ORIGINAL ARTICLE

The Neurobiology of Giving Versus Receiving

Support: The Role of Stress-Related and Social
Reward–Related Neural Activity
Tristen K. Inagaki, PhD, Kate E. Bryne Haltom, BA, Shosuke Suzuki, BA, Ivana Jevtic, BA,
Erica Hornstein, MA, Julienne E. Bower, PhD, and Naomi I. Eisenberger, PhD

ABSTRACT
Objectives: There is a strong association between supportive ties and health. However, most research has focused on the
health benefits that come from the support one receives while largely ignoring the support giver and how giving may con-
tribute to good health. Moreover, few studies have examined the neural mechanisms associated with support giving or how
giving support compares to receiving support.
Method: The current study assessed the relationships: a) between self-reported receiving and giving social support and vul-
nerability for negative psychological outcomes and b) between receiving and giving social support and neural activity to
socially rewarding and stressful tasks. Thirty-six participants (mean [standard deviation] age = 22.36 [3.78] years, 44%
female) completed three tasks in the functional magnetic resonance imaging scanner: 1) a stress task (mental arithmetic
under evaluative threat), b) an affiliative task (viewing images of close others), and c) a prosocial task.
Results: Both self-reported receiving and giving social support were associated with reduced vulnerability for negative psy-
chological outcomes. However, across the three neuroimaging tasks, giving but not receiving support was related to reduced
stress-related activity (dorsal anterior cingulate cortex [r = −0.27], left [r = −0.28] and right anterior insula [r = −0.33], and
left [r = −0.32] and right amygdala [r = −0.32]) to a stress task, greater reward-related activity (left [r = 0.42] and right ven-
tral striatum [VS; r = 0.41]) to an affiliative task, and greater caregiving-related activity (left VS [r = 0.31], right VS
[r = 0.31], and septal area [r = 0.39]) to a prosocial task.
Conclusions: These results contribute to an emerging literature suggesting that support giving is an overlooked contributor
to how social support can benefit health.
Key words: social support, helping, caregiving, providing social support, social relationships and health.

S ocial relationships are critical to health and well-being

(1–3). Those without close ties fare poorly across a
range of important outcomes (1,4,5) including mental and
to examine the associations between both receiving and
giving support with a) vulnerability to negative psycholog-
ical outcomes and b) neural activity to stressful and socially
physical health (2–6), whereas those with more social ties rewarding tasks.
show enhanced mental and physical health. To date, the re- Since the 1970s, studies have shown associations be-
search literature has focused on how “social support”—the tween supportive ties and reduced morbidity and mortality
perception or experience that one is loved and cared for by (8). Theories on social support suggest that these health ben-
others, esteemed and valued, and part of a social network of efits may come from the support we receive from others,
mutual assistance and obligations (7)—relates to positive which then helps to reduce stress responding (9–11). For
health outcomes. Because social support, by definition, fo- example, undergoing stressful tasks with a friend present
cuses more on the consequences of receiving support, stud-
ies linking social support and health have largely focused
on the support recipient while ignoring the individual giv- AI = anterior insula, dACC = dorsal anterior cingulate cortex,
ROI = region of interest, SA = septal area, VS = ventral striatum
ing support and care. Therefore, the following study aimed

From the Department of Psychology (Inagaki), University of Pittsburgh, Pittsburgh, Pennsylvania; and Department of Psychology (Byrne Haltom,
Suzuki, Jevtic, Hornstein, Bower, Eisenberger), University of California, Los Angeles, California.
Address correspondence and reprint requests to Tristen K. Inagaki, PhD, Department of Psychology, University of Pittsburgh, 210 South Bouquet St,
Pittsburgh, PA 15260. E-mail: [email protected]; Naomi I. Eisenberger, PhD, UCLA-Psych Soc, Box 951563, 4444 Franz Hall, Los Angeles, CA
90095-1563. E-mail: [email protected]
Received for publication May 22, 2015; revision received November 4, 2015.
DOI: 10.1097/PSY.0000000000000302
Copyright © 2016 by the American Psychosomatic Society

Psychosomatic Medicine, V 78 • 443-453 443 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 ORIGINAL ARTICLE

(versus alone) leads to reduced cardiovascular responses Hence, the current study examined the associations be-
to the stressors ((12); see Ref. (13) for a review). In addi- tween both receiving support and giving support with neu-
tion, self-reports of social support have been associated ral activity in response to stressful and socially rewarding
with reduced distress-related neural activity (in the dorsal tasks. Specifically, we examined how both individual dif-
anterior cingulate cortex [dACC]) to an experience of social ferences in receiving and giving support related to neural
rejection (14,15) and receiving support during a painful activity in response to a stress task, an affiliative task (in
experience is associated with reduced pain-related neural which participants viewed images of their close others),
activity (16,17). Although the benefits of receiving sup- and a prosocial task.
port are well established, receiving support can some- There were two main goals. First, we examined whether
times be ineffective or, in worst cases, backfire (e.g., self-reported support receiving related to negative psycho-
Refs. (18,19)). Therefore, it is possible that the support that logical outcomes and neural activity to the three tasks.
one receives may not explain the relationship between so- The current literature on receiving social support and
cial support and health in its entirety. health-related outcomes suggests that receiving social sup-
There has been increasing interest in the support giver port is sometimes helpful, particularly at reducing stress,
and the health benefits that may be accrued by giving to and sometimes not. Therefore, receiving support was hy-
others. Thus, support giving, much like support receiving, pothesized to be associated with less negative psychologi-
is associated with a number of beneficial mental and phys- cal outcomes; however, given the mixed findings in the
ical health outcomes (20–24). Specifically, giving to others literature, no directional hypotheses were made for the
is associated with lower mortality rates over a 5-year period neural outcomes. Second, we examined the associations be-
(20), fewer sick days (24), and reduced cardiovascular ac- tween self-reported support giving and negative psycho-
tivity over a 24-hour period (22). The mechanisms leading logical outcomes and neural activity to the three tasks.
to support giving's beneficial effects may have evolved out Based on emerging findings that giving to others might
of a caregiving system that helped nurture and support in- be good for health and relationships, support giving was
fants and others in need (25–30). In particular, giving to hypothesized to be associated with less negative psycho-
others may rely on distinct neural regions that serve to rein- logical outcomes, decreased stress-related activity to the
force the caregiving behavior and are also involved in the stressful task, increased reward-related activity to viewing
regulation of stress responses (31,32). close others, and increased caregiving-related activity to
In fact, it has been shown that the neural regions in- the prosocial task.
volved in maternal caregiving behavior in animals and the
provision of social support in humans are also known for
processing basic rewards; these include the ventral striatum
(VS) and septal area (SA; (31,32)). Moreover, these care- METHODS
giving regions have inhibitory connections with regions
known to be involved in threat and stress responding such Participants
as the amygdala (33,34). This activity in caregiving regions Forty-seven individuals participated as part of a larger study on the neural
may dampen stress responding to facilitate adaptive care mechanisms associated with social support. The sample was then
during times of stress, which over time may have health constrained to those who completed all three neuroimaging tasks. A single
benefits. For example, giving support to a romantic partner outlier who was consistently greater than 3 standard deviations (SDs) away
from the mean (both below and above) across multiple tasks was removed
in need (versus not giving support) activated both the VS leaving a final sample of 36 participants (Table 1). The self-reported ethnic
and SA (32). Furthermore, one of these regions, the SA, composition included 8.3% black/African American, 22.2% Latino/
was negatively correlated with amygdala activity when Chicano, 33.3% white, 30.6% Asian, and 5.6% other. Procedures were
participants gave support to their partners, suggesting run between June and October of 2014 in accordance with institutional re-
that caregiving-related circuitry may help dampen stress view board guidelines, and all participants provided informed consent.
responding. Indeed, in a separate study, giving support
to a friend (versus a control condition) led to reduced
sympathetic-related responding to a laboratory-based stress
task (21). Together, these findings suggest that giving to TABLE 1. Participant Characteristics (n = 36)
others may be stress reducing on its own and therefore Variable Value
may be an overlooked contributor to the mental and physi-
cal health benefits of social support. Age, y 22.36 (3.78)
Still, little work has examined both receiving and giving Sex 16 women/20 men
support in terms of how they relate to stress-related neural Self-reported receiving support 23.74 (3.12)
activity. Furthermore, no studies have assessed social sup- Self-reported giving support 19.19 (3.46)
port and neural activity to positive, stress-free experiences.

Psychosomatic Medicine, V 78 • 443-453 444 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 Social Support and Neural Outcomes

SELF-REPORT MEASURES possible answers. The subsequent screen provided the par-
ticipant with feedback about whether or not they answered
Support Receiving and Giving correctly and their performance relative to the “typical
Support receiving and giving was assessed with the 21-item UCLA student.” Over time, the feedback indicated that
two-way social support scale (35). Questions are rated on a participants' performance (based on how quickly and ac-
0 (not at all) to 5 (always) scale, with higher numbers indi- curately they responded relative to the average student)
cating a greater amount of support received and given. became increasingly worse as the typical student's perfor-
Sample items to assess receiving support include the fol- mance grew better, thus amplifying the social evaluative
lowing: “when I am feeling down there is someone I can nature and uncontrollability of the situation.
lean on,” and “I have someone to help me if I am physically Participants completed a total of eight 56-second blocks
unwell.” Giving support is evaluated by items such as separated by 5 seconds of fixation crosshair: 4 practice
“I look for ways to cheer people up when they are feeling blocks and 4 test blocks. Within each block, eight math
down,” and “I have helped someone with their responsibil- problems were presented for 5 seconds each. For the prac-
ities when they were unable to fulfill them.” Participants re- tice blocks, math problems were followed by 2 seconds of
ported receiving more support than they gave (t(35) = 8.26, rest, but for the test blocks, math problems were followed
p < .001). Reports of receiving support were also moder- by 2 seconds of feedback (1 second indicating whether
ately associated with reports of support giving (r = 0.501, the participant correctly answered and 1 second indicating
p = .002), suggesting that both forms of social support are how they performed relative to others). After each block,
related but separate constructs. participants rated how stressful they found the previous
block (“how stressful was that”), rated from 1 (not at all
Vulnerability Index stressed) to 4 (extremely stressed). As expected, partici-
Because receiving social support has historically been asso- pants found the test blocks more stressful (mean [M;
ciated with less stress, we examined the associations be- SD] = 2.90 [0.73]) than the practice blocks (M [SD] = 1.98
tween self-reported receiving and giving social support [0.72], t(35) = 8.74, p < .001), indicating that the test
and a broad array of negative psychological outcomes. A blocks were indeed stressful. Stress-related neural activity
“vulnerability index” was created that included depression was evaluated by comparing the test blocks to the practice
(Beck Depression Inventory (36)), sensitivity to social re- blocks.
jection (Mehrabian Rejection Sensitivity (37)), perceptions
of stress (Perceived Stress Scale (38)), and feelings of Affiliative Task
loneliness (University of California, Los Angeles [UCLA] To examine associations between social support and
Loneliness Scale (39)). The scales were standardized to reward-related activity in the VS to viewing images of close
the same scale by converting the raw scores to z scores. others, participants completed a modified affiliative task
Scores were averaged to create a single value per partici- previously shown to activate reward-related neural circuitry
pant (α = .82). (41–43). Before their experimental session, participants
e-mailed two digital photographs of two different close
NEUROIMAGING MEASURES others and ratings to two items using a 1-10 scale: “how
The association between questionnaire measures of re- close are you to this person,” anchored by “not at all” and
ceiving and giving social support and neural activity was “extremely” and “to what extent is this person a source of
assessed across three separate tasks social support to you” anchored by “not a source of sup-
port” and “tremendous source of support.” Participants
Stress Task were indeed able to identify people with whom they had
Stress-related neural activity was manipulated with a mod- close, supportive relationships (M [SD] closeness = 9.21
ified version of the Montreal Imaging Stress Task (40) [1.31]; M [SD] supportiveness = 9.03 [1.26]). One outlier,
where participants calculated math problems of varying dif- who rated one of their close others a 4 on closeness and sup-
ficulty under the guise that the experimenters were compar- portiveness, was removed from any analyses evaluating the
ing their performance to other students at UCLA. As a affiliative task. Therefore, the affiliative task is based on a
control condition, participants completed “practice” blocks sample of 35 participants.
where they answered easy math problems (e.g., 1 + 2 + 4=) Images of close others were reformatted into standard
silently in their heads. Participants pushed a button once space and presented along with gender, race, and age-
they came up with an answer. No answer choices or re- matched strangers in a block design. For each block, two
sponses were displayed. During the “test” blocks, the blocks images of a single person (either a close other or a stranger)
designed to elicit stress, participants mentally calculated were presented for a total of 16 seconds (8 seconds for each
more difficult math problems (e.g., 14  3/6 =) under time of the two images). In between blocks of images, partici-
pressure and then selected an answer from among a set of pants completed easy mental arithmetic silently in their

Psychosomatic Medicine, V 78 • 443-453 445 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 ORIGINAL ARTICLE

heads (e.g., count back by 3's from 639) in 12-second (44,45). To assess caregiving-related neural activity in re-
blocks. This was included as a neutral baseline condition sponse to acting prosocially, costly giving trials were com-
intended to prevent participants from continuing to think pared with both control and self-reward trials.
about their close others in between trials. In total, 16 blocks
were presented, 8 mental arithmetic, 4 close other, and 4 Functional Magnetic Resonance Imaging
stranger. Neural activity to the affiliative task was evaluated Data Acquisition
by comparing blocks of close others to blocks of strangers. Participants were scanned at UCLA's Staglin IMHRO Cen-
ter for Cognitive Neuroscience on a Siemens 3-T “Tim Trio”
Prosocial Task MRI scanner. Two anatomical scans, a high-resolution T1-
In addition to exploring associations between social support weighted echo-planar imaging volume (spin-echo, repetition
and neural activity to a stressful and affiliative task, we time [TR] = 5000 milliseconds; echo time [TE] = 33 millisec-
wanted to understand the associations between social sup- onds; matrix size 128  128; 36 axial slices; field of view
port and neural responses to acting prosocially. Thus, par- [FOV] = 20 cm; 3 mm thick, skip 1 mm) and T2-weighted
ticipants completed a modified version of a donation task matched-bandwidth (slice thickness = 3 mm, gap = 1 mm,
(44,45). In this task, participants had the chance to earn raf- 36 slices, TR = 5000 milliseconds, TE = 34 milliseconds, flip
fle tickets for themselves and for “someone you know that angle = 90°, matrix = 128  128, FOV = 20 cm) were col-
could use some money right now.” Before the task began, lected. Functional scans for each of the three tasks (stress
participants were asked to name the person they would be task: 9 minutes, 31 seconds; affiliative task: 4 minutes,
playing for and to provide the email and phone number of 14 seconds; prosocial task: 10 minutes, 12 seconds) were
this person so that the experimenters could later contact then acquired (echo planar T2*-weighted gradient echo,
them directly if they won. After the study was over, the TR = 2000 milliseconds, TE = 25 milliseconds, flip
tickets were entered into a drawing (including subjects as angle = 90°, matrix size 64  64, 36 axial slices, FOV = 20 cm;
well as those they were playing for), in which two people 3 mm thick, skip 1 mm). The affiliative and prosocial tasks
won $300 each. The more tickets a participant won, the were counterbalanced across participants, but the stress task
greater chance that he/she or the person he/she was playing was always presented after these other two tasks to ensure
for could win the $300. that the stressfulness of completing the stress task did not
During the task, participants saw four different trial alter the potentially pleasant experience of completing the
types. On each trial, participants viewed a proposed distri- other two tasks.
bution of raffle tickets split across themselves and the other
person they chose to play for (3.5 seconds followed by a Statistical Analyses
jittered fixation period, lasting 3 seconds on average [range,
2.28–6.07]), and participants could then choose whether to Self-Report Correlations
accept or reject the proposed split. During the key prosocial Self-report measures of receiving and giving social support
trials—costly giving trials, participants chose whether to ac- were correlated with the vulnerability index, the measure of
cept offers where they could give raffle tickets to the other current negative psychological outcomes, using SPSS ver-
person at a cost of tickets for themselves (e.g., YOU −10, sion 20.0.
OTHER +50). During the self-reward trials, participants
could win raffle tickets for themselves without any cost to Functional Magnetic Resonance Imaging Data
the other person (e.g., YOU +40, OTHER +0). On control Data were preprocessed using the DARTEL procedure in
trials, participants saw trials in which no tickets could be SPM8 (Wellcome Department of Imaging Neuroscience,
won or lost (YOU +0, OTHER +0). Costly reward (e.g., London). Images were realigned, then normalized to the
YOU +30, OTHER −10) trials were also included in the T2-weighted matched bandwidth and warped into Montreal
task to keep participants engaged and interested in the task, Neurologic Institute space, and finally smoothed with an
but were not analyzed. 8-mm Gaussian kernel, full width at half maximum. The
The prosocial task included 90 trials (40 costly giving— general linear model was used to estimate first-level effects
to account for the fact that some trials would not be ac- to each condition of interest compared with their respective
cepted, 20 pure reward, 20 control, 10 costly reward). The controls. Group level analyses were then computed using
range of tickets that participants could win or lose varied the first-level contrast images for each participant.
between −30 tickets and +70 tickets. The running total of
tickets was not shown. During data analysis, trials were Region-of-Interest Analyses
binned based on the acceptance of an offer. Thus, if a partic- Given that the main goals of the current study were to ex-
ipant accepted a costly giving trial, that trial was modeled amine the association between self-reported social support
as a prosocial giving trial; otherwise, it was modeled sep- and stress, reward, and caregiving-related neural circuitry
arately and not examined (based on prior procedures) to three separate tasks, analyses were constrained to a priori

Psychosomatic Medicine, V 78 • 443-453 446 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 Social Support and Neural Outcomes

hypothesized regions of interest (ROIs; see Fig. 1). In addi- (49–51), although it is still a relatively small region, and
tion, two regions that were unrelated to the hypotheses thus, results should be interpreted with caution.
tested in the current study were created to test the specificity Finally, “control” regions were created based on their
of the associations. irrelevance to the current tasks to establish whether the as-
For the stress task, we examined regions known to signal sociations between social support and neural activity were
threat and stress including the dACC, left and right anterior specific to the hypothesized regions. The supplementary
insula (AI), and left and right amygdala. Stress-related ROIs motor area (SMA) was chosen as a control region for the
were structurally defined using the Automated Anatomical affiliative task because this task did not involve any motor
Labeling (AAL) atlas (46). To further refine the dACC ROI, actions (i.e., there were no button responses required) and
we constrained the region at 32 < y < 0 on the basis of sum- the olfactory cortex, including the olfactory tubercle and
mary data on cingulate activations to physical pain (47), and Broca's olfactory cortex located under the corpus callo-
for the AI ROI, we divided the insula at y = 8 the approxi- sum, was chosen as a control region for the stress and
mate boundary between the dysgranular and granular sectors. prosocial tasks because there is no hypothesized function
For the affiliative task, left and right VS ROIs were of this region in stress or prosocial processes. Both re-
structurally defined by combining the caudate and putamen gions were defined using the AAL atlas using the Wake
from the AAL atlas and then constraining the regions at Forest University Pickatlas (52).
−24 < x < 24, 4 < y < 18, −12 < z < 0. Based on previous All ROI analyses were run in Marsbar (http://marbar.
work on the neural correlates of support giving (32), the an- sourceforge.net) and thresholded at p < .05.
atomical ROIs of the VS and an additional ROI of the SA
were used to explore activity in maternal caregiving regions
to the prosocial task. The SA ROI was previously defined RESULTS
for a study examining prosocial emotions (48) according
to microscopic sections between y = 0 and y = 14 that show Correlation Between Receiving/Giving Social
the location of the septal nuclei through the anterior com- Support and the Vulnerability Index
missure and anterior to the optic chiasma. Although the sep- In line with the current literature linking social support with
tal nuclei are too small to be imaged with the 3-T functional well-being, both self-reported receiving support (r = −0.60,
magnetic resonance imaging, the SA ROI used in the cur- p < .001) and self-reported giving support (r = −0.64,
rent study encompasses the larger surrounding area. The p < .001) were negatively correlated with the vulnerability
SA has also been imaged and reported on previously index (Fig. 2). That is, both receiving and giving more

FIGURE 1. ROIs for each of the three tasks. Analyses were targeted to stress-related activity in the AI, dACC, and amygdala for the stress
task; reward-related activity in the VS for the close other task; and caregiving-related activity in the VS and SA for the prosocial task.
ROIs = regions of interest; VS = ventral striatum, AI = anterior insula; dACC = dorsal anterior cingulate cortex; SA = septal area.
Color image is available only in online version (www.psychosomaticmedicine.org).

Psychosomatic Medicine, V 78 • 443-453 447 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 ORIGINAL ARTICLE

FIGURE 2. Associations between self-reported social support and the vulnerability index. Self-reported negative psychological outcomes
were negatively correlated with receiving support (A) and giving support (B) such that less vulnerability was associated with greater
support receiving and giving.

support were related to lower reported negative psycholog- support giving were significantly different from the correla-
ical outcomes. tions with self-reported support receiving across all of these
regions (dACC z = 2.01, p = .022; left AI z = 2.13, p = .016;
Neural Results right AI z = 2.02, p = .014; left amygdala z = 1.71, p = .043;
right amygdala z = 1.81, p = .035). Thus, those who re-
Stress Task ported the most support giving also displayed the least
Main Effect of Task amount of threat-related activity to the socially evaluative
For the stressful math task, we looked for activity in stress stressor, which is consistent with hypotheses that support
and threat-related regions (dACC, AI, amygdala) when par- giving might be stress reducing.
ticipants completed test blocks compared with when partici- Supporting the specificity of this effect, there were no
pants completed practice blocks. Validating the stressfulness associations between olfactory cortex activity (the chosen
of the task, dACC (M [SD] = 0.21 [0.61]) and left (M [SD] = control region) and either self-reported receiving support
0.14 [0.51]) and right AI (M [SD] = 0.08 [0.55]) activity (r = 0.164, p = .170) or self-reported support giving
was greater during the test blocks compared with the prac- (r = 0.007, p = .491) in response to the stress task (test ver-
tice blocks (M [SD] dACC = 0.06 [0.48], t(35) = 2.07, p = sus practice blocks).
.023; M [SD] left AI = 0.03 [0.42], t(35) = 1.63, p = .056; M
[SD] right AI = −0.07 [0.43], t(35) = 2.29, p = .014). Affiliative Task
Amygdala activity, however, did not differ between test Main Effect of Task
(M [SD] left amygdala = −0.25 [0.59]; M [SD] right VS activity was assessed as participants viewed images of
amygdala = −0.30 [0.63]) and practice blocks (M [SD] close, supportive others and strangers. Consistent with the im-
left amygdala = −0.30 [0.63], t(35) = .74, p = .23; M right portance of reward-related processing to close social relation-
amygdala [SD] = −0.32 [0.48], t(35) = .16, p = .44). ships (42), viewing images of close others (versus strangers)
led to increased activity in the left (M [SD] close other = 0.04
Correlations Between Receiving/Giving Support
and Neural Activity [0.22], M [SD] stranger = −0.03 [0.20], t(34) = 1.88, p = .034)
We then examined whether questionnaire measures of so- and right VS (M [SD] close other = −0.03 [0.20], M [SD]
cial support were related to neural activity to the stress task. stranger = −0.10 [0.21], t(34) = 1.71, p = .049).
Although receiving support was not associated with dACC Correlations Between Receiving/Giving Support and
(r = 0.07), AI (left r = 0.08, right r = 0.03), or amygdala ac- Neural Activity
tivity (left r = −0.04, right r = −0.02), support giving was Correlational analyses revealed that although support re-
negatively correlated with dACC (r = −0.27, p = .054), left ceiving was not associated with VS activity (left r = 0.03,
(r = −0.28, p = .048) and right AI (r = −0.33, p = .024), and right r = −0.02), support giving was associated with greater
left (r = −0.32, p = .028) and right amygdala (r = −0.32, left (r = 0.42, p = .006) and right VS (r = 0.41, p = .008) to
p = .028) activity to test (versus practice) blocks (Fig. 3). viewing images of close others (versus strangers). More-
The correlations between neural activity and self-reported over, these associations were significantly different from

Psychosomatic Medicine, V 78 • 443-453 448 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 Social Support and Neural Outcomes

FIGURE 3. Associations between stress and threat-related activity and self-reported social support. A, Parameter estimates from the AI,
dACC, and amygdala while participants completed stressful blocks of mental arithmetic (versus practice blocks) were not associated with
how much support they reported receiving. B, Stress-related activity was, however, negatively correlated with support giving such that less
AI, dACC, and amygdala to stressful blocks (versus practice blocks) were associated with greater support giving. AI = anterior insula;
dACC = dorsal anterior cingulate cortex.

those between receiving support and VS activity (left VS right VS and SA activity to costly giving and reward trials
z = 2.34, p = .009; right VS z = 2.58, p = .005; Fig. 4). (versus control) separately. Consistent with prior studies
To examine the specificity of these effects, we examined (44,45), VS activity was greater during the costly giving tri-
the correlation between activity in the control region (SMA) als, when participants chose to give raffle tickets to someone
and questionnaire measures of social support. As expected, else (M [SD] left VS = .26 [0.50]; M [SD] right VS = .42
there were no associations between SMA activity and either [0.62]), compared with the control condition (M [SD]
self-reported receiving support (r = −0.024, p = .446) or left VS = .04 [0.20], t(35) = 2.795, p = .004; M [SD] right
self-reported support giving (r = 0.191, p = .136) in re- VS = .07 [0.24], t(35) = 3.937, p < .001). Moreover, in re-
sponse to the affiliative task. sponse to self-reward (versus control) trials, there was greater
activity in the right VS (M [SD] = 0.31 [0.75], t(35) = 2.193,
Prosocial Task p = .018), but not the left VS (M [SD] = 0.16 [0.72],
Main Effect of Task t(35) = 1.15, p = .13). SA activity was not significantly dif-
To ensure that the prosocial task activated the hypothesized ferent between costly giving trials (M [SD] = −0.18 [1.55])
caregiving regions, we evaluated differences in the left and and control trials (M [SD] = −0.18 [0.55], t(35) = −0.01,

FIGURE 4. Associations between VS activity to the affiliative task and social support. A, VS to viewing images of close others (versus
strangers) was not associated with receiving support. B, However, parameter estimates from the VS ROI to viewing images of close others
(versus strangers) were associated with more support giving. VS = ventral striatum; ROI = region of interest.

Psychosomatic Medicine, V 78 • 443-453 449 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 ORIGINAL ARTICLE

p = .50) or between self-reward trials (M [SD] = −0.50 [1.78]) measure of receiving support (r = 0.004, p = .491) or the
compared with control trials (t(35) = −1.34, p = .094). There questionnaire measure of support giving (r = 0.021,
were no differences in VS and SA to costly giving versus p = .452).
self-reward (p values > .14), indicating that both giving and
receiving lottery tickets elicited reward-related activity.
DISCUSSION
Correlations Between Receiving/Giving Support and To date, our understanding of the association between so-
Neural Activity cial ties and health has been incomplete insofar as there is
Correlational analyses showed no association between self- a fair amount known about the benefits of receiving sup-
reported receiving support and either left (r = 0.11, p = .26) port, but less is known about the benefits that may come
or right VS activity (r = 0.09, p = .31) in response to costly from giving to others. The current study examined the asso-
giving (versus self-reward) trials. The association between ciations between questionnaire measures of receiving and
receiving support and SA activity during costly giving (ver- giving support to others, self-reported negative psycholog-
sus self-reward) was marginal (r = 24, p = .084). However, ical outcomes, and neural activity to stressful and socially
there was a significant association between support giving rewarding tasks. Although both receiving and giving social
and left VS (r = 0.31, p = .033), right VS (r = 0.31, support were related to lower reported negative psycholog-
p = .034), and SA (r = 0.39, p = .095) activity (Fig. 5). That ical outcomes, at the level of the brain, only support giving
is, those who reported giving the most support to others was associated with beneficial outcomes. Specifically, sup-
demonstrated the greatest VS and SA activity to costly giving port giving was associated with lower threat-related activity
trials relative to self-reward trials. Thus, the more participants to a stress task, greater activity in the VS to viewing images
reported giving support to others, the more caregiving-related of close others (versus strangers), and greater caregiving-
neural activity they showed to trials in which they chose to related neural activity to acting prosocially (versus self-
give to another person in need, but no such association was ishly). These results add to the accumulating literature
found between how much support people received and neu- suggesting that giving to others might be beneficial for
ral activity to the prosocial task. The correlations between health and well-being (24,32,44,53).
self-reported support giving and the VS were marginally The current findings suggest that giving to others may act
different from the correlation between self-reported receiv- through multiple routes: a) via reduced stress-related mecha-
ing support and the VS (left: z = −1.32, p = .094; right nisms, b) via increased reward-related activity in response to
z = −1.19, p = .12); however, the correlation between self- viewing close others, and c) via increased caregiving-related
reported support giving and SA activity was not signifi- activity in response to acting prosocially. The findings from
cantly different from the correlation between self-reported each of the three tasks are discussed below.
receiving support and SA activity (p = .17). Although physical health outcomes were not measured
Finally, we examined the correlation between activity in this study, the current results shed light on possible neural
in the control region (olfactory cortex) and questionnaire pathways by which giving to others may be associated with
measures of social support. There were no associations be- better health outcomes (20,22), namely, by reducing ac-
tween olfactory cortex activity and either the questionnaire tivity in stress and threat-related regions during stressful

FIGURE 5. Correlations between parameter estimates from caregiving-related neural regions during the prosocial task and social support.
A, Whereas activity in the VS and SA when participants acted prosocially (versus selfishly) was not significantly associated with receiving
support, (B) greater VS and SA activity during prosocial (versus selfish) decisions was associated with more support giving. VS = ventral
striatum; SA = septal area.

Psychosomatic Medicine, V 78 • 443-453 450 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 Social Support and Neural Outcomes

experiences. Indeed, stressor-evoked dACC, AI, and amyg- association between VS activity and reports of receiving
dala activity, as studied here, have all been associated with support. The VS is particularly sensitive to social (versus
increased autonomic (sympathetic nervous system) responding nonsocial) rewards (62), and also to the degree of closeness
(see Ref. (31) for review). between two individuals (63). For instance, sharing mone-
Why might giving support, but not receiving support, be tary rewards with a friend activates the VS more than shar-
associated with reduced stress-related neural activity? Giv- ing rewards with a stranger. Furthermore, those reporting
ing support might bypass some of the conditions that result higher closeness ratings with the friend display greater VS
in the detrimental effects associated with receiving support. to the shared wins with the friend (versus a stranger). Sim-
For instance, receiving support sometimes backfires be- ilarly, the number of years married is associated with in-
cause it mismatches one's personal preferences for support creased VS to images of romantic partners (versus close
(54), leaves one feeling indebted (18), and signals that friends; (41)). Based on these findings, it is possible that
something distressing or negative has happened (55). Giv- VS activity to “social rewards” might also signal the degree
ing support, on the other hand, allows an individual to con- of closeness and connection with others. Thus, the positive
trol when and how support is given, can lead to a greater correlation between support giving and VS to images of
sense of autonomy and self-efficacy (56), and is associated close others may suggest that there is something unique
with increased feelings of social connection (32) and in- about giving support, but not necessarily receiving support,
creased happiness (57). Thus, to the extent that giving that binds people closer together (26).
to others simultaneously bypasses the problems that re- The prosocial task assessed caregiving-related neural ac-
ceiving support sometimes elicits and increases one's psy- tivity as participants had the opportunity to give to another
chological resources, giving, relative to receiving support, individual. Self-reported support giving, but not support re-
may result in more effective stress reduction. The cur- ceiving, was positively correlated with VS and SA activity
rent results are correlational, however, and so experimental in response to acting prosocially (versus selfishly). These
work that directly compares both receiving and giving findings suggest that those who tend to give more support
support on physical health-related outcomes is needed might also be more rewarded by acting prosocially toward
to understand when and how each kind of support is a known other, which may encourage more subsequent sup-
uniquely beneficial or whether those who give support are port giving behavior. Although the current task assessed
simply less prone to stress-related responding to socially prosocial behavior toward a single target, caregiving-
evaluative stressors. related activity while acting prosocially toward others more
The hypothesis that support giving might be stress re- broadly may predict better health outcomes on its own.
ducing on its own seems to contradict the large and well- Consistent with this possibility, one recent study found that
known literature on the negative health effects of chronic VS activity to giving, but not to receiving monetary rewards
caregiving (e.g., (58,59)). It is possible that when the care- for oneself, was associated with reduced depression levels a
giving needs exceed the demands of psychological, mate- year later (45). Future work directly linking caregiving cir-
rial, or social resources available to an individual, giving cuitry during prosocial behavior and health outcomes out-
support may no longer be health protective. However, stud- side the scanner will clarify the role of the VS and SA in
ies linking chronic caregiving to poor health outcomes have any health benefits that may come from giving to others.
thus far failed to control for the emotional distress of wit- An important future direction for understanding how
nessing the deterioration of a loved one. Therefore, it re- support giving relates to beneficial outcomes is to isolate
mains unclear whether the support giving component of the unique contribution of support giving itself from other
chronic caregiving is the “active ingredient” in the associa- individual difference factors, such as extraversion or gen-
tion between caregiving and bad health or if there are other eral positive affect. Previous work exploring the health ben-
factors at play. Some studies suggest that caregiving can be efits of support giving has shown that the associations
associated with beneficial, rather than detrimental, health between support giving and health remain after controlling
outcomes (60,61), pointing to the possibility that the care- for subjective well-being, baseline health factors, and extra-
giving itself may not be the cause of the negative health ef- version (20,23), suggesting that there is something uniquely
fects. Thus, much like receiving support, there are a number beneficial about giving to others. However, controlling for
of factors that contribute to whether and when giving to these additional personality factors will help clarify the
others is or is not beneficial for health. unique role that support giving plays in health.
In addition to the relationship between support giving In sum, the results of the current study suggest that ex-
and reduced stress-related neural activity, support giving amining the psychological and physical health benefits of
was also associated with increased VS activity to viewing giving support to others deserves greater empirical attention
images of close others who reported giving more support and that gaining a full understanding of how and why social
to others also displayed greater reward-related activity ties are so important to well-being requires the consider-
to images of their own loved ones. There was no such ation of both the support that is received and given.

Psychosomatic Medicine, V 78 • 443-453 451 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 ORIGINAL ARTICLE

The authors acknowledge the Staglin IMHRO Center 17. Younger J, Aron A, Parke S, Chatterjee N, Mackey S. View-
for Cognitive Neuroscience. ing pictures of a romantic partner reduces experimental pain:
involvement of neural reward systems. PLoS One 2010;5:
The content of this publication does not necessarily re- e133309.
flect the views of policies of the Department of Health and 18. Gleason MEJ, Iida M, Bolger N, Shrout PE. Daily supportive
Human Services, nor does mention of trade names, com- equity in close relationships. Pers Soc Psychol Bull 2003;29:
1036–45.
mercial products, or organizations imply endorsement by 19. Bolger N, Amarel D. Effects of social support visibility on ad-
the US government. justment to stress: experimental evidence. J Pers Soc Psychol
Source of Funding and Conflicts of Interest: This project 2007;92:458–75.
20. Brown SL, Nesse RM, Vinokur AD, Smith DM. Providing so-
has been funded in whole or in part with Federal funds cial support may be more beneficial than receiving it: results from
from the National Cancer Institute, National Institutes of a prospective study of mortality. Psychol Sci 2003;14:320–7.
Health, under Contract No. HHSN261200800001E. The 21. Inagaki TK, Eisenberger NI. Giving support to others re-
authors declared no conflicts of interest. duces sympathetic nervous system-related responses to stress.
Psychophysiology 2015; In Press.
REFERENCES 22. Piferi RL, Lawler KA. Social support and ambulatory blood
pressure: an examination of both receiving and giving. Int J
1. Bowlby JA. Secure Base: Parent-Child Attachment And Healthy Psychophysiol 2006;62:328–36.
Human Development. New York, NY: Basic Books; 1988. 23. Poulin MJ, Brown SL, Dillard AJ, Smith DM. Giving to
2. House JS, Landis KR, Umberson D. Social relationships and others and the association between stress and mortality. Am
health. Science 1988;241:540–5. J Public Health 2013;103:1649–55.
3. Holt-Lunstad J, Smith TB, Layton JB. Social relationships 24. Vaananen A, Buunk BP, Kivimaki M, Pentti J, Vahtera J.
and mortality risk: a meta-analytic review. PLoS Med 2010; When it is better to give than to receive: long-term health ef-
7:e1000316. fects of perceived reciprocity in support exchange. J Pers Soc
4. Gunnar MR, Bruce J, Grotevant HD. International adoption Psychol 2005;89:176–93.
of institutionally reared children: research and policy. Dev 25. Brown SL, Brown RM. Connecting prosocial behavior to im-
Psychopathol 2000;12:677–93. proved physical health: Contributions from the neurobiology
5. Tottenham N, Sheridan MA. A review of adversity, the amyg- of parenting. Neurosci Biobehav Rev 2015;55:1–17.
dala and the hippocampus: a consideration of developmental 26. Brown SL, Brown RM. Selective investment theory: recasting
timing. Front Hum Neurosci 2010;3:68. the functional significance of close relationships. Psychol Inq
6. Lutgendorf SK, Sood AK. Biobehavioral factors and cancer pro- 2006;17:1–29.
gression: physiological pathways and mechanisms. Psychosom 27. Brown S, Brown M, Preston S. The human caregiving system.
Med 2011;73:724–30. In: Brown SL, Brown M, Penner LA, editors. Moving Be-
7. Wills TA. Social support and interpersonal relationships. In: yond Self Interest: Perspectives from Evolutionary Biology,
Clark MS, ed. Prosocial Behavior, Newbury Park: Sage; 1991; Neuroscience, and the Social Sciences. Oxford: University
265–89. Press; 2011:76–88.
8. Berkman LF, Syme SL. Social networks, host resistance, and 28. Preston S. The origins of altruism in offspring care. Psychol
mortality: a nine-year follow-up study of alameda county res- Bull 2013;139:1305–41.
idents. Am J Epidemiol 1979;109:186–204. 29. Taylor SE, Klein LC, Lewis BP, Gruenewald TL, Gurung
9. Cassel J. The contribution of the social environment to host RAR, Updegraff JA. Biobehavioral responses to stress in
resistance: the Fourth Wade Hampton Frost Lecture. Am J females: tend-and-befriend, not fight-or-flight. Psychol Rev
Epidemiol 1976;104:107–23. 2000;107:411–29.
10. Cobb S. Presidential Address-1976. Social support as a mod- 30. Feeney BC, Collins NL. Predictors of caregiving in adult in-
erator of life stress. Psychosom Med 1976;38:300–14. timate relationships: an attachment theoretical perspective. J
11. Cohen S, Wills TA. Stress, social support, and the buffering Pers Soc Psychol 2001;80:972–94.
hypothesis. Psychol Bull 1985;98:310–57. 31. Eisenberger NI, Cole SW. Social neuroscience and health:
12. Kamarck TW, Manuck SB, Jennings JR. Social support re- neurophysiological mechanisms linking social ties with phys-
duces cardiovascular reactivity to psychological challenge: a ical he. Nat Neurosci 2012;15:669–74.
laboratory model. Psychosom Med 1990;52:42–58. 32. Inagaki TK, Eisenberger NI. Neural correlates of giving sup-
13. Thorsteinsson EB, James JE. A meta-analysis of the effects of port to a loved one. Psychosom Med 2012;74:3–7.
experimental manipulations of social support during labora- 33. Adolphs R, Tranel D, Damasio H, Damasio AR. Fear and the
tory stress. Psychol Health 1999;14:869–86. human amygdala. J Neurosci 1995;15:5879–91.
14. Eisenberger NI, Taylor SE, Gable SL, Hilmert CJ, Lieberman 34. Anderson AK, Phelps EA. Lesions of the human amygdala
MD. Neural pathways link social support to attenuated neuro- impair enhanced perception of emotionally salient events. Na-
endocrine stress responses. Neuroimage 2007;35:1601–2. ture 2001;411:305–9.
15. Masten CL, Telzer EH, Fuligni A, Lieberman MD, 35. Shakespeare-Finch J, Obst PL. The development of the 2-way
Eisenberger NI. Time spent with friends in adolescence re- social support scale: a measure of giving and receiving emo-
lates to less neural sensitivity to later peer rejection. Soc Cogn tional and instrumental support. J Pers Assess 2011;93:483–90.
Affect Neurosci 2012;7:106–14. 36. Beck AT, Steer RA, Carbin MG. Psychometric properties of
16. Eisenberger NI, Master SL, Inagaki TK, Taylor SE, Shirinyan the Beck Depression Inventory: twenty-five years of evalua-
D, Lieberman MD, Naliboff BD. Attachment figures activate tion. Clin Psychol Rev 1988;8:77–100.
a safety signal-related neural region and reduce pain experi- 37. Mehrabian A. Questionnaire measures of affiliative tendency
ence. Proc Natl Acad Sci U S A 2011;108:11721–6. and sensitivity to rejection. Psychol Rep 1976;38:199–209.

Psychosomatic Medicine, V 78 • 443-453 452 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
 Social Support and Neural Outcomes

38. Cohen S, Kamarck T, Mermelstein R. A global measure of 50. Krueger F, McCabe K, Moll J, Kriegeskorte N, Zahn R,
perceived stress. J Health Soc Behav 1983;24:385–96. Strenziok M, Heinecke A, Grafman J. Neural correlates of
39. Russell DW. UCLA Loneliness Scale (version 3): reliability, trust. Proc Natl Acad Sci U S A 2007;104:20084–9.
validity, and factor structure. J Pers Assess 1996;66:20–40. 51. Morelli SA, Rameson LT, Lieberman MD. The neural com-
40. Dedovic K, Renwick R, Mahani NK, Engert V, Lupien SJ, ponents of empathy: predicting daily prosocial behavior. Soc
Pruessner JC. The Montreal imaging stress task: using func- Cogn Affect Neurosci 2014;9:39–47.
tional imaging to investigate the effects of perceiving and pro- 52. Maldjian JA, Laurienti PJ, Kraft RA, Burdette JH. An auto-
cessing psychosocial stress in the human brain. J Psychiatry mated method for neuroanatomic and cytoarchitectonic atlas-
Neurosci 2005;30:319–25. based interrogation of fMRI data sets. Neuroimage 2003;19:
41. Acevedo BP, Aron A, Fisher HE, Brown LL. Neural corre- 1233–9.
lates of long-term intense romantic love. Soc Cogn Affect 53. Dunn EW, Aknin LB, Norton MI. Spending money on others
Neurosci 2012;7:145–59. promotes happiness. Science 2008;319:1687–8.
42. Aron A, Fisher H, Mashek DJ, Strong G, Li H, Brown LL. 54. Cutrona CE. Stress and social support: in search of optimal
Reward, motivation, and emotion systems associated with matching. J Soc Clin Psychol 1990;9:3–14.
early-stage intense romantic love. J Neurophysiol 2005;93: 55. Seidman G, Shrout PE, Bolger N. Why is enacted social sup-
327–37. port associated with increased distress? Using simulation to
43. Inagaki TK, Muscatell KA, Irwin MR, Moieni M, Dutcher test two possible sources of spuriousness. Pers Soc Psychol
JM, Jevtic I, Breen EC, Eisenberger NI. The role of the ventral Bull 2006;32:52–65.
striatum in inflammatory-induced approach toward support 56. Gruenewald TL, Liao DH, Seeman TE. Contributing to
figures. Brain Behav Immun 2015;44:247–52. others, contributing to oneself: perceptions of generativity
44. Moll J, Krueger F, Zahn R, Pardini M, de Oliveira-Souza R, and health in later life. J Gerontol B Psychol Sci Soc Sci
Grafman J. Human fronto-mesolimbic networks guide deci- 2012;67:660–5.
sions about charitable donation. Proc Natl Acad Sci U S A 57. Harbaugh WT, Mayr U, Burghart DR. Neural responses to
2006;103:15623–8. taxation and voluntary giving reveal motives for charitable
45. Telzer EH, Fuligni AJ, Lieberman MD, Gálvan A. Neural donations. Science 2007;316:1622–5.
sensitivity to eudaimonic and hedonic rewards differentially 58. Schulz R, Beach SR. Caregiving as a risk factor for mortality:
predict adolescent depressive symptoms over time. Proc Natl the caregiver health effects study. JAMA 1999;22:15–9.
Acad Sci U S A 2014;111:6600–5. 59. Vitaliano PP, Zhang J, Scanlan JM. Is caregiving hazardous to
46. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello one's physical health? A meta-analysis. Psychol Bull 2003;
F, Etard O, Delcroix N, Mazoyer B, Joliot M. Automated an- 129:946–72.
atomical labeling of activations in SPM using a macroscopic 60. Brown SL, Smith DM, Schulz R, Kabeto MU, Ubel PA,
anatomical parcellation of the MNI MRI single-subject brain. Poulin M, Yi J, Kim C, Langa KM. Caregiving behavior is
Neuroimage 2002;15:273–89. associated with decreased mortality risk. Psychol Sci 2009;
47. Vogt BA, Berger GR, Derbyshire SW. Structural and functional 20:488–94.
dichotomy of human midcingulate cortex. Eur J Neurosci 61. Amirkhanyan AA, Wolf DA. Caregiver stress and noncare-
2003;18:3134–44. giver stress: exploring the pathways of psychiatric morbidity.
48. Zahn R, Moll J, Paiva M, Garrido G, Krueger F, Huey ED, Gerontologist 2003;48:817–27.
Grafman J. The neural basis of human social values: evidence 62. Izuma K, Saito DN, Sadato N. Processing of social and mone-
from functional MRI. Cereb Cortex 2009;19:276–83. tary rewards in the human striatum. Neuron 2008;58:284–94.
49. Ho SS, Konrath S, Brown S, Swain JE. Empathy and stress 63. Fareri DS, Niznikiewicz MA, Lee VK, Delgado MR. Social
related neural responses in maternal decision making. Front network modulation of reward-related signals. J Neurosci
Neurosci 2014;8:152. 2012;32:9045–52.

Psychosomatic Medicine, V 78 • 443-453 453 May 2016

Copyright © 2016 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.