Vasopressin selectively impairs emotion recognition
Florina Uzefovskya, Idan Shalevb, Salomon Israela, Ariel Knafoa,
Richard P. Ebsteinb,c,*
aPsychology Department, Hebrew University, Jerusalem 91501, Israel
bNeurobiology, Hebrew University, Jerusalem 91501, Israel
cPsychology Department, National University of Singapore, Singapore
Received 19 January 2011; received in revised form 24 July 2011; accepted 26 July 2011
At the core of human social behavior is the ability to
empathize, that is, the ability to accurately perceive
another’s emotions and to respond to them in an appropriate
way (Davis, 1980). Empathy, which molds the fabric of social
interactions and underlies all forms of social communication,
has been hypothesized to drive prevalent social processes
Facial expressions are salient representatives of our emo-
tions and an important focus of social communication
appears to be a robust indicator of empathy (Baron-Cohen
et al., 1999). Accordingly, the ‘Reading the Mind in the Eyes
Test’ (RMET), employed in the current investigation, is a
widely used instrument for evaluating emotion recognition
ability in both nonclinical and clinical settings (Baron-Cohen
et al., 1999).
In the past several years the neural substrates and
specific brain regions underlying empathy have become
better understood. Neural circuits involved in empathy
and emotion recognition have been localized primarily in
Psychoneuroendocrinology (2012) 37, 576—580
Reading the Mind in the
and to respond to them appropriately, are only recently becoming better understood. This report
focuses on the nonapeptide arginine-vasopressin (AVP), which plays an important role in
modulating social behavior in animals, especially promoting aggressive behavior. Towards
clarifying the role of AVP in human social perception we used the Reading of the Mind in the
Eyes Test and intranasal administration of AVP to show that AVP leads to a significant decrease in
emotion recognition. Moreover, when comparing photos of males vs. females, all viewed by
males, AVP had an effect on gender-matched photos only. Furthermore, the effect of AVP was
restricted to recognition of negative emotions while leaving recognition of positive emotions
and empathy, a core element in all human social interactions.
# 2011 Elsevier Ltd. All rights reserved.
The biological mechanisms underlying empathy, the ability to recognize emotions
* Corresponding author at: Psychology Department, Faculty of Arts
and Sciences, Block AS4, 9 Arts Link, National University of
Singapore, Singapore 117570.
Tel.: +65 91140010; fax: +65 6773 1843.
E-mail address: firstname.lastname@example.org (R.P. Ebstein).
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/psyneuen
0306-4530/$ — see front matter # 2011 Elsevier Ltd. All rights reserved.
the limbic system, prefrontal cortex, and frontoparietal
networks (for a meta-analysis see Lamm et al., 2010).
Despite these impressive advances in understanding the
neuroanatomical circuitry of empathy, much less is known
about the underlying neurochemical basis for this salient
construct. Arginine vasopressin (AVP) is an important neuro-
modulator of activation in various brain areas involved in
emotion regulation, including the limbic system (subgenual
cingulate cortex, amygdala) (Zink et al., 2010). Conse-
quently, AVP is considered a prime candidate to modulate
human social perception and communication.
Most of the knowledge on the effects of AVPin modulating
primarily linked to a host of social behaviors in males (Carter
the brains ofmale vs. female voles (Bamshad et al., 1993) and
even humans (Ishunina and Swaab, 1999). Mating induces
et al., 1994), this further implicates AVP in mating-induced
changes in male voles’ social behaviors. As part of the post-
mating male social behavior, AVP causes aggression upon
intrusions to one’s territory. Moreover, AVP antagonists
injected centrally inhibit creation of mate preference and
aggressive behaviors towards intruders, while infusion of AVP
induces partner preference behavior even without mating
(Winslow et al., 1993) and induces long lasting tendencies
towards male-to-male aggression (Stribley and Carter, 1999).
AVP’s role as the initiator of male social behavior is further
supported by the finding that androgens control AVP synthesis
in several brain areas (De Vries et al., 1994).
Despite these advances in animal models, much less is
known about the role of AVP in human social behavior. Our
initial studies of AVP and human social cognition have pre-
sented evidence for a role of the arginine vasopressin 1a
receptor gene in a host of social behaviors in nonclinical
subjects (reviewed in Ebstein et al., 2010). Studies using
intranasal administration (INA) to investigate the role of AVP
inhumansocialcognition arescarce,but theworkdonesofar
suggest a similar effect to the one found in rodents, that is,
AVP has a sex-specific effect on social interactions, specifi-
cally promoting aggression in males. Cerebrospinal fluid
levels of AVP were positively correlated with aggressive
behaviors in subjects suffering from personality disorders
(Coccaro et al., 1998). Notably, this correlation was stronger
in males than in females. Thompson et al. showed that INA of
AVP stimulates agonistic facial motor patterns in response to
the faces of unfamiliar men and decreases perceptions of the
friendliness of those faces. In contrast, in women, AVPhad an
opposite effect of increasing perceptions of the friendliness
of those faces (Thompson et al., 2006). Taken together, these
findings point to a major role of AVP in modulating social
communication. Given that AVP seems to have a more pro-
minent role in male-typical behavior, we have conducted this
study on males, hypothesizing that AVP would selectively
impair emotion recognition.
Participants were 39 male Israeli university students (mean
age = 24.69, SD = 2.52), recruited by word of mouth and
advertisements on campus bulletin boards for a study on
neurobiology and personality. Selection criteria stipulated
that all subjects are males aged 18—35, have no self-report
history of psychiatric disorders, chronic illness or drug taking
and are non-smokers.
All sessions were scheduled for a fixed time-window
(14:00—18:00) to control for circadian rhythm effects.
Additionally, participants were asked to refrain from phy-
sical activity, eating, or drinking (except for water) for 2 h
prior to the experiment. At the beginning of the session,
participants were provided with an explanation of the
procedure and the physiological effects of the drug. Upon
receiving the explanations, subjects were asked to read and
sign an informed consent form. The project was approved
by the S. Herzog Hospital IRB committee and the Israeli
Ministry of Health. Importantly, there were no reports of
adverse side effects or water retention reported by parti-
cipants. All subjects were reimbursed for participating in
In a between-subject, double-blind design, participants
were randomly assigned to placebo or treatment groups.
Participants self-administered into both nostrils a 250 ml
solution containing 20IU of AVP (diluted in 0.9% NaCl, Sigma,
Germany) or placebo (sterile saline, 0.9% NaCl) using a
medical dropper. For maximum absorption, participants
did this seated and tilting their heads backward, with the
experimenter present. After a 45 min waiting period to allow
the hormone to take effect, the experiment began (Thomp-
son et al., 2006). A computerized version of RMET (Baron-
Cohen et al., 1999) was administered on a 2200computer
screen.Thetest comprisesof36photographs oftheeyearea,
and one practice photo. For each photo, participants were
asked to choose among 4 possible words one which best suits
the emotion displayed in the photo. Before seeing each test
picture, participants were able to see the four possible
emotions from which they would have to choose later. The
emotions presented were variable and demanded a complex
understanding of the different shades of human emotions.
For example, the target emotions included desire, fantasiz-
ing, and despondent. A few of the target words were
repeated, but for the most part, there was no repetition
of target emotions. Including the foils, 93 mental states are
represented in the RMET, with 27 different target emotions
(the correct response), of which 21 were presented only
once. The photographs were balanced for sex (Baron-Cohen
et al., 2001). To ensure that participants could make
informed selections, definitions for each emotion were pro-
vided by clicking on each word. After presenting all four
alternatives on the screen, participants clicked on a ‘Ready’
button which causedthe picture toappear,and then selected
the most appropriate emotion to describe the photo.
Statistical analysis was conducted using SPSS version 15.
We first examined the overall effect of treatment (AVP vs.
predictor and found, as predicted, that the AVP group made
p = 0.034,D = 0.72).ResultsareillustratedinFig.1.
AVP selectively impairs emotion recognition in men 577
Given the known gender-specific effect of AVP (Thomp-
son et al., 2006) we hypothesized that the gender of the
individuals photographed might be a factor in how AVP
affects emotion recognition. The RMET includes photo-
graphs of both men and women. Gender recognition of
pictures has been validated in previous experiments
(Baron-Cohen et al., 2001), as well as in our mini-pilot
of 5 participants, achieving 100% accuracy of sex differ-
entiation. When scores were stratified by the sex of the
observed pictures, the effect of AVP was not found when
p = 0.510). In contrast, in gender-matched photos; that
is when males observed male photographs, a large
(D = 0.91) effect was found for the administration of AVP
(t(37)= ?2.77, p = 0.009) (Fig. 2, panel a). Under placebo,
subjects were better at identifying emotions in males than
in females (paired t-test; t(16)= 2.03, p = 0.029), however,
under AVP, this difference disappeared (paired t-test;
t(19)= ?0.52, p = 0.608).
We further hypothesized that the valence of the emotion
exhibited would play a role in how AVP affects emotion
recognition, since AVP was shown to induce aggressive emo-
tions in men (Thompson et al., 2006). We therefore sorted
the RMET pictures by those displaying negative vs. positive
emotions. The effect of AVPon reducing emotion recognition
was only observed for negative emotions (t(37)= ?2.38,
p = 0.023) and not for positive emotions (t(37)= 0.05,
p = 0.960) (Fig. 2, panel b).
This study provides an important step in understanding the
role of AVP in modulating human social communication and
demonstrates that AVP has a specific effect on emotion
recognition in males. AVP negates the increase in men’s
ability to recognize the emotions of other men vs. other
women. Importantly, this finding echoes previous knowledge
derived from animal models showing a role of AVP in aggres-
sion related male social behavior. For example, in male
prairie voles, mating-mediated AVP release induces aggres-
sion towards other males and strange females (Winslow
et al., 1993). In voles, aggression was directed towards both
sexes, while in this report the effect was limited to photos of
males. In humans, physical aggressive behavior by males is
more common than by females, and it is more commonly
directed towards other males. Notably, feelings of aggression
which increase the propensity for a physical attack are
accompanied by diminished ability to empathize with the
possible victim (Preston and de Waal, 2003). Therefore, AVP
would be hypothesized to diminish the ability of a male to
recognize the emotion of other males, and not of females, as
is reflected in our current findings.
recognition across the board, men receiving AVP significantly
misinterpret negative emotions but not positive emotions.
Whereas AVP enhances mistakes regarding negative emo-
tions, suchas upset, accusing,hostility,and suspicion, under-
standing of positive emotions remained unaffected by
treatment. The lack of effect of AVP on positive emotions
reinforces the notion that this hormone resonates more with
negative social behavior, emotions and feelings.
INA AVP promotes aggressive behavioral responses to
social stimuli (i.e. faces), as well as autonomic responsive-
ness to social threat and anxiety associated with angry facial
expressions (Thompson et al., 2006), and enhances social
stress (Shalev et al., 2011). The current report strengthens
the role of AVP in male emotional perception and extends
recognition, considered a core element in all human social
A recent report (Zink et al., 2010) showed, in healthy
human subjects, that INA of AVP affects activity and con-
nectivity in the cortical component of a medial prefrontal
cortex—amygdala circuit. In addition, the limbic system has
been implicated in empathy as well (Lamm et al., 2010).
Percent of errors in RMET (?SE) by treatment (AVP
(b) Percent of errors (?SE) by the valence of the displayed emotion.
Percent of errors in RMET by treatment (AVPand placebo). (a) Percent of errors (?SE) by the sex of the observed pictures.
578 F. Uzefovsky et al.
Specifically, the task used in this report, RMET, has been
shown to increase amygdala activation (Baron-Cohen et al.,
1999). Taken together, the findings might suggest that the
limbic system is AVP’s location of action on empathy and
AVP affects empathy through limbic system processes.
It is important to discuss also the role of oxytocin (OT) a
neuropeptide closely related to AVP, which differs from it by
only two amino acids, and has also been linked to various
social behaviors. Importantly, INA of OT has been found to
elevate performance in the RMET for males (Domes et al.,
2006). Taken together, the current report suggests that the
two hormones affect social cognition and behavior in oppo-
site directions. Strengthening the notion that OT and AVP
have opposite effects on social cognition are recent neuro-
physiological studies. In a recent study extracellular single-
unit recordings were used to identify two major populations
of neurons in the central amygdala, one excited by vasopres-
sin but inhibited by oxytocin, the other only excited by
oxytocin and unresponsive to vasopressin (Viviani and Stoop,
2008). Future studies would benefit from looking at both
hormones simultaneously to understand how they interact
and in what context each comes into play.
In our study we used the RMET, as it is a widely used test
of emotion recognition, with high face validity (Baron-
Cohen et al., 2001). Despite the documented strengths of
this test, for current purposes a limitation emerged, as the
photographs were not counter-balanced across emotional
valence and sex (meaning, there were more photos of males
exhibiting negative emotions than women exhibiting nega-
tive emotions), this, taken together with the limited num-
ber of items in each category, did not allow for a parametric
test of interaction effects. Therefore we were unable to
analyze potential interaction effects of AVP with both sex
and valence on performance accuracy. Another issue
emerged with respect to the directionality of AVP’s effect
on emotion recognition. We conducted an informal analysis
to see whether AVP increased negative judgments (mistake
of valence) or whether AVP diminished the fine-grained
ability to distinguish among negative valence emotions,
but no specific pattern of errors emerged. In addition,
there was no evidence for a consistent bias in the erroneous
response (e.g. no higher likelihood of erroneously identify-
ing an emotion as angry). Because the foils were not
systematically construed to accommodate this kind of ana-
lysis, future work should clarify this finding by examining
the specific consequence of AVP treatment on emotion
The current report extends previous findings suggesting
that AVP may influence male aggression by predisposing
individuals to respond to negative emotional stimuli with
lack of empathy, possibly exhibiting a propensity towards
aggression. Given the differential effect of AVP on males vs.
females, as well as the consistent findings of gender differ-
ences in levels of empathy, future studies should investigate
whether the effect of AVP on males described in this report
would be dissimilar to its effect on females.
Several psychological disorders involve social impair-
ments, including highly aggressive behavior (e.g. personality
disorder, conduct disorder) and lack of empathy (autism), in
which there is a marked difficulty in accurate interpretation
of emotions. The current results can provide an insight into
the biological underpinnings of such disorders which cause
serious impairments in day-to-day social functioning.
Role of funding source
This research was supported in part by Autism Speaks (R.P .E.)
and by a fellowship from the Arianne de Rothschild Program
(F.U.). None had any further role instudy design; inthe collec-
tion, analysis and interpretation of data; in the writing of the
Author F.U designed the study, wrote the protocol, under-
took statistical analysis and wrote the first draft of the paper.
Author I.S. helped with protocol construction and assisted in
writing the paper. Author S.I. participated in writing the
paper and in statistical analysis. Authors A.K. and R.E. wrote
the paper. All authors contributed to and have approved the
Conflict of interest statement
All authors declare that they have no conflict of interests.
We are grateful to all the participants and to the research
assistants who recruited them. We are also grateful to Shany
Edelman for her helpful comments on the manuscript.
Bamshad, M., Novak, M.A., De Vries, G.J., 1993. Sex and Species
Differences in the Vasopressin Innervation of Sexually Naive and
Parental Prairie Voles, Microtus ochrogaster and Meadow Voles
Microtus pennsylvanicus, vol. 5. Blackwell Publishing Ltd., pp.
Baron-Cohen, S., Ring, H.A., Wheelwright, S., Bullmore, E.T., Bram-
mer, M.J., Simmons, A., Williams, S.C.R., 1999. Social Intelli-
gence in the Normal and Autistic Brain: An fMRI Study, vol. 11.
Blackwell Science Ltd., pp. 1891—1898.
Baron-Cohen, S., Wheelwright, S., Hill, J., Raste, Y., Plumb, I., 2001.
The ‘‘Reading the Mind in the Eyes’’ test revised version: a study
with normal adults, and adults with Asperger syndrome or high-
functioning autism. J. Child Psychol. Psychiatry 42 (2), 241—251.
Batson, C.D., Batson, J.G., Slingsby, J.K., Harrell, K.L., Peekna,
H.M., Todd, R.M., 1991. Empathic joy and the empathy-altruism
hypothesis. J. Pers. Soc. Psychol. 61 (3), 413—426.
Carter, C.S., Grippo, A.J., Pournajafi-Nazarloo, H., Ruscio, M.G.,
Porges, S.W., Inga, D.N., Rainer, L., 2008. Oxytocin, Vasopressin
and Sociality Progress in Brain Research, vol. 170. Elsevier, pp.
Coccaro, E.F., Kavoussi, R.J., Hauger, R.L., Cooper, T.B., Ferris, C.F.,
1998. Cerebrospinal fluid vasopressin levels: correlates with
aggression and serotonin function in personality-disordered sub-
jects. Arch. Gen. Psychiatry 55 (8), 708—714.
Davis, M., 1980. A multidimensional approach to individual differ-
ences in empathy. JSAS Catalog Select. Doc. Psychol. 10 (4), 85.
De Vries, G., Wang, Z., Bullock, N., Numan, S., 1994. Sex differences
in the effects of testosterone and its metabolites on vasopressin
messenger RNA levels in the bed nucleus of the stria terminalis of
rats. J. Neurosci. 14 (3), 1789.
Domes, G., Heinrichs, M., Michel, A., Berger, C., Herpertz, S.C.,
2006. Oxytocin improves ‘‘Mind-Reading’’ in humans. Biol. Psy-
AVP selectively impairs emotion recognition in men579
Ebstein, R., Israel, S., Chew, S., Zhong, S., Knafo, A., 2010. Genetics Download full-text
of human social behavior. Neuron 65 (6), 831—844.
Ishunina, T., Swaab, D., 1999. Vasopressin and oxytocin neurons of
the human supraoptic and paraventricular nucleus; size changes
in relation to age and sex. J. Clin. Endocrinol. Metabol. 84 (12),
Lamm, C., Decety, J., Singer, T., 2010. Meta-analytic evidence for
common and distinct neural networks associated with directly
experienced pain and empathy for pain. Neuroimage.
Preston, S., de Waal, F., 2003. Empathy: its ultimate and proximate
bases. Behav. Brain Sci. 25 (01), 1—20.
Shalev, I., Israel, S., Uzefovsky, F., Gritsenko, I., Kaitz, M., Ebstein,
R.P., 2011. Vasopressin needs an audience: neuropeptide elicited
stress responses are contingent upon perceived social evaluative
threats. Horm. Behav. 60 (1), 121—127.
Stribley, J.M., Carter, C.S., 1999. Developmental exposure to vaso-
pressin increases aggression in adult prairie voles. Proc. Natl.
Acad. Sci. 96 (22), 12601—12604.
Thompson, R.R., George, K., Walton, J.C., Orr, S.P., Benson, J., 2006.
tion. Proc. Natl. Acad. Sci. U. S. A. 103 (20), 7889—7894.
Viviani, D., Stoop, R., 2008. Opposite effects of oxytocin and vaso-
pressin on the emotional expression of the fear response. Prog.
Brain Res. 170, 207—218.
Wang, Z., Smith, W., Major, D., De Vries, G., 1994. Sex and species
differences in the effects of cohabitation on vasopressin messen-
ger RNA expression in the bed nucleus of the stria terminalis in
prairie voles (Microtus ochrogaster) and meadow voles (Microtus
pennsylvanicus). Brain Res. 650 (2), 212—218.
Winslow, J.T., Hastings, N., Carter, C.S., Harbaugh, C.R., Insel, T.R.,
1993. A role for central vasopressin in pair bonding in monoga-
mous prairie voles. Nature 365 (6446), 545—548.
Zink, C.F., Stein, J.L., Kempf, L., Hakimi, S., Meyer-Lindenberg, A.,
2010. Vasopressin modulates medial prefrontal cortex—amygdala
circuitry during emotion processing in humans. J. Neurosci. 30
580 F. Uzefovsky et al.