levels than other men (Mazur & Michalek, 1998), consistent with the prediction that these men are allocating less effort to competing for mates (Gray, Kahlenberg, Barrett, Lipson, & Ellison, 2002). Important tweaks to this relation provide further support: Men in committed relationships but who are open to an extra-pair affair have higher testosterone levels than their monogamous peers (McIntyre et al., 2006). ...
Chapter
Full-text available
Chapter
Wide-ranging and inclusive, this text provides an invaluable review of an expansive selection of topics in human evolution, variation and adaptability for professionals and students in biological anthropology, evolutionary biology, medical sciences and psychology. The chapters are organized around four broad themes, with sections devoted to phenotypic and genetic variation within and between human populations, reproductive physiology and behavior, growth and development, and human health from evolutionary and ecological perspectives. An introductory section provides readers with the historical, theoretical and methodological foundations needed to understand the more complex ideas presented later. Two hundred discussion questions provide starting points for class debate and assignments to test student understanding.
Chapter
Wide-ranging and inclusive, this text provides an invaluable review of an expansive selection of topics in human evolution, variation and adaptability for professionals and students in biological anthropology, evolutionary biology, medical sciences and psychology. The chapters are organized around four broad themes, with sections devoted to phenotypic and genetic variation within and between human populations, reproductive physiology and behavior, growth and development, and human health from evolutionary and ecological perspectives. An introductory section provides readers with the historical, theoretical and methodological foundations needed to understand the more complex ideas presented later. Two hundred discussion questions provide starting points for class debate and assignments to test student understanding.
Chapter
Wide-ranging and inclusive, this text provides an invaluable review of an expansive selection of topics in human evolution, variation and adaptability for professionals and students in biological anthropology, evolutionary biology, medical sciences and psychology. The chapters are organized around four broad themes, with sections devoted to phenotypic and genetic variation within and between human populations, reproductive physiology and behavior, growth and development, and human health from evolutionary and ecological perspectives. An introductory section provides readers with the historical, theoretical and methodological foundations needed to understand the more complex ideas presented later. Two hundred discussion questions provide starting points for class debate and assignments to test student understanding.
Chapter
Wide-ranging and inclusive, this text provides an invaluable review of an expansive selection of topics in human evolution, variation and adaptability for professionals and students in biological anthropology, evolutionary biology, medical sciences and psychology. The chapters are organized around four broad themes, with sections devoted to phenotypic and genetic variation within and between human populations, reproductive physiology and behavior, growth and development, and human health from evolutionary and ecological perspectives. An introductory section provides readers with the historical, theoretical and methodological foundations needed to understand the more complex ideas presented later. Two hundred discussion questions provide starting points for class debate and assignments to test student understanding.
Article
Vocal characteristics are important cues to form social impressions. Previous studies indicated that men with masculine voices are perceived as engaging in higher rates of infidelity and being less committed to their relationship. In the current study, we examined how women in China perceive information regarding infidelity and relationship investment conveyed by the voices (voice pitch and vocal tract length) of males, and whether different vocal characteristics play a similar role in driving these impressions. In addition, we examined whether these perceptions are consistent in Chinese and English language contexts. The results indicated that women perceived men with more masculine voices (lower voice pitch and longer vocal tract length) as showing a lower likelihood of infidelity and higher relationship investment; further, women who preferred more masculine voices in long-term relationships, but not in short-term relationships, were more likely to perceive men with masculine voices as less likely to engage in infidelity and more likely to invest in their relationship. Moreover, the participants formed very similar impressions irrespective of whether the voices spoke native (Chinese) or foreign (English) languages. These results provide new evidence for the role of the voice in women's choices in selecting long-term partners.
Chapter
Research on testosterone has long been dominated by a focus on “high testosterone” behaviors, such as aggression, competition, and dominance. The vast majority of this work, including in humans, has also been conducted in exclusively male samples, based in part on presumed links between testosterone and masculinity. Yet testosterone is implicated in many psychological and interpersonal processes for both men and women, and “low testosterone” behaviors may be particularly critical for ongoing close relationships. This fairly narrow focus on high testosterone, in men, leaves major gaps in our understanding of the social neuroendocrinology of close relationships, particularly as related to positive processes like caregiving, support-seeking, and intimacy. The goal of this review is to integrate the literature on testosterone in close relationships, in both men and women, with an eye toward closeness, intimacy, and other positive processes that likely contribute to and are supported by individual differences in testosterone and changes in testosterone over time. I focus on testosterone in the context of romantic and parent-child relationships, and highlight directions for future research that can help to fill important gaps in this literature. Further, I argue that, because close relationships are inherently dynamic and dyadic, longitudinal research that includes both men and women, and ideally both couple members, is critical for a complete understanding of the role of testosterone in close relationship processes.
Article
Full-text available
This article evaluates theories of the origins of sex differences in human behavior. It reviews the cross-cultural evidence on the behavior of women and men in nonindustrial societies, especially the activities that contribute to the sex-typed division of labor and patriarchy. To explain the cross-cultural findings, the authors consider social constructionism, evolutionary psychology, and their own biosocial theory. Supporting the biosocial analysis, sex differences derive from the interaction between the physical specialization of the sexes, especially female reproductive capacity, and the economic and social structural aspects of societies. This biosocial approach treats the psychological attributes of women and men as emergent given the evolved characteristics of the sexes, their developmental experiences, and their situated activity in society.
Article
Full-text available
In this article, we attempt to distinguish between the properties of moderator and mediator variables at a number of levels. First, we seek to make theorists and researchers aware of the importance of not using the terms moderator and mediator interchangeably by carefully elaborating, both conceptually and strategically, the many ways in which moderators and mediators differ. We then go beyond this largely pedagogical function and delineate the conceptual and strategic implications of making use of such distinctions with regard to a wide range of phenomena, including control and stress, attitudes, and personality traits. We also provide a specific compendium of analytic procedures appropriate for making the most effective use of the moderator and mediator distinction, both separately and in terms of a broader causal system that includes both moderators and mediators. (46 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Article
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Article
In a study of possible links between testosterone and dominance, 119 men and 114 women provided saliva samples for testosterone assay and posed smiling and not smiling for portrait photographs. Expert judges viewing the photographs found smaller smiles among high than low testosterone men, with less zygomatic major (raising the corners of the mouth) and orbicularis oculi (raising the cheeks and crinkling around the corners of the eyes) muscle activity. Naive judges viewing individual photographs gave higher potency ratings to smiling high testosterone men than smiling low testosterone men. Naive judges viewing photographs grouped into high and low testosterone sets gave higher potency and lower goodness ratings to high than to low testosterone men, regardless of whether they were smiling. Among women, judges found only slight relationships between testosterone and facial appearance. The pattern among men of less smiling with higher testosterone levels fits with research linking testosterone to face-to-face dominance.
Article
This research explored the relation of endogenous testosterone levels to behavior in brief social encounters. In four studies, 358 college students whose testosterone levels were known entered a room and (1) stood and spoke to a video camera, (2) stood and talked with an experimenter, (3) sat and talked with an interviewer, or (4) sat and talked with a peer. High-testosterone students entered more quickly, focused more directly on their targets, and displayed a more forward and independent manner. Results were similar for men and women. Correlates of testosterone are visible in thin slices of everyday behavior lasting only a few seconds. The effect of this behavior on a social interaction partner remains to be determined.
Article
The hormone testosterone (T) has a central role in recent theories about allocation of status ranks during face-to-face competition. It has been methodologically convenient to test the hypothesized T mechanism in physically taxing athletic contests, where results have been supportive, although their generalizability to normal social competition is questionable. Competition among chess players is a step closer to normal social competition because it does not require physical struggle, and it is the arena for tests of the T mechanism which are reported here. We find that winners of chess tournaments show higher T levels than do losers. Also, in certain circumstances, competitors show rises in T before their games, as if in preparation for the contests. These results generally support recent theories about the role of T in the allocation of status ranks.
Article
In order to study the hormonal correlates of the tradeoff between mating and parenting effort in human males, we examined the salivary testosterone (T) levels of 58 Boston-area men who were either unmarried (n=29), married without children (n=14), or married with children (n=15). Additionally, we asked participants to complete a questionnaire that surveyed their demographic, marital, and parenting backgrounds. We tested the hypotheses that (1) T levels will be lower in married than in unmarried men and (2) married men with children will have lower T levels than unmarried men and married men without children. We also tested a series of hypotheses relating variation in parenting and spousal relationships to T. We found that married men with and without children had significantly lower evening T than unmarried men. No significant differences in T were found among the groups in morning samples. Among married men without children, higher scores on a “spousal investment” measure and more hours spent with a man's wife on his last day off work were both associated with lower T levels. We suggest that lower T levels during the day among fathers may facilitate paternal care in humans by decreasing the likelihood that a father will engage in competitive and/or mating behavior.