Does Rejection Hurt? An fMRI Study of Social Exclusion

Abstract

A neuroimaging study examined the neural correlates of social exclusion and tested the hypothesis that the brain bases of
social pain are similar to those of physical pain. Participants were scanned while playing a virtual ball-tossing game in
which they were ultimately excluded. Paralleling results from physical pain studies, the anterior cingulate cortex (ACC) was
more active during exclusion than during inclusion and correlated positively with self-reported distress. Right ventral prefrontal
cortex (RVPFC) was active during exclusion and correlated negatively with self-reported distress. ACC changes mediated the
RVPFC-distress correlation, suggesting that RVPFC regulates the distress of social exclusion by disrupting ACC activity.

Full text

DOI: 10.1126/science.1089134
, 290 (2003); 302Science et al.Naomi I. Eisenberger,
Exclusion
Does Rejection Hurt? An fMRI Study of Social
www.sciencemag.org (this information is current as of September 30, 2008 ):
The following resources related to this article are available online at
http://www.sciencemag.org/cgi/content/full/302/5643/290
version of this article at: including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/cgi/content/full/302/5643/290/DC1
can be found at: Supporting Online Material
found at: can berelated to this articleA list of selected additional articles on the Science Web sites
http://www.sciencemag.org/cgi/content/full/302/5643/290#related-content
http://www.sciencemag.org/cgi/content/full/302/5643/290#otherarticles
, 7 of which can be accessed for free: cites 17 articlesThis article
220 article(s) on the ISI Web of Science. cited byThis article has been
http://www.sciencemag.org/cgi/content/full/302/5643/290#otherarticles
37 articles hosted by HighWire Press; see: cited byThis article has been
http://www.sciencemag.org/cgi/collection/neuroscience
Neuroscience : subject collectionsThis article appears in the following
http://www.sciencemag.org/about/permissions.dtl
in whole or in part can be found at: this article permission to reproduce of this article or about obtaining reprintsInformation about obtaining
registered trademark of AAAS. is aScience2003 by the American Association for the Advancement of Science; all rights reserved. The title CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on September 30, 2008 www.sciencemag.orgDownloaded from

subtilis (fig. S2). The RLP of B. subtilis
includes both those amino acid residues of
RuBisCO that are responsible for binding
the phosphate on C1 of RuBP and those
required for activation by CO
2
. However,
the residues of RuBisCO that are responsi-
ble for binding the other phosphate group
of RuBP and the residues of loop 6, which
are essential for RuBisCO activity (2,3),
are replaced by different amino acids in
RLP (Fig. 1B). The reaction catalyzed by
RuBisCO consists of three sequential, par-
tial reactions: enolization, carboxylation or
oxygenation, and hydrolysis (2,3,26). De-
letion of loop 6 from RuBisCO prevents it
from catalyzing the carboxylation/oxygen-
ation reactions (27). However, it retains the
ability to catalyze the enolization reaction
(27). This observation supports the hypoth-
esis that the RLP-catalyzed enolization of
DK-MTP-1-P does not require the amino
acid residues that bind the phosphate group
on C5 of RuBP and the loop 6. Moreover,
the structure of DK-MTP-1-P is very sim-
ilar to that of RuBP. In photosynthetic
RuBisCO, these additional structures may
hinder the DK-MTP-1-P enolase reaction,
and they may also explain the slow growth
of ykrW
–
/rbcL
⫹
cells (Fig. 4C). In this
context, our results with the RLP of B.
subtilis suggest that RLPs of other bacteria
may also catalyze a reaction similar to one
of the partial reactions of RuBisCO in a
bacterial metabolic pathway.
Our analysis shows that RLP of B. sub-
tilis functions as a DK-MTP-1-P enolase,
which has no RuBP-carboxylation activity,
in the methionine salvage pathway. More-
over, this function of RLP is conserved in
the RuBisCO from a photosynthetic bacte-
rium. In a standard phylogenetic tree of the
large subunits of RuBisCO, the RLP from
B. subtilis is not included on any branches
that include RuBisCO or on branches that
include other RLPs with RuBP-carboxyla-
tion activity (Fig. 1A). The codon usage
and the G ⫹C content of the gene for RLP
are typical of the organism. The literature
(28) suggests that genes such as the gene
for RLP were probably not derived by lat-
eral transfer of a gene for a RuBP-carbox-
ylating enzyme from another unrelated or-
ganism, for example, in this case, an ar-
chaeon or photosynthetic bacterium. Thus,
it is possible that the gene for RLP, which
in B. subtilis is part of the methionine
salvage pathway, and the gene for photo-
synthetic RuBisCO originated from a com-
mon ancestral gene (supporting online
text). However, bacteria and Archaea that
have RLPs first appeared on Earth (29)
long before the Calvin cycle developed in
photosynthetic bacteria (30), thus we sug-
gest that RLPs may be the ancestral en-
zymes of photosynthetic RuBisCO.
References and Notes
1. R. J. Ellis, Trend. Biochem. Sci. 4, 241 (1979).
2. T. J. Andrews, G. H. Lorimer, in Biochemistry of Plants,
vol. 10, M. D. Hatch, N. K. Boardman, Eds. (Academic
Press, New York, 1987), pp. 131–218.
3. H. Roy, T. J. Andrews, in Photosynthesis, vol. 9, R. C.
Leegood, T. D. Sharkey, S. von Caemmerer, Eds. (Klu-
wer, Dordrecht, Netherlands, 2000), pp. 53–83.
4. G. M. Watson, F. R. Tabita, FEMS Microbiol. Lett.146,
13 (1997).
5. G. M. Watson, J. Yu, F. R. Tabita, J. Bacteriol.181,
1569 (1999).
6. S. Ezaki, N. Maeda, T. Kishimoto, H. Atomi, T.
Imanaka, J. Biol. Chem.274, 5078 (1999).
7. F. Kunst et al., Nature 390, 249 (1997).
8. T. E. Hanson, F. R. Tabita, Proc. Natl. Acad. Sci. U.S.A.
98, 4397 (2001).
9. J. A. Eisen et al., Proc. Natl. Acad. Sci.U.S.A.99, 9509
(2002).
10. H. P. Klenk et al., Nature 390, 364 (1997).
11. J. F. Grundy, M. T. Henkin, in Bacillus subtilis and Its
Closest Relatives: from Genes to Cells, A. L. Sonen-
shein et al., Eds. (ASM Press, Washington, DC, 2002),
pp. 245–254.
12. B. A. Murphy, F. J. Grundy, T. M. Henkin, J. Bacteriol.
184, 2314 (2002).
13. A. Sekowska, A. Danchin, BMC Microbiol.2, 8 (2002).
14. E. S. Furfine, R. H. Abeles, J. Biol. Chem. 263, 9598
(1988).
15. R. W. Myers, J. W. Wray, S. Fish, R. H. Abeles, J. Biol.
Chem.268, 24785 (1993).
16. J. W. Wray, R. H. Abeles, J. Biol. Chem. 270, 3147
(1995).
17. Y. Dai, T. C. Pochapsky, R. H. Abeles, Biochemistry 40,
6379 (2001).
18. J. Heilbronn, J. Wilson, B. J. Berger, J. Bacteriol. 181,
1739 (1999).
19. Materials and Methods are available as supporting
material on Science Online.
20. A. Sekowska, L. Mulard, S. Krogh, J. K. Tse, A. Danchin,
BMC Microbiol.1, 15 (2001).
21. G. Avigad, Methods Enzymol.41, 27 (1975).
22. N. C. Kyrpides, C. R. Woese, Proc. Natl. Acad. Sci.
U.S.A.95, 224 (1998).
23. Y. R. Chen, F. C. Hartman, J. Biol. Chem.270, 11741
(1995).
24. H. Ashida et al., unpublished observations.
25. V. Vagner, E. Dervyn, S. D. Ehrlich, Microbiology 144,
3097 (1998).
26. W. W. Cleland, T. J. Andrews, S. Gutteridge, F.C.
Hartman, G. H. Lorimer, Chem. Rev. 98, 549 (1998).
27. E. M. Larson, F. W. Larimer, F. C. Hartman, Biochem-
istry 34, 4531 (1995).
28. I. Moszer, E. P. Rogha, A. Danchin, Curr. Opin. Micro-
biol.2, 524 (1999).
29. C. R. Woese, O. Kandler, M. L. Wheelis, Proc. Natl.
Acad. Sci. U.S.A. 87, 4576 (1990).
30. H. Hartman, Origins Life Evol. Biosphere 28, 515
(1998).
31. We thank W. L. Ogren and A. R. Portis Jr., for review-
ing the manuscript. We also thank M. Inui, RITE, for
providing the plasmid pRR2119, and J. Tsukamoto for
assistance with mass analysis. This study was sup-
ported by a Grant-in-Aid for Scientific Research (no.
10460043) from the Ministry of Education, Science,
Sports and Culture of Japan, and by the “Research for
the Future” programs (JSPS-RFTF97R16001 and
JSPS-00L01604) of the Japan Society for the Promo-
tion of Science.
Supporting Online Material
www.sciencemag.org/cgi/content/full/302/5643/286/
DC1
Materials and Methods
SOM Text
Figs. S1 and S2
References
19 May 2003; accepted 26 August 2003
Does Rejection Hurt? An fMRI
Study of Social Exclusion
Naomi I. Eisenberger,
1
* Matthew D. Lieberman,
1
Kipling D. Williams
2
A neuroimaging study examined the neural correlates of social exclusion and
tested the hypothesis that the brain bases of social pain are similar to those
of physical pain. Participants were scanned while playing a virtual ball-
tossing game in which they were ultimately excluded. Paralleling results
from physical pain studies, the anterior cingulate cortex (ACC) was more
active during exclusion than during inclusion and correlated positively with
self-reported distress. Right ventral prefrontal cortex (RVPFC) was active
during exclusion and correlated negatively with self-reported distress.
ACC changes mediated the RVPFC-distress correlation, suggesting that
RVPFC regulates the distress of social exclusion by disrupting ACC
activity.
It is a basic feature of human experience to
feel soothed in the presence of close others
and to feel distressed when left behind.
Many languages reflect this experience in
the assignment of physical pain words
(“hurt feelings”) to describe experiences of
social separation (1). However, the notion
that the pain associated with losing some-
one is similar to the pain experienced upon
physical injury seems more metaphorical
than real. Nonetheless, evidence suggests
that some of the same neural machinery
recruited in the experience of physical pain
may also be involved in the experience of
pain associated with social separation or
1
Department of Psychology, Franz Hall, University of
California, Los Angeles, Los Angeles, CA 90095–1563,
USA.
2
Department of Psychology, Macquarie Univer-
sity, Sydney NSW 2109, Australia.
*To whom correspondence should be addressed. E-
mail: neisenbe@ucla.edu
REPORTS
10 OCTOBER 2003 VOL 302 SCIENCE www.sciencemag.org290
on September 30, 2008 www.sciencemag.orgDownloaded from

rejection (2). Because of the adaptive value
of mammalian social bonds, the social at-
tachment system, which keeps young near
caregivers, may have piggybacked onto the
physical pain system to promote survival
(3). We conducted a functional magnetic
resonance imaging (fMRI) study of social
exclusion to determine whether the regions
activated by social pain are similar to those
found in studies of physical pain.
The anterior cingulate cortex (ACC) is
believed to act as a neural “alarm system”
or conflict monitor, detecting when an au-
tomatic response is inappropriate or in con-
flict with current goals (4–6). Not surpris-
ingly, pain, the most primitive signal that
“something is wrong,”activates the ACC
(7,8). More specifically, dorsal ACC ac-
tivity is primarily associated with the affec-
tively distressing rather than the sensory
component of pain (7–9).
Because of the importance of social
bonds for the survival of most mammalian
species, the social attachment system may
have adopted the neural computations of
the ACC, involved in pain and conflict
detection processes, to promote the goal of
social connectedness. Ablating the cingu-
late in hamster mothers disrupts maternal
behavior aimed at keeping pups near (10),
and ablating the cingulate in squirrel mon-
keys eliminates the spontaneous production
of the separation cry, emitted to reestablish
contact with the social group (11). In hu-
man mothers, the ACC is activated by the
sound of infant cries (12). However, to
date, no studies have examined whether the
ACC is also activated upon social separa-
tion or social rejection in human subjects.
Right ventral prefrontal cortex (RVPFC)
has been implicated in the regulation or
inhibition of pain distress and negative af-
fect (13–16). The primate homolog of
VPFC has efferent connections to the re-
gion of the ACC associated with pain dis-
tress (17,18), suggesting that RVPFC may
partially regulate the ACC. Additionally,
electrical stimulation of VPFC in rats di-
minishes pain behavior in response to pain-
ful stimulation (19). More recently in hu-
mans, heightened RVPFC activation has
been associated with improvement of pain
symptoms in a placebo-pain study (16).
Given that even the mildest forms of
social exclusion can generate social pain
(20), we investigated the neural response
during two types of social exclusion: (i)
explicit social exclusion (ESE), in which
individuals were prevented from participat-
ing in a social activity by other players
engaged in the social activity, and (ii) im-
plicit social exclusion (ISE), in which par-
ticipants, because of extenuating circum-
stances, were not able to join in a social
activity with other players.
fMRI scans were acquired while partic-
ipants played a virtual ball-tossing game
(“CyberBall”) with what they believed to
be two other players, also in fMRI scan-
ners, during which the players eventually
excluded the participant (21). In reality,
there were no other players; participants
were playing with a preset computer pro-
gram and were given a cover story to en-
sure that they believed the other players
were real (22).
In the first scan (ISE), the participant
watched the other “players”play Cyber-
Ball. Participants were told that, because of
technical difficulties, the link to the other two
scanners could not yet be made and thus, at
first, they would be able to watch but not play
with the other two players. This cover story was
intended to allow participants to view a
scene visually identical to ESE without par-
ticipants believing they were being exclud-
ed. In the second scan (inclusion), partici-
pants played with the other two players. In
the final scan (ESE), participants received
seven throws and were then excluded when
the two players stopped throwing partici-
pants the ball for the remainder of the scan
(⬃45 throws). Afterward, participants
filled out questionnaires assessing how ex-
cluded they felt and their level of social
distress during the ESE scan (22).
Behavioral results indicated that partic-
ipants felt ignored and excluded during
ESE (t⫽5.33, P⬍0.05). As predicted,
group analysis of the fMRI data indicated
that dorsal ACC (Fig. 1A) (x⫽–8, y⫽20,
z⫽40) was more active during ESE than
during inclusion (t⫽3.36, r⫽0.71, P⬍
0.005) (23,24). Self-reported distress was
positively correlated with ACC activity in
this contrast (Fig. 2A) (x⫽–6, y⫽8, z⫽
45, r⫽0.88, P⬍0.005; x⫽–4, y⫽31,
z⫽41, r⫽0.75, P⬍0.005), suggesting
that dorsal ACC activation during ESE was
associated with emotional distress parallel-
ing previous studies of physical pain (7,8).
The anterior insula (x⫽42, y⫽16, z⫽1)
was also active in this comparison (t⫽
4.07, r⫽0.78, P⬍0.005); however, it was
not associated with self-reported distress.
Two regions of RVPFC were more ac-
tive during ESE than during inclusion (Fig.
1B) (x⫽42, y⫽27, z⫽–11, t⫽4.26, r⫽
0.79, P⬍0.005; x⫽37, y⫽50, z⫽1, t⫽
4.96, r⫽0.83, P⬍0.005). Self-reported
distress was negatively correlated with
RVPFC activity during ESE, relative to
inclusion (Fig. 2B) (x⫽30, y⫽34, z⫽–3,
r⫽–0.68, P⬍0.005). Additionally,
RVPFC activation (x⫽34, y⫽36, z⫽–3)
was negatively correlated with ACC activ-
ity (x⫽–6, y⫽8, z⫽45) during ESE,
relative to inclusion (r⫽–0.81, P⬍0.005)
(Fig. 2C), suggesting that RVPFC may play
a self-regulatory role in mitigating the dis-
tressing effects of social exclusion.
ACC activity (x⫽–6, y⫽8, z⫽45)
mediated the direct path from RVPFC (x⫽
34, y⫽36, z⫽–3) to distress (Sobel test,
Z⫽3.16, P⬍0.005). After controlling for
ACC activity, the remaining path from
RVPFC to distress was no longer signifi-
cant (␤⫽–0.17, P⬎0.5). This mediation-
al model is nearly identical to the results
from previous research on the self-regula-
tion of physical pain (16).
ISE, relative to inclusion, also produced
significant activation of ACC (x⫽–6, y⫽
21, z⫽41; (z⫽41, t⫽4.34, I ⫽0.78, P⬍
0.005). To preserve the cover story, self-
reported distress was not assessed after this
condition, and thus we could not assess the
relation between ACC activity during ISE
and perceived distress. However, no
RVPFC activity was found in this compar-
ison, even at a P⫽.05 significance level,
suggesting that the ACC registered this ISE
but did not generate a self-regula-
tory response.
In summary, a pattern of activations very
similar to those found in studies of physical
pain emerged during social exclusion, pro-
viding evidence that the experience and reg-
ulation of social and physical pain share a
common neuroanatomical basis. Activity in
dorsal ACC, previously linked to the experi-
ence of pain distress, was associated with
increased distress after social exclusion. Fur-
thermore, activity in RVPFC, previously
linked to the regulation of pain distress, was
associated with diminished distress after so-
cial exclusion.
The neural correlates of social pain were
also activated by the mere visual appear-
Fig. 1. (A) Increased
activity in anterior cin-
gulate cortex (ACC)
during exclusion rela-
tive to inclusion. (B) In-
creased activity in
right ventral prefron-
tal cortex (RVPFC)
during exclusion rela-
tive to inclusion.
REPORTS
www.sciencemag.org SCIENCE VOL 302 10 OCTOBER 2003 291
on September 30, 2008 www.sciencemag.orgDownloaded from

ance of exclusion in the absence of actual
exclusion. The pattern of neural activity
associated with ISE and ESE provides
some challenges to the way we currently
understand exclusion and its consequences.
Although the neural correlates of distress
were observed in both ISE and ESE, the
self-regulation of this distress only oc-
curred in response to ESE. Explicit aware-
ness of exclusion may be required before
individuals can make appropriate attribu-
tions and regulate the associated distress.
Dorsal ACC activation during ESE
could reflect enhanced attentional process-
ing, previously associated with ACC activ-
ity (4,5), rather than an underlying distress
due to exclusion. Two pieces of evidence
make this possibility unlikely. First, ACC
activity was strongly correlated with per-
ceived distress after exclusion, indicating
that the ACC activity was associated with
changes in participants’self-reported feel-
ing states. Second, although inclusion is
likely to require greater attentional process-
ing than does ISE to facilitate participation
in the game, there was greater ACC activity
during ISE than during inclusion, indicat-
ing that the ACC activity was not fully
attributable to heightened attention.
Because of the need to maintain a realistic
situation in which participants would genu-
inely feel excluded, the study did not contain
some of the controls typical of most neuro-
imaging studies. For instance, the conditions
were always implemented in the same order
so as to keep expectations consistent from
one scan to the next across participants. It
was especially critical that ESE came last to
prevent expectations of possible exclusion
from contaminating the other conditions.
There was only a single ESE period to pre-
serve ecological validity. This modification,
however, diminishes, rather than increases,
the likelihood of Type I errors.
This study suggests that social pain is
analogous in its neurocognitive function to
physical pain, alerting us when we have
sustained injury to our social connections,
allowing restorative measures to be taken.
Understanding the underlying commonalities
between physical and social pain unearths
new perspectives on issues such as why phys-
ical and social pain are affected similarly by
both social support and neurochemical inter-
ventions (2,3,25), and why it “hurts”to lose
someone we love (1).
References and Notes
1. M. R. Leary, C. A. Springer, in Behaving Badly: Aversive
Behaviors in Interpersonal Relationships, R. Kowalski,
Ed. (American Psychological Association, Washing-
ton, DC, 2001), pp. 151–175.
2. J. Panksepp, Affective Neuroscience (Oxford Univer-
sity Press, New York, 1998).
3. E. Nelson, J. Panksepp, Neurosci. Biobehav. Rev. 22,
437 (1998).
4. G. Bush, P. Luu, M. I. Posner, Trends Cognit. Sci. 4, 215
(2000).
5. C. S. Carter et al.,Proc. Natl. Acad. Sci. U.S.A. 97,
1944 (2000).
6. T. A. Kimbrell et al.,Biol. Psychol. 46, 454 (1999).
7. P. Rainville, G. H. Duncan, D. D. Price, B. Carrier, M. C.
Bushnell, Science 277, 968 (1997).
8. N. Sawamoto et al.,J. Neurosci. 20, 7438 (2000).
9. E. L. Foltz, E. W. White, J. Neurosurg. 19, 89 (1962).
10. M. R. Murphy, P. D. MacLean, S. C. Hamilton, Science
213, 459 (1981).
11. P. D. MacLean, J. D. Newman, Brain Res. 450, 111
(1988).
12. J. P. Lorberbaum et al.,Depress. Anxiety 10,99
(1999).
13. P. Petrovic, E. Kalso, K. M. Petersson, M. Ingvar,
Science 295, 1737 (2002).
14. A. R. Hariri, S. Y. Bookheimer, J. C. Mazziotta, Neuro-
report 11, 43 (2000).
15. D. M. Small, R. J. Zatorre, A. Dagher, A. C. Evans, M.
Jones-Gotman, Brain 124, 1720 (2001).
16. M. D. Lieberman et al., in preparation.
17. S. T. Carmichael, J. L. Price, J. Comp. Neurol. 363, 615
(1995).
18. C. Cavada, T. Company, J. Tejedor, R. J. Cruz-Rizzolo,
F. Reinoso-Suarez, Cerebr. Cort. 10, 220 (2000).
19. Y. Q. Zhang, J. S. Tang, B. Yuan, H. Jia, Pain 72, 127
(1997).
20. L. Zadro, K. D. Williams, reported in K. D. Williams,
T. I. Case, C. L. Govan, in Responding to the Social
World: Implicit and Explicit Processes in Social Judg-
ments and Decisions, J. Forgas, K. Williams, W. von
Hippel, Eds. (Cambridge University Press, New York,
in press).
21. K. D. Williams, C. Cheung, W. Choi, J. Personal. Soc.
Psychol. 79, 748 (2000).
22. Materials and methods are available as supportive
material on Science Online.
23. The correction for multiple comparisons was carried
out with an uncorrected Pvalue of 0.005 and a
cluster size threshold of 10, corresponding to a per-
voxel false positive probability of less than 0.000001
(24).
24. S. D. Forman et al.,Magn. Reson. Med. 33, 636
(1995).
25. J. L. Brown, D. Sheffield, M. R. Leary, M. E. Robinson,
Psychosom. Med. 65, 276 (2003).
26. We thank A. Satpute and the staff of the University
of California, Los Angeles, Brain Mapping Center
for their assistance. This research was funded by a
grant from the National Institutes of Mental
Health (R21MH66709-01) to M.D.L.
Supporting Online Material
www.sciencemag.org/cgi/content/full/302/5643/290/
DC1
Materials and Methods
14 July 2003; accepted 15 August 2003
Fig. 2. Scatterplots showing the
relation during exclusion, rela-
tive to inclusion, between (A)
ACC activity and self-reported
distress, (B) RVPFC and self-
reported distress, and (C) ACC
and RVPFC activity. Each point
represents the data from a single
participant.
REPORTS
10 OCTOBER 2003 VOL 302 SCIENCE www.sciencemag.org292
on September 30, 2008 www.sciencemag.orgDownloaded from