The effects of gratitude expression on neural activity

Full text

The effects of gratitude expression on neural activity
Prathik Kini, Joel Wong, Sydney McInnis, Nicole Gabana, Joshua W. Brown ⁎
Indiana University, Bloomington, United States
abstractarticle info
Article history:
Received 23 July 2015
Accepted 23 December 2015
Available online 30 December 2015
Gratitude is a common aspect of social interaction, yet relatively little is known about the neural bases of grati-
tude expression, nor how gratitude expression may lead to longer-term effects on brain activity. To address
these twin issues, we recruited subjects who coincidentally were entering psychotherapy for depression and/
or anxiety. One group participated in a gratitude writing intervention, which required them to write letters ex-
pressing gratitude. The therapy-as-usual control group did not perform a writing intervention. After three
months, subjects performed a “Pay It Forward”task in the fMRI scanner. In the task, subjects were repeatedly
endowed with a monetary gift and then asked to pass it on to a charitable cause to the extent they felt grateful
for the gift. Operationalizing gratitude as monetary gifts allowed us to engage the subjects and quantify the grat-
itude expression for subsequent analyses. We measured brain activity and found regionswhere activity correlat-
ed with self-reported gratitude experience during the task, even including related constructs such as guilt
motivation and desire to help as statistical controls. These were mostly distinct from brain regions activated by
empathy or theory of mind. Also, ourbetween groups cross-sectional study found that a simplegratitude writing
intervention was associated with significantly greater and lasting neural sensitivity to gratitude –subjects who
participated in gratitude letter writing showed both behavioral increases in gratitude and significantly greater
neural modulation by gratitude in the medial prefrontal cortex three months later.
© 2015 Elsevier Inc. All rights reserved.
Keywords:
Gratitude
fmri
Plasticity
Counseling
Empathy
Introduction
Gratitude is an essential part of human prosocial behavior. A number
of recent studies have shown the benefits of gratitude interventions on
well being, mainly using gratitude letter writing or keeping a gratitude
diary, both of which can be effective (Kaczmarek et al., 2015). Gratitude
interventions have recently been shown as effective at increasing well
being in students (Flinchbaugh et al., 2012), those with chronic pain
(Baxter et al., 2012), depression (Cheng et al., 2015), and older adults
(Killen and Macaskill, 2015). Gratitude interventions have similar effec-
tiveness compared with mindfulness interventions (O’Leary and
Dockray, 2015) and practicing kindness (Kerr et al., 2015).
Despite the recent findings regarding gratitude intervention effec-
tiveness, the basic neural mechanisms involved in gratitude are rela-
tively unknown, as are the neurological mechanisms mediating the
effects of gratitude interventions (Layous et al., 2011). On the one
hand, gratitude may involve affective processes. Gratitude has been
characterized as a positive moral affect alongside other moral affects
such as empathy, sympathy, guilt, and shame, and as a force that helps
people maintain positive social reciprocity (McCullough et al., 2001).
On the other hand, other work has cast gratitude as related to a more
cognitive process of benefitappraisal(Wood et al., 2008).
Gratitude may also be delineated in terms of experience vs. expres-
sion. The recipients (or observers) of a generous or prosocial act may
experience emotions related to gratitude, such as positive affect,
empathy, and increased inclination toward prosocial behavior
(Emmons and Stern, 2013). The experience of gratitude may natural-
ly lead to an expression of gratitude, which typically takes the form
of a verbal recognition (“thank you”)orareciprocalgift(suchas
generous restaurant tipping). In this paper we address both experience
and expression of gratitude in three inter-related ways. First, subjects in
an experimental group express gratitude in the form of a written letter
three months prior to fMRI scanning. Second, during scanning, we
model gratitude as an expression, with monetary gifting as a quantifi-
able operationalization of gratitude expression. Third, during scanning,
we also ask subjects to evaluate their experience of gratitude on each
trial, with self-reports on a Likert scale.
Gratitude has been studied with neuroimaging as part of more gen-
eral investigationsof human social value (Zahnet al., 2009), but the spe-
cific neural correlates of gratitude have only recently been explored
(Fox et al., 2015). A number of similar social cognitive and affective con-
structs have been studied, including trust and reciprocity (King-Casas
et al., 2005), fairness (De Quervain et al., 2004; Sanfey et al., 2003),
and empathy (Singer et al., 2004, 2006). Collectively these studies high-
light a number of brain regions as central to social interaction, including
prominently the anterior cingulate cortex (ACC), anterior insula (AI),
ventromedial prefrontal cortex (vmPFC), and striatum, which are gen-
erally limbic regions associated with affect and valuation. The anterior
NeuroImage 128 (2016) 1–10
⁎Corresponding author at: Dept. of Psychological and Brain Sciences, 1101 E Tenth St.,
Bloomington, IN 47405-7007, United States.
E-mail address: jwmbrown@indiana.edu (J.W. Brown).
http://dx.doi.org/10.1016/j.neuroimage.2015.12.040
1053-8119/© 2015 Elsevier Inc. All rights reserved.
Contents lists available at ScienceDirect
NeuroImage
journal homepage: www.elsevier.com/locate/ynimg

cingulate cortex in particular may play a key role in predicting and eval-
uating the outcomes of actions (Alexander and Brown, 2010, 2011), in-
cluding in social contexts (Chiu et al., 2008; King-Casas et al., 2005;
Tomlin et al., 2006). ACC activity reflects and drives avoidance behavior,
especially of potential losses (Brown and Braver, 2007; Fukunaga et al.,
2012) as well as potential regret (Coricelli et al., 2005) and social affect
(Harris et al., 2007), and has been shown to correlate with empathy
(Rameson et al., 2012), especially for pain (Singer et al., 2006). If grati-
tude involves primarily an affective process, then gratitude expression
should correlate with activity in the limbic regions.
Gratitude may also involve cognitive processes, including computa-
tions and appraisals of appropriate reciprocity (Wood et al., 2008).
Mental calculations generally involve more dorsolateral prefrontal and
parietal brain regions (Dehaene et al., 2004). Thus, if gratitude involves
more cognitive than affective processes, then gratitude expression
should correlate with activity in the parietal and dorsolateral prefrontal
regions.
In addition to questions about the basic neural mechanisms of grat-
itude, there is potential clinical relevance. Growing evidence supports
the mental health benefits of gratitude interventions (e.g.Boehm et al.,
2011; Emmons and McCullough, 2003; Froh et al., 2008; Lyubomirsky
et al., 2011; Seligman et al., 2005; Sheldon and Lyubomirsky, 2006;
Watkins et al., 2003). Research on such interventions has focused main-
ly on non-clinical populations (e.g., college and high school students).
Still, the basic neural mechanisms through which gratitude interven-
tions might positively influence mental health are not well understood.
Several mechanisms have been proposed (Emmons and Mishra, 2012;
Geraghty et al., 2010) but only one peer-reviewed study so far has
identified a mediator (i.e., feelings of gratitude) of the positive effects
of a gratitude intervention (Emmons and McCullough, 2003).
Here our aim is to elucidate specifically the basic neural correlates of
gratitude expression, and we do not investigate or report clinical effects.
Our study focuses on four related questions: first, what are the immedi-
ate neural activity correlates associated with acts of expressing
gratitude (operationalized as monetary gifts)? Second, is the neural ac-
tivity more consistent with gratitude as a cognitive process or as an
emotional process? Third, to what extent does gratitude expression in-
volve the brain regions associated with empathy and theory of mind
processes? Fourth, what are the long-term effects of written gratitude
expression (as distinct from grateful monetary gifting) on neural sensi-
tivity to gratitude?
To addressthese questions, we developed a variant of the trust game
(Berg et al., 1995), called the “Pay It Forward”task, in which subjects ex-
press gratitude as monetary gifts while undergoing fMRI. We adminis-
tered a short gratitude writing intervention to half of the subjects and
explored the effect of that intervention on brain activity during the grat-
itude task several months later. This approach allows us to measure
both the immediate neural activity associated with gratitude and the
lasting effects of gratitude expression on brain activity. Here we show
two main results: first, that gratitude correlates with activity in specific
set of brain regions; and second, that a simple gratitude writing inter-
vention is associated with significant increases in both gratefulness
and neural sensitivity to gratitude over the course of weeks to months.
Methods
Subject recruitment
Subjects were recruited from a population of psychotherapy cli-
ents seeking clinical counseling. All subjects provided written in-
formed consent, and all procedures were approved by the Indiana
University IRB. Subjects were randomized to one of three conditions:
a gratitude writing intervention, an expressive writing intervention,
or a psychotherapy-only condition. The randomization was per-
formed with successive subjects to keep the running group sizes as
equal as possible, without regard for particular demographic factors.
Subjects initially completed the six-item Gratitude Questionnaire
(GQ) (Mccullough et al., 2002), the three-item gratitude adjectives
scale (GAC3) (Mccullough et al., 2002), which assess self-reports of
how grateful one feels in daily life, and the BHM-20 scale, which
briefly assesses mental health, including anxiety and depression
(Kopta and Lowry, 2002). Higher BHM-20 scores reflect better men-
tal health. For the gratitude writing intervention, subjects were
asked to spend 20 min writing a letter to someone expressing grati-
tude. They did this during three consecutive sessions on the first, sec-
ond, and third week of counseling. They were instructed that they
could choose whether or not to actually send the letters to the recip-
ient. Subjects in the expressive writing condition were asked to write
about their most stressful episodes in life. Three months after
counseling, the gratitude writing and psychotherapy-only subjects
were recruited to participate in the fMRI task.
A total of 43 clients (22 in the gratitude intervention condition and
21 in the psychotherapy only condition, age range 18 to 34, mean
22.98, 32 females, all right handed) were recruited for the fMRI study.
We did not scan the expressive writing subjects for cost reasons, al-
though future studies might consider doing so as an additional control.
The subject demographics are summarized in Table 1. There was no dif-
ference between the fMRI gratitude writing vs. psychotherapy-only
groups with respect to their age (p=0.50), gender (p = 0.73) initial
GQ (trait gratitude) scores (p=0.41), or symptoms of anxiety/depres-
sion as measured on the BHM-20 symptom subscale (p=0.55). The av-
erage BHM-20 score at intake was 2.39 for the gratitude group and 2.50
for the therapy-as-usual group (Table 1), which is consistent with aver-
age scores of 2.33 for psychotherapy outpatients and 2.68 for college
counseling clients, as reported by the BHM-20 developer (Kopta and
Lowry, 2002), who also reported for comparison that the average
healthy college student BHM-20 score was 3.13 (SD=0.51). Lower
BHM-20 scores indicate more symptoms of anxiety and/or depression.
fMRI task
In order to perform the study in a controlled setting, the expression
of gratitude was operationalized as money, with a function similar to
tipping a restaurant server. To partly dissociate gratitude from guilt
aversion, we modified the trust game (Berg et al., 1995) to make it a
“Pay It Forward”(PIF) task (Fig. 1). In the PIF task, subjects acted as a
Trustee,who received a sum of money between $1 and $20 from a ben-
efactor, whose picture wasshown on the screen. Subjects were told that
the benefactor was a real person, not a computer, although the endow-
ments were in fact determined by a computer. The subjects were then
shown a potential third party beneficiary, with whom the trustee
could share any portion of the endowment given by the benefactor.
They were told that the benefactor did not want the money back, but
that the benefactor wanted them to pass on what they had received if
they felt that they wanted to express gratitude for the endowment.
The beneficiary was said to not have immediate need of the money,
but would nevertheless appreciate it. We could have quantified grati-
tude simply as a Likert rating, but we chose not to. Instead, we chose
to operationalize gratitude here as money given, for several reasons.
First, it renders gratitude quantifiable in terms of monetary value,
which is necessary for the quantitative analyses we perform. Second,
Table 1
Demographics. Subjects showed no significant differences between groups regardingage,
gender, initial gratitude (GQ), or initial anxiety/depression symptoms (BHM-20).
Variable Gratitude Control P
Gender F= 17, M=5 F= 15, M=6 0.73 (Fisher exact 2 sided)
Age (range 18–34) 23.41 22.52 0.50 (2 tailed)
GQ 5.38 5.67 0.41 (2 tailed)
Symptoms
(initial BHM-20)
2.39 2.50 0.55 (2 tailed)
2P. Kini et al. / NeuroImage 128 (2016) 1–10

we reasoned that a costly expression of gratitude was more likely to en-
gage the subjects than a cost-free expression, such as a Likert rating of
experienced gratitude. Third, monetary gifts as expressions of gratitude
are ecologically valid as for example in tipping restaurant servers, with
the caveat that our task here involves payment to a third party rather
than reciprocity to the benefactor. Of note, we still measured the self-
reported experience of gratitude separately on each trial as described
below.
Subjects were placed in the fMRI scanner and performed repeated it-
erations of the PIF task. They performed five runs of 7.5-minute fMRI
blocks. Each trial began with one of ten possible benefactor faces and
names, 5 male and 5 female with neutrallevels of emotional expression
(Fig. 1). The fictitious benefactor’s name appeared above the face, and
the amount of the endowment for the trial was displayed below the
benefactor’s face. The endowment range was between $1 and $20 for
each trial. The endowment value was visible for 3 seconds followed by
a blank screen for a jittered interval with a minimum of 1 second dura-
tion, followed by a potential beneficiary, with the fictitious name of the
beneficiary above. The beneficiary was one of eight possibilities, either a
fictitious person, (neutral emotional expression, p= 0.5 female, p=0.5
Caucasian), or a politically neutral charity (Aid for Africa, Indiana Wild-
life Federation, Mother Hubbard’s Kitchen, or the Rainforest Founda-
tion). The beneficiary was visible for 3 seconds followed by a jittered
interval of a minimum one second duration, followed by a cue to select
a donation amount. Once subjects selected a donation amount via but-
ton presses, they were asked to rate on 5-point Likert scale their
a) gratitude motivation, (b) their desire to help the particularbeneficia-
ry, (c) their guilt motivation. The order of these three questions was
consistent across trials, i.e. not counterbalanced, for two reasons. First,
we wanted to ensure that gratitude self-report was always first and
thus not contaminated by order effects from subsequent questions. Sec-
ond, the consistent order of the remaining two questions made them
more predictable, which we expected would reduce confusion about
the questions and lower reaction times. This mattered because the
responses were self-paced in order to ensure that subjects provided a
response to all self-report questions on all trials. The tradeoff with our
design choice is that we cannot rule out order effects of the self-report
question responses. Still, the self-paced design maximizes trial yield
while ensuring that subjects answer every question. For validity, sub-
jects were informed that one of their interactions would be chosen at
random and actually paid out according to their choices, with the desig-
nated gift going to the charity and the remaining portion being paid to
the subject. We did in fact choose one of each subject’s choices and
pay it to the designated charity, except that when the beneficiary was
an individual, then themoney wentto a charitable organization instead
for purposes of accountability. After the self report phase of each trial
there was a jittered interval with a minimum one second duration
until the start of the following trial. Each run began with and ended
with 15 seconds of blank screen to establish a baseline of neuralactivity.
There were 27 trials per block, which led to 135 trials per subject.
We counterbalanced the order of button presses across subjects, so
that for half of the subjects, the left hand pressed buttons1 (little finger)
through 5 (thumb), and for the other half of subjects, the left hand
pressed buttons 0 (little finger), 9 (ring finger), 8 (middle finger), 7
(index finger), and 6 (thumb). The finger-to-button mapping was pre-
sented visually to the subjects during the experiment. This motor
counterbalancing avoids potential confounds between motor-related
activation and higher-level factors such as gratitude, guilt, and desire
to help, allowing the motor activation to load instead on motor regres-
sors at the time of button press. Subjects indicated the amount of money
they wished to give by pressing two buttons in succession, correspond-
ing to the tens and ones places of the number of dollars they wished to
give.
fMRI methods and analysis
Images were acquired on a 3T Siemens TIM Trio scanner using a 32-
channel head coil. Functional BOLD data were collected at a 30° angle
Fig. 1. The “Pay It Forward”task. Each trial of the task began with a benefactor, who endowed the subject with a dollar value between $1 and $20. After a variable delay, the subject was
shown a charitable cause (person or organization) to which they could donate. They were told that the benefactor requested them to donate to the charity to the extent that they
appreciated the benefactor’s gift. Subjects were then given an opportunity to type in a donation amount. Then they were asked to rate on a Likert scale how much the factors of
gratitude, desire to help the cause, and guilt influenced their decision ab out how much to give. All jittered delays lasted between 1 and 7 seconds, with an aver age of 2.14 s,
exponentially distributed. A wide range of charities and benefactor ethnicities were used in the experiment in order to maximize the variance in the responses. One response from
each subject was chosen to be actually paid —the subject was given the amount of money they chose to keep, and the donation amount was given to the charity.
3P. Kini et al. / NeuroImage 128 (2016) 1–10

from the anterior commissure–posterior commissure line in order to
maximize signal-to-noise ratio in the orbital and ventral regions of the
brain (Deichmann et al. 2003). FunctionalT2∗-weighted images were
acquired using a gradient echo planar imaging sequence with 35 axial
slices and 3.44 × 3.44 × 3.8 mm voxels; TR= 2000 ms; TE = 25 ms;
64 × 64 voxel matrix; flip angle = 70; field of view =
220mm × 220mm. High resolution T1-weighted MPRAGE images
were collected for spatial normalization excitation consisting of
160 sagittal slices (256 × 256 × 160 voxel matrix of 1 × 1 × 1 mm
voxels, TR = 1800 ms; TE = 2.67 ms; flip angle = 9) at the end of
each session.
Functional data were spike-corrected on a voxel-by-voxel basis to
reduce the impact of artifacts using AFNI's 3dDespike (http://afni.
nimh.nih.gov/afni). Subsequent preprocessing was done using SPM5
(Wellcome Department of Imaging Neuroscience London, UK; www.
fil.ion.ucl.ac.uk/spm/). Functional images were corrected for differences
in slice timing using sinc-interpolation (Oppenheim et al., 1999) and
head movement using a least-squares approach and a 6 parameter
rigid body spatial transformation. Once the resulting images were
coregistered to the structural image and normalized to the SPM stan-
dard 152T1 atlas template Montreal Neurological Institute (MNI)
space usingthe standard SPM5 normalization functions withboth affine
and nonlinear warping, the resulting functional images were then spa-
tially smoothed with an 8-mm3 full-width-at-half-maximum isotropic
Gaussian kernel. Functional neuroimaging data were statistically ana-
lyzed based on a general linear model (GLM) with random effects im-
plemented in SPM5. Each individual subject’sGLMwasestimatedwith
a canonical hemodynamic response function with no derivatives, a
microtime resolution of 16 time bins per scan, a high-pass filter cutoff
of 128 seconds using a residual forming matrix, autoregressive
AR(1) to account for serial correlations, and restricted maximum likeli-
hood (ReML) for model estimation.
For each subject, we constructed GLMs with the aim of measuring
the processes involved in the decision to express a level of gratitude
with a monetary gift. We included one regressor for each of the follow-
ing event types, each with events modeled as having zero duration: the
onset of the endowment event, the onset of the beneficiary information,
the onset of the decision time (i. e. the prompt to enter a donation
amount), the onset of the first button press (when subjects indicated
how much they wanted to “pay forward”), the onset of the second
button press (again for how much to pay forward), the onset of the
gratitude rating button press (a single digit on a Likert scale), the
onset of the desire to help rating button press (also on a Likert scale),
and the onset of the guilt motivation rating button press (also on a
Likert scale). Although the information about which button was pressed
provided valuable information (and that information was recorded),
the motor-related activity of the button press itself was unimportant,
so the button press regressors were largely nuisance regressors to ac-
count for motor-related activation that was not of interest. In addition,
we added four parametrically modulated (PM) regressors, with zero-
duration events modeled at the onset of the decision time. The height
of each modeled event was determined by the corresponding PM,
which was mean-centered for each subject. These PM regressors were
independent of the button press response regressors, because the PM
regressors were mean-centered and parametrically modulated, and
they occurred at somewhat different times. The varying parametric
modulation across events decorrelates the PM regressors from regres-
sors that model the main effect of an event. The first PM was the grati-
tude rating; the second was the guilt rating; the third was the desire
to help rating, and the fourth was the percent of the initial endowment
actually given. These PM regressors afforded an estimate of how much
each self-reported emotion correlated with activity at the time of deci-
sion. We were primarily interested in the gratitude PM regressor,
while the others served as statistical controls. It is important to note
that the order matters when entering the PM, because successive PMs
are orthogonalized with respect to the preceding ones. This means
that the first PM regressor entered has priority in accounting for any
variance shared with subsequent regressors. We also included motion
regressors for those subjects (eight in total) who had more than 3 mm
total movement across a session or 0.5 mm from one image acquisition
to another. In that case, we included 24 motion regressors, using a
Volterra expansion (Friston et al., 1996): the six degrees of freedom in-
cluding three for translation and three for rotation, the squaredvalues of
the six degrees of freedom, the scan-to-scan difference (i.e. time deriv-
ative) of the six motion regressors, and the squared values of the six
time derivative regressors. Selectively including motion regressors for
subjects with substantial movement maximizes the sensitivity to real
effects while minimizing the spurious effects of motion (Johnstone
et al., 2006). GLM regressor contrasts were computed for each subject
and then evaluated with random effects tests at the population level.
Unless otherwise noted, we analyzed each regressor across the popula-
tion by looking at the whole brain, with an initial threshold of p b0.001,
and we report as significant those regions which further passed a
cluster-correction for multiple comparisons with p b0.05, using SPM5
standard cluster correction based on random field theory.
Results
The gratitude writing intervention led to better clinical outcomes in
a larger cohort of subjects than we analyze here, but there were no sig-
nificant differences in clinical outcomes between the therapy-as-usual
control group and the gratitude group fMRI subjects here, as we elabo-
rate below. We treat the clinical questions and results more fully in a
separate paper (Wong et al., in preparation). Briefly, in a larger cohort
of several hundred subjects, those in the gratitude writing condition (in-
cluding all subjects, both those that were fMRI participants and those
that were not) reported significantly better mental health than those
in the expressive writing and therapy-as-usual control conditions
about 4 weeks and 12 weeks after the conclusion of the writing
interventions. Additionally, when the gratitude writing condition was
compared to the expressive writing condition, a lower proportion of
negative emotion words in subjects’writing mediated the effect of
condition on mental health. That is, subjects in the gratitude writing
condition used a lower percentage of negative words than those in the
expressive writing condition, which was in turn associated with better
mental health. Detailed information about these findings are described
in a separate manuscript (Wong et al., in preparation).
With regard to only the subjects who participated in the present
fMRI study, we found no significant differences in the clinical outcomes
of those in the gratitude group vs. those in the therapy-as-usual control
group. We compared the difference in BHM-20 scores one week after
the writing sessions relative to intake (two weeks before completion
of all the writing sessions). The gratitude subjects showed a trend of
greater improvement relative to therapy-as-usual controls, but this
was not significant (Gratitude group increase = 0.36, Control group
increase= 0.17, t(39)=1.44,p =0.08, one-tail). However, thegratitude
subjects did show a greater increase in gratitude (GAC) scores rela-
tive to therapy-as-usual controls at one week after the writing ses-
sions relative to intake (Gratitude = 1.27, Control = 0.06, t(40)=
2.25, p = 0.015, one-tail). This suggests that our gratitude writing in-
tervention was effective at increasing gratitude in the fMRI subjects
and provides a behavioral basis for the between groups fMRI effects
shown below.
For the Pay It Forward task, subjects gave an average of 60.5%
(range 18.4% to 91.5%) of their endowment on each trial to the ben-
eficiaries, indicating that the task was effective in eliciting altruistic
donations, and subjects neither gave all the money away nor kept
it all for themselves. Also, the average gratitude rating across sub-
jects was 2.71 (range 1 to 4.66), which was significantly greater
than the minimum possible gratitude rating (p b0.001). The average
guilt rating was 2.41 (range 1 to 4.3), which was significantly greater
than the minimum possible (p b0.001). The average desire to help
4P. Kini et al. / NeuroImage 128 (2016) 1–10

rating was 2.97 (range 1.41 to 4.26), which likewise was significantly
greater than the minimum possible (p b0.001). All of these suggest
that the subjects experienced gratitude, guilt, and desire to help as
significant factors in their decisions.
Parametric modulators
We first collapsed across both intervention groups to explore
whether the self-reported gratitude, guilt, and desire-to-help ratings
were correlated within subjects. We found that the self-reported desire
to help and guilt regressors were not correlated within a given subject
(average r= 0.01, t(42)=0.14, p = 0.89, Fishers r-to-Z transformed).
The gratitude and desire-to-help ratings were positively correlated
(average r = 0.36, t(42)=7.04, p b10
−7
, Fishers r-to-Z transformed).
The gratitude and guilt ratings were positively correlated as well
(average r = 0.10, t(42)=2.3, p = 0.026, Fishers r-to-Z transformed).
The gratitude ratings and percent of endowment given were positively
correlated (average r = 0.21, t(42)=4.53, p b0.00005, Fishers r-to-Z
transformed). This suggests that gratitude was most closely associated
with a desire to “Pay It Forward”to help the designated cause, and
while guilt may have played some role, that role was less significant.
These findings also suggest that brain activity correlating with a de-
sire to help may share some variance with brain activity correlating
with gratitude in this study, and vice versa.Furthermore,although
the percent given was unsurprisingly correlated positively with the
total amount given (average r = 0.37, t(42)=9.01, p b10
−10
,Fishers
r-to-Z transformed), there was apparently a ceiling effect –alarger
initial endowment correlated negatively with the percent given (av-
erage r=-.23, t(42)=−6.50, pb10
−7
,Fisher’s r-to-Z transformed),
indicating that subjects were reluctant to give large absolute dollar
amounts.
Neuroimaging
Across the population, the gratitude PM regressor loaded signifi-
cantly positively on four regions (Fig. 2A): in the left superior parie-
tal lobule, left superior frontal gyrus, left inferior frontal gyrus, and
right middle occipital gyrus (Table 2), using an initial uncorrected
threshold of pb0.001 to identify regions and then a cluster-correced
threshold of p b0.05 to determine which clusters were significant
after correcting for multiple comparisons. We also explored whether
any regions loaded negatively on the parametrically modulated
gratitude rating regressor. With a slightly more liberal threshold of p b
0.005 for determining candidate clusters, we found several additional
regions that passed cluster correction (Table 2), in the medial frontal
gyrus, parietal lobe, and visual cortex. We further interrogated the lim-
bic regions including the amygdala, ventral striatum, ventromedial pre-
frontal cortex, and insula, but we found no other significant loading on
the gratitude rating regressor, either positively or negatively, even at an
uncorrected threshold of pb0.001.
For the Guilt PM regressor, there were no brain regions that passed
correction. For the Desire to help PM regressor, there were four brain
regions that reached significance, in the left mid-occipital region, left
precental gyrus (BA6), left superior parietal lobule (BA7), and left
cerebellar declive (Fig. 2B, Table 3).
Since the gratitude rating regressor was entered first, it had priority
for accounting for the variance in the BOLD signal at the time of deci-
sions. We tested how robust the gratitude effect was by constructing
two additional GLMs: the “guilt-then-gratitude”GLM was identical to
the original GLM, except that we entered the guilt regressor first, and
then gratitude (followed by desire to help, then percent given). This af-
fords a test of the gratitude effect while “partialing out”the effects of
guilt. The gratitude effect cluster remained significant in the left superi-
or parietal lobule (MNI −22, −58, 54; p b0.001, cluster-corrected). The
Regressor loading
Gratitude (positive)
Gratitude (negative)
Desire-to-help (positive)
Z=56 Z=28
Z=16
Z=3.09 Z=4
B
Gratitude
A
Desire-to-help
Fig. 2. Neural correlates of motivations during decision-making. (A) Gratitude related neural activity. A regressor at the time of decision-making is parametrically modulated by self-
reported gratitude for each trial. The regressor is mean-centered. Regions in red represent areas that showed a significant positive loading on the gratitude rating regressor across the
population of all subjects. Regions in green show negative loading. Shown at an uncorrected threshold of p b0.001 for visualization purposes. Significant clusters are shown in Table 2.
(B) Desire-to-help related neural activity. As in Fig. 2A, except that the parametrically-modulated regressors track the self-reported desire to help the beneficiary on each trial.
5P. Kini et al. / NeuroImage 128 (2016) 1–10

peak voxels in BA6 and BA9 listed in Table 2 remained significant (p b
0.001, uncorrected), and the region in BA18 was not significant.
In the “help-then-gratitude”GLM, we entered the desire-to-help re-
gressor first, and then gratitude (followed byguilt, then percent given).
In this GLM, the gratitude effect remained significant in a cluster of 60
voxels in the superior parietal lobule region identified earlier (peak
MNI −26, −58, 56, p b0.001, uncorrected), however the cluster as a
whole did not pass cluster correction. No other regions showed a signif-
icant effect. This confirms that gratitude-related activation shares vari-
ance with a desire to help, but neither desire to help nor guilt can
completely account for neural activity that correlates with gratitude.
We hypothesized that neural activity related to gratitude in the grat-
itude task would correlate positively with trait measures of gratitude, in
particular the GQ and the GAC3 (Mccullough et al., 2002). To test this,
we collapsed across both intervention groups and tested for a correla-
tion between the GQ and GAC3 self-report measures and each of the
four PM regressors of gratitude, guilt, desire to help, and percent of
endowment given. We found no correlations between self-reported
gratitude and the parametrically modulated regressors for gratitude or
guilt. We did find positive correlations between the GQ and the
desire-to-help regressor in the bilateral supplementary motor area
(MNI 6, −12, 62; p b0.001, cluster corrected), which is activated with
willful actions (Debaere et al., 2003). A scatter plot of this region
showed that the result was strongly driven by an outlier (Fig. 3). To
investigate this further, we recomputed the correlation without the
outlier and found that the correlation was still significant and positive
(r-squared = 0.154, t(40)=2.70, p= 0.01). We also found positive cor-
relations between the GAC3 and the percent given parametric modula-
tor in the medial prefrontal cortex (MNI −6, 42, 30, p b0.001, cluster
corrected), as shown in Fig. 3. No other regions correlated with the
trait gratitude measures.
Effect of gratitude intervention
We hypothesized that the gratitude writing intervention would lead
to measurable changes in brain activity, and specifically in greater
neural activity related to the expression of gratitude. We thus explored
the GLM beta weights for the gratitude PM regressor at the time of
decision making. We compared these beta weights voxel-by-voxel in
those that received the gratitude writing intervention relative to those
that received only psychotherapy. Using a between-groups random
effects t-test, we found that relative to the therapy-as-usual control
group, the gratitude-writing group showed greater loading on the
gratitude PM regressor in a single region, the perigenual anterior cingu-
late cortex (p b0.05, cluster corrected, MNI −10, 38, 2), as shown in
Fig. 4. This dovetails with our finding above that the gratitude writing
intervention led to significantly greater self-reported gratefulness in
the two weeks following the intervention relative to the therapy-as-
usual control group. Also, there was significantly greater loading on
the percent-given PM regressor in the gratitude intervention group rel-
ative to the therapy-as-usual control group, in the right thalamus at MNI
10, −28, 0 (p b0.05, cluster corrected). There was no between-group
effect of the intervention on the guilt PM regressor, nor on the desire-
to-help PM regressor. These results are consistent with an effect of neu-
ral plasticity, but given that we did not collect baseline functional scans
on subjects prior to the gratitude writing intervention, we cannot defin-
itively rule out a pre-existing group difference. This is less likely though
given that our subjects were randomized to the three intervention
groups, and there were no age, gender, or initial symptom differences
between the groups.
Discussion
Until now, very few studies have attempted to determinethe neural
correlates of gratitude. Structural imaging studies have shown relation-
ships between cortical volume, especially larger right inferior temporal
cortical volume, and gratitude traits (Zahn et al., 2014). An earlier study
showed that individuals who identify gratitude in a social narrative
more often showed greater hypothalamic activity while reading
sentences that describe a social interaction (Zahn et al., 2009). This
may reflect significant physiological effects of recognizing gratitude.
Our findings (Fig. 2) show that greater gratitude expression generally
correlated more with activity in parietal and lateral prefrontal cortex
rather than with activity in the limbic regions. We found activity corre-
lating with gratitude specifically extending across the intraparietal sul-
cus and inferior frontal gyrus, both of which have previously been
implicated in mental arithmetic specifically (Dehaene et al., 2004;
Simon et al., 2002). This is consistent with the nature of our task,
which required subjects to operationalize and quantify their gratitude
as a payment amount. A very recent paper also shows a correlation be-
tween self-reported gratitude experience and pre-genual medial pre-
frontal cortex activity (Fox et al., 2015), in a region that appears to
overlap with the region showing between-groups effect of the gratitude
intervention in our study (Fig. 4). This provides converging evidence for
this region’s role in gratitude-related cognitive processes.
Gratitude consists of both experience and expression, and our task
also required a reflective self-report of gratitude experience after each
gratitude expression. For our GLM analysis, we chose to model the
gratitude-related neural activity at the time when gratitude was
expressed, i.e. when they chose how much money to pay forward. An
alternative possibility would be to model gratitude earlier, at the time
when the initial endowment was received and gratitude may have
been experienced, or later when subjects self-report their degree of
gratitude motivation. Our analysis reflects a construction of gratitude
as an expression rather than an experience per se. Put another way,
gratitude has prosocial effects by virtue of its expression —if one
experiences positive affect as a consequence of receiving benefit, the
Table 2
Gratitude-related activation. Regions that load on the gratitude parametric modulator at the time of decision making. Clusters defined at an uncorrected threshold of p b0.001 (positive
modulation) or p b0.005 (negative modulation). The “Cluster P”shows the cluster-corrected probability of the cluster.
Positive modulation Cluster P Cluster size MNI (x, y ,z) Area
0.001 596 −26, −58, 54 BA7, left superior parietal lobule
0.002 332 −24, −6, 60 BA6, left superior frontal gyrus
0.003 307 −52, 8, 26 BA9, left inferior frontal gyrus
0.023 196 38, −82, −14 BA18, right middle occipital gyrus
Negative modulation 0.013 513 6, 52, 16 BA10, right medial frontal gyrus
0.048 390 60, −48, 32 BA40, right supramarginal gyrus
b0.001 1032 −10, −84, 28 BA19, left visual cortex
Table 3
Desire to helpeffects. Regions thatload on the desire-to-helpparametric modulator at the
time of decision making. Clusters defined at an initial uncorrected threshold of p b0.001,
and the “Cluster P”shows the cluster-corrected probability of the cluster.
Cluster P Cluster size MNI (x, y ,z) Area
0.001 404 −30, −80, 36 Left mid-occipital
0.004 232 −30, −12, 58 BA6, left precentral gyrus
0.001 279 −30, −56, 58 BA7, left superior parietal lobule
0.025 157 −20, −74, −18 Left cerebellar declive
6P. Kini et al. / NeuroImage 128 (2016) 1–10

experience has no direct prosocial value unless it is converted into an
expression of gratitude. It is the activity during the expression of grati-
tude that we measure here. Of note, subjects also evaluated and self-
reported their experience of gratitude in deciding how much money
to give, butthat self-report happened later in each trial. Our neuroimag-
ing analyses thus modeled neural activity at the time when subjects de-
cided how much to give, and the regressors were parametrically
modulated by the subsequently self-reported gratitude, desire-to-help,
and guilt motivation values in each trial.
It is also noteworthy that our “Pay It Forward”task involves grati-
tude expressed as monetary gifting to a third party. This differs some-
what from a more typical expression of gratitude as reciprocal, in that
gratitude is expressed to the person one is grateful to. We believe this
approach is justified on the basis that the benefactor specifically re-
quests that gratitude be expressed this way, and also many subjects
did report experiencing substantial gratitude as a motivating factor in
their task decisions, as measured by the self-report during the task.
Still, our analysis subtly assumes that subjects modulated their donation
amount in proportion to how grateful they felt, per the instructions,
such that the donation amount was directly driven by the felt gratitude.
This assumption would be violated to the extent that subjects were not
intrinsically motivated to donate more out of gratitude, but instead
were simply attempting to comply with the experimenter instructions.
Nevertheless, in either case, the regions where activity correlates with
self-reported gratitude ratings may reflect an operationalized expres-
sion of gratitude.
Besides gratitude experience vs. expression vs. evaluation, the motor
effects of button pressing present a potential confound, so we carefully
counterbalanced the subjects to control motor confounds and included
separate motor regressors as nuisance covariates. We counterbalanced
across subjects the motor (button press) mappings to different levels
of gratitude expression (operationalized as monetary gifts), so neural
activity related to gratitude expression is not confounded with simple
motor activity here.
-3
-2
-1
0
1
2
0510
GQ vs. HelpRating
-3
-2
-1
0
1
2
0510
GAC3 vs. Pct. Given
X=4
Z=3.09
Z=4
Z=3.09
Z=4
Y=48Y=-12
Fig. 3. Neural correlates of trait gratitude measures. The region in orange shows a significant correlationacross all subjects between the GQ self-report measure (Mccullough et al., 2002)
and the GLM regressor that is parametrically modulated by the desire to help (peak MNI= 6,-12,62. p b0.001 cluster corrected, 399 voxels. Peak voxel Z = 4.40). The corresponding
scatterplot is shown in orange. The correlation in this region remains significant even when excluding the outlier (r-squared = 0.154, p= 0.01). The medial prefrontal region in violet
shows a significant correlation across all subjects between the GAC3 self report measure (Mccullough et al., 2002) and the GLM regressor that is parametrically modulated by the
percent of the initial endowment given (peak MNI = −6, 42, 30. pb0.001, cluster corrected, 622 voxels. Peak voxel Z=4.38). The corresponding scatterplot is shown in violet. All
contrasts are visualized at p b0.001 uncorrected.
Z=3.09 Z=4
-0.3
-0.15
0
0.15
Y=36X=-10 Z=4
Fig. 4. Effects of prior gratitude intervention on brain activity three months post-treatment. The contrast shows the greater loading of the gratitude rating parametrically modulated
regressor on the gratitude intervention group relative to the control group. The pregenual anterior cingulate cluster shown (Peak MNI = −10, 38, 2) is significant at the level of the
cluster, p b0.05. Inset: The gratitude PM regressor loads positively on the region in the gratitude group (arbitrary units), but it loads negatively in the therapy-as-usual control group.
7P. Kini et al. / NeuroImage 128 (2016) 1–10

Tables 2 and 3 summarize the brain regions that are modulated by
self-reported gratitude and desire to help. It is essential to recall that
the activities are measured at the time when a decision is made regard-
ing how much to give, and not at the time later when motivations are
reported. We find that a number of regions in the frontal, parietal, and
occipital lobes show greater activity when gratitude motivations is
greater. Some regions notably show activity that is decreased when
gratitude motivation is greater.
We also explored whether gratitude involves brain regions related
to social and affective processes, and we found some evidence in sup-
port of this hypothesis. Gratitude may involve positive affect and a
focus on other people rather than self. In that sense, it is perhaps most
related to empathy, which may be either positively or negatively
valenced, and that is other-oriented as well (Tangney et al., 2007). The
neurobiology of empathy has been studied thoroughly and generally in-
volves the anterior cingulate cortex and anterior insula (Singer et al.,
2004, 2009), as well as the amygdala and pars opercularis in the inferior
frontal gyrus (IFG) (Bzdok et al., 2012; Carr et al., 2003; Moll et al., 2002;
Shamay-Tsoory et al., 2009). The region we found that loaded on the
gratitude parametric modulated regressor (Fig. 2A) was found in BA9,
which was distinct from earlier reports of empathy effects in the pars
opercularis BA44. Nevertheless, our between-groups finding of greater
activity in the pregenual anterior cingulate cortex are consistent with
recent reports (Fox et al., 2015), which may be consistent with empa-
thy, theory of mind, and moral cognition-related activation.
The constructs of cognitive mentalizing and theory of mind (ToM)
are closely related and have been found to involve overlapping brain re-
gions (Bzdok et al., 2012), though they were found to be distinct in that
they are connected with different networks (Vollm et al., 2006). Cogni-
tive mentalizing and ToM operations involve overlapping activations in
the medial prefrontal cortex, superior temporoparietal junction (TPJ),
and temporal poles, however the ToM also includes activations in
the frontal gyrus, cuneus, and superior temporal gyrus (Vollm et al.,
2006). Notably, there are no common regions of brain activation
involved in cognitive mentalizing that overlap with regions found to
be activated in the present study. Other than the activation in the IFG,
significant activations were found in the left superior parietal lobule,
left superior frontal gyrus, and the right middle occipital gyrus, all of
which are not within the same regions as common literature observing
the cognitive mentalizing system. Still, our results show activation relat-
ed to gratitude and the desire to help (Fig. 2B) in the left superior pari-
etal lobule (BA7). This region is at least adjacent to the TPJ, which is
involved in ToM (Saxe and Kanwisher, 2003), among other functions
(Mitchell, 2008). Overall, a recent meta-analysis has identified regions
associated with ToM, empathy, and moral cognition (Bzdok et al.,
2012), but we find little if any overlap between the regions identified
by the meta-analysis and the regions showing significant effects of
gratitude here.
It was particularly noteworthy that the gratitude intervention was
associated with such a lasting increase in pregenual anterior cingulate
responsiveness to gratitude, even three months later (Fig. 4). The inter-
vention lasted around an hour in all, which is relatively short given the
lasting nature of the effects. This finding may reflect a neural plasticity
effect, but it must also be treated with caution. This was a cross-
sectional study. The subjects were randomized into groups, and there
were no significant differences in the age, gender, trait gratitude (GQ)
or clinical symptom scores between groups. A significant constraint on
interpreting our results stems from the fact that we did not perform
initial fMRI scans pre-intervention due to budget constraints. Thus, we
cannot definitively rule out a possible pre-existing between-group
difference in neural sensitivity to gratitude, despite the absence of
pre-existing differences in trait gratitude. The pre-intervention scans
would be necessary to conclude that there is a neural plasticity effect,
and a follow-up study would be warranted by these results. Also, the
therapy-as-usual control group can be considered a passive control, as
we simply omitted the gratitude writing intervention. This means that
part of the group differences could be due to the writing itself, apart
from the specific content of the writing. We considered using the
expressive writing condition as an active control group, but given the
absence of similar previous studies, we elected to cast a wide net and
leave the further dissection of gratitude writing effects to future studies.
The particular region showing between-groups effects in Fig. 4 over-
laps with another regionvery recently reported as correlating with grat-
itude (Fox et al., 2015). It also overlaps with a similar region found to be
altered by mindfulness interventions (Allen et al., 2012), which have
also been shown to increase gratitude (Shapiro et al., 2002). Functional-
ly, the pregenual anterior cingulate has been shown to be involved in
predicting the outcomes of actions (Alexander and Brown, 2011; Jahn
et al., 2014) and in dual task performance (Dreher, 2003). This suggests
a possible mechanistic account of the gratitude intervention: specifical-
ly, it may increase the neural activity related to predicting the effects of
one’s actions on another person. To the extent one predicts and evalu-
ates the likely effects of one’s actions on others, one might be more will-
ing to direct those actions towards having a positive impact on others.
For example, individuals who role-play typically report greater empa-
thy for the subjects whose roles they are playing (Poorman, 2002).
The measures of trait gratitude (especially the GQ) showed a corre-
lation between gratitude and the activity related to desire to help in the
supplementary motor area (SMA), as shown in Fig. 3. The desire to help
is conceptually related to altruism, which may also be driven by empa-
thy (De Waal, 2008). In any case, the SMAis generally involved in motor
functions and may reflect a greater neural preparation to act as an ex-
pression of gratitude. Studies of altruism have generally shown activity
in the posterior superior temporal cortex, which has been suggested to
be related to prediction of the beliefs of others or their actions in an en-
vironment (Saxe and Kanwisher, 2003; Singer et al., 2004; Tankersley
et al., 2007; Vollm et al., 2006). Thus, our results suggest that gratitude
and the neural underpinnings of a desire to help in the SMA may be
distinct from effects related to altruism.
Similarly, trait gratitude (especially the GAC3) showed a correlation
with neural activity in the medial prefrontal cortex related to the per-
cent of the endowment given. This region is adjacent to the region
that showed lasting effects of the gratitude intervention and may reflect
a cognitive process related to predicting the results of one’sactions
(Jahn et al., 2014).
Our studyhas a few limitations. First, because the gratitude rating re-
gressor and the desire to help regressor were partially correlated in the
GLM, we cannot be sure that the regions showing a correlation with
gratitude would not also show some correlation with the desire to
help, and vice versa. We partially addressed this by entering the guilt
and desire to help regressors before gratitude in a follow-up GLM anal-
ysis. This essentially partialed out the guilt and desire to help signals in
the neural activity, and we found that the gratitude rating regressor
loadings remained significant in several of the regions. Still, future stud-
ies may be able to decorrelate these factors in the experimental design,
but that was not possible with our self-report approach. Second, it is
possible that the scanner environment and static pictures of benefactors
and beneficiaries may not fully recreate the social experience of
interacting with people in a prosocial manner. This is a limitation that
confronts all such studies of social interactions to varying degrees. This
may partly account for the relative lack of affective brain region activity
observed here. Nevertheless, the task was valid in that participants
knew that they would get to keep the part of an endowment they
chose to keep and that the remaining part would in fact be given to a
charitable cause. All subjects did in fact choose to donate part of their
endowments and reported experiencing gratitude, which suggests
that the task was successful in evoking prosocial experiences and
actions.
Another potential limitation of our study is that we recruited from a
population seeking counseling for anxiety and/or depression. There was
no significant difference between the two groups in terms of the mental
health scores either before or after the interventions, but the average
8P. Kini et al. / NeuroImage 128 (2016) 1–10

mental health score at intake reflected more anxiety and/or depression
than the normal healthy population. For this reason, we cannot rule out
the possibility that the effects we observed may differ in a normal
healthy population.
Overall, our results suggest a nuanced view of the neural mecha-
nisms of gratitude. It is somewhat distinct at the neural level from em-
pathy, theory of mind, and altruism, despite some modest overlap. It
involves neural mechanisms associated with predicting the effects of
one’s actions, mental arithmetic and calculations, and carrying out mul-
tiple tasks at once. We show here that even brief expressions of grati-
tude may have profound and lasting effects on neural activity and
sensitivity, perhaps related to monitoring of self and others, which
may have implications for practices and interventions involving grati-
tude expression.
Acknowledgments
This work was supported in part by the Expanding the Science and
Practice of Gratitude Project run by UC Berkeley's Greater Good Science
Center in partnership with UC Davis with funding from the John
Templeton Foundation. We thank K. McKinney and S. Berry for help
with MR scanning.
References
Alexander, W.H., Brown, J.W., 2010. Computational models of performance monitoring
and cognitive control. Top. Cogn. Sci. 2, 658–677.
Alexander, W.H., Brown, J.W., 2011. Medial prefrontal cortex as an action-outcome pre-
dictor. Nat. Neurosci. 14, 1338–1344. http://dx.doi.org/10.1038/nn.2921.
Allen, M., Dietz, M., Blair, K.S., van Beek, M., Rees, G., Vestergaard-Poulsen, P., Lutz, A.,
Roepstorff, A., 2012. Cognitive–affective neural plasticity following active-controlled
mindfulness intervention. J. Neurosci. http://dx.doi.org/10.1523/JNEUROSCI.2957-
12.2012.
Baxter, H.J., Johnson, M.H., Bean, D., 2012. Efficacy of a character strengths and gratitude
intervention for people with chronic back pain. Aust. J. Rehabil. Couns. 18, 135–147.
http://dx.doi.org/10.1017/jrc.2012.14.
Berg, J., Dickhaut, J., McCabe, K., 1995. Trust, reciprocity, and social history. Games Econ.
Behav. 10, 122–142.
Boehm, J.K., Lyubomirsky, S., Sheldon, K.M.,2011. A longitudinal experimental study com-
paring the effectiveness of happiness-enhancing strategies in Anglo Americans and
Asian Americans. Cogn. Emot. http://dx.doi.org/10.1080/02699931.2010.541227.
Brown, J.W., Braver, T.S., 2007. Risk prediction and aversion by anterior cingulate cortex.
Cogn. Affect. Behav. Neurosci. 7, 266–277.
Bzdok, D., Schilbach, L., Vogeley, K., Schneider, K., Laird, A.R., Langner, R., Eickhoff, S.B.,
2012. Parsing the neural correlates of moral cognition: ALE meta-analysis on moral-
ity, theoryof mind, and empathy. Brain Struct. Funct. 217, 783–796.http://dx.doi.org/
10.1007/s00429-012-0380-y.
Carr, L., Iacoboni, M., Dubeau, M.-C., Mazziotta, J.C., Lenzi, G.L., 2003. Neural mechanisms
of empathy in humans: a relay from neural systems for imitation to limbic areas.
Proc. Natl. Acad. Sci. U. S. A. 100, 5497–5502. http://dx.doi.org/10.1 073/pnas.
0935845100.
Cheng, S.-T., Tsui, P.K., Lam, J.H.M., 2015. Improving mental health in health care practi-
tioners: random ized controlled tr ial of a gratitude inte rvention. J. Consult. Clin.
Psychol. 83, 177–186. http://dx.doi.org/10.1037/a0037895.
Chiu, P.H.,Kayali, M.A., Kishida, K.T., Tomlin, D., Klinger, L.G., Klinger, M.R.,Montague, P.R.,
2008. Self responses along cingulate cortex reveal quantitative neural phenotype for
high-functioning autism. Neuron 57, 463–473. http://dx.doi.org/10.1016/j.neuron.
2007.12.020.
Coricelli,G., Critchley, H., Joffily, M., O’Doherty, J., Sirigu, A., Dolan,R.J., 2005. Regret and its
avoidance: a neuroimaging study of choice behavior. Nat. Neurosci. 8, 1255–1262.
De Quervain, D.J.-F., Fischbacher, U., Treyer, V., Schellhammer, M., Schnyder, U., Buck, A.,
Fehr, E., 2004. The neural basis of altruistic punishment. Science 305, 1254–1258.
http://dx.doi.org/10.1126/science.1100735.
De Waal, F.B.M., 2008. Putting the altruism back into altruism: the evolution of empathy.
Annu.Rev.Psychol.59,279–300. http://dx.doi.org/10.1146/annurev.psych.59.
103006.093625.
Debaere, F., Wenderoth, N., Sunaert, S., Van Hecke, P., Swinnen, S.P., 2003. Internal vs ex-
ternal generation of movements: differential neural pathways involved in bimanual
coordination performed in the presence or absence of augmented visual feedback.
NeuroImage 19, 764–776. http://dx.doi.org/10.1016/S1053-8119(03)00148-4.
Dehaene, S., Molko, N., Cohen, L., Wilson, A.J., 2004. Arithmetic and the brain. Curr. Opin.
Neurobiol. http://dx.doi.org/10.1016/j.conb.2004.03.008.
Deichmann, R., Gottfried, J.A., Hutton, C., Turner, R., 2003. Optimized EPI for fMRI studies
of the orbitofrontal cortex. NeuroImage 19, 430–441.
Dreher, J.-C., 2003. Dissociating the roles of the rostral anterior cingulate and the lateral
prefrontal cortices in performing two tasks simultaneously or successively. Cereb.
Cortex 13, 329–339. http://dx.doi.org/10.1093/cercor/13.4.329.
Emmons,R.A., McCullough,M.E., 2003. Counting blessingsversus burdens: anexperimen-
tal investigati on of gratitude and su bjective well-b eing in daily life. J. Pers. Soc.
Psychol. http://dx.doi.org/10.1037/0022-3514.84.2.377.
Emmons, R.A., Mishra, A., 2012. Why gratitude enh ances well-being: what we know,
what we need to know. In: Kashdan, T., Steger, M.F. (Eds.), Designing the Future of
Positive Psychology: Taking Stock and Moving Forward. Oxford Univer sity Press,
New York.
Emmons, R.A., Stern, R., 2013. Gratitude as a psychotherapeutic Intervention. J. Clin.
Psychol. 69, 846–855. http://dx.doi.org/10.1002/jclp.22020.
Flinchbaugh, C.L., Moore, E.W.G., Chang, Y.K., May, D.R., 2012. Student well-being inter-
ventions: the effects of stress management techniques and gratitude journaling in
the management education classroom. J. Manag. Educ. 36, 191–219. http://dx.doi.
org/10.1177/1052562911430062.
Fox, G.R., Kaplan, J., Damasio, H., Damasio, A., 2015. Neural correlates of gratitude. Front.
Psychol. 6. http://dx.doi.org/10.3389/fpsyg.2015.01491.
Friston, K.J., William s, S., Howard, R., Frac kowiak, R.S.J., Turner , R., 1996. Movement-
related effects in fMRI time-series. Magn. Reson. Med. 35, 346–355.
Froh, J.J., Sefick, W.J., Emmons, R.A., 2008. Counting blessings in early adolescents: an ex-
perimental study of gratitude and subjective well-being. J. Sch. Psychol. 46, 213–233.
http://dx.doi.org/10.1016/j.jsp.2007.03.005.
Fukunaga, R., Brown, J.W., Bogg, T., 2012. Decision making in the Balloon Analogue Risk
Task (BART): anterior cingulate cortex signals loss aversion but not the infrequency
of risky choices. Cogn. Affect. Behav. Neurosci. 12, 479–490. http://dx.doi.org/10.
3758/s13415-012-0102-1.
Geraghty, A.W.A., Wood, A.M., Hyland, M.E., 2010. Dissociating the facets of hope: agency
and pathways predict dropout from unguided self-help therapy in opposite direc-
tions. J. Res. Pers. 44, 155–158. http://dx.doi.org/10.1016/j.jrp.2009.12.003.
Harris, L.T., McClure, S.M., van den Bos, W., Cohen, J.D., Fiske, S.T., 2007. Regions of the
MPFC differentially tuned to social and nonsocial affective evaluation. Cogn. Affect.
Behav. Neurosci. 7, 309–316. http://dx.doi.org/10.3758/CABN.7.4.309.
Jahn, A., Nee, D.E., Alexander, W.H., Brown, J.W., 201 4. Distinct regions of anterior
cingulate cortex signal prediction and outcome evaluation. NeuroImage 95, 80–89.
http://dx.doi.org/10.1016/j.neuroimage.2014.03.050.
Johnstone, T., Ores Walsh, K.S., Greischar, L.L., Alexander, A.L., Fox, A.S., Davidson, R.J.,
Oakes, T.R., 2006. Motion correction and the use of motion covariates in multiple-
subject fMRI analysis. Hum. Brain Mapp. 27, 779–788. http://dx.doi.org/10.1002/
hbm.20219.
Kaczmarek, L.D., Kashdan, T.B., Drążkowski, D., Enko, J., Kosakowski, M., Szäefer, A., Bujacz, A.,
2015. Why do people prefer gratitude journaling over gratitude letters? The influence
of individual differences in motivation and personality on web-based interventions.
Personal.Individ.Differ.75,1–6. http://dx.doi.org/10.1016/j.paid.2014.11.004.
Kerr, S.L., O’Donovan, A., Pepping, C.A., 2015. Can gratitude and kindness interventions
enhance well-being in a clinical sample? J. Happiness Stud. 16, 17–36. http://dx.doi.
org/10.1007/s10902-013-9492-1.
Killen, A., Macaskill, A., 2015. Using a gratitude intervention to enhance well-being in
older adults. J. Happiness Stud. 16, 947–964. http://dx.doi.org/10.1007/s10902-014-
9542-3.
King-Casas, B., Tomlin, D., Anen, C., Camerer, C.F., Quartz, S.R., Montague, P.R., 2005. Get-
ting to know you: reputation and trust in a two-person economic exchange. Science
308, 78–83. http://dx.doi.org/10.1126/science.1108062 (80-., doi:308/5718/78 [pii]).
Kopta, S.M., Low ry, J.L., 2002. Psy chometric evaluation of the beha vioral health
questionnaire-20: a brief instrument for assessing global mental health and the
three phases of psychotherapy outcome. Psychother. Res. 12, 413–426. http://dx.
doi.org/10.1093/ptr/12.4.413.
Layous, K., Chancellor, J., Lyubomirsky, S., Wang, L., Doraiswamy, P.M., 2011. Delivering
happiness: translating positive psychology intervention research for treating major
and minor depressive disorders. J. Altern. Complement. M ed. 17 (8), 675–68 3.
http://dx.doi.org/10.1089/acm.2011.0139.
Lyubomirsky, S., Dickerhoof, R., Boehm, J.K., Sheldon, K.M., 2011. Becoming happier takes
both a will and a proper way: an experimental longitudinal intervention to boost
well-being. Emotion 11, 391–402. http://dx.doi.org/10.1037/a0022575.
McCullough, M.E., Kilpatrick, S.D., Emmons, R.A., Larson, D.B., 2001. Is gratitude a moral
affect? Psychol. Bull. 127, 249–266. http://dx.doi.org/10.1037/0033-2909.127.2.249.
Mccullough, M.E., Emmons, R.A., Tsang, J.-A., 2002. The grateful disposition: a conceptual
and empirical topography. J. Pers. Soc. Psychol. 82, 112–127. http://dx.doi.org/10.
1037/0022-3514.82.1.112.
Mitchell, J.P., 2008. Activity in right temporo-parietal junction is not selective for theory-
of-mind. Cereb. Cortex 18, 262–271. http://dx.doi.org/10.1093/cercor/bhm051.
Moll, J., de Oliveira-Souza, R., Eslinger, P.J., Bramati, I.E., Mourão-Miranda, J., Andreiuolo, P.A.,
Pessoa, L., 2002. The neural correlates of moral sensitivity: a functional magnetic reso-
nance imaging investigation of basic and moral emotions. J. Neurosci. (doi:20026214).
O’Leary, K., Dockray, S., 2015. The effects of two novel gratitude and mindfulness inter-
ventions on well-being. J. Altern. Complement. Med. 21, 243–245. http://dx.doi.org/
10.1089/acm.2014.0119.
Oppenheim, A.V., Shafer, R.W., Buck, J.R., 1999. Discrete Time-Signal Processing. Prentice
Hall, Upper Saddle River, NJ.
Poorman, P.B., 2002. Biography and role playing: foster empathyin abnormal psychology.
Teach. Ps ychol. 29, 32 –36. http://dx.doi.org/10.1207/S15328023TOP2901_08.
Rameson, L.T., Morelli, S.A., Lieberman, M.D., 2012. The neural correlates of empathy:
experience, automaticity, and prosocial behavior. J. Cogn. Neurosci. 24, 235–24 5.
http://dx.doi.org/10.1162/jocn_a_00130.
Sanfey, A.G., Rilling, J.K., Aronson, J.A., Nystrom, L.E., Cohen, J.D., 2003. The neural basis of
economic decision-making in the ultimatum game. Science 300, 1755–1758 (80-.).
Saxe, R., Kanwisher, N., 2003. People thinking about thinking peo ple: the role of
the temporo-parietal junction in “th eory of mind.”. NeuroImage 19, 1835–1842.
http://dx.doi.org/10.1016/S1053-8119(03)00230-1.
9P. Kini et al. / NeuroImage 128 (2016) 1–10

Seligman, M.E.P., Steen, T.A., Park,N., Peterson, C.,2005. Positive psychology progress. Am.
Psychol. 60, 410–421. http://dx.doi.org/10.1037/0003-066X.60.5.410.
Shamay-Tsoory, S.G., Aharon-Peretz, J., Perry, D., 2009. Two systems for empathy: a dou-
ble dissociation between emotional and cognitive empathy in inferior frontal gyrus
versus ventromedial prefrontal lesions. Brain 132, 617–627. http://dx.doi.org/10.
1093/brain/awn279.
Shapiro, S.L., Schwartz, G.E.R., Santerre, C., Shapiro, S.L., Gary Schwartz, C.S., 2002. Medita-
tion and positive psychology. Handbook of Positive Psychology, pp. 632–645.
Sheldon, K.M., Lyubomirsky, S., 2006. How to increase and sustain positive emotion: the
effects of expressing gratitude and visualizing best possible selves. J. Posit. Psychol.
http://dx.doi.org/10.1080/17439760500510676.
Simon, O., Mangin, J.F., Cohen, L., Le Bihan, D., Dehaene, S., 2002. Topographical layout of
hand, eye, calculation, andlanguage-relatedareas in the human parietallobe. Neuron
33, 475–487. http://dx.doi.org/10.1016/S0896-6273(02)00575-5.
Singer, T., Seymour, B., O’Doherty, J., Kaube, H., Dolan, R.J., Frith, C.D., 2004. Empathy for
pain involves the aff ective but not senso ry components of pain. Science 303,
1157–1162. http://dx.doi.org/10.1126/science.1093535 303/5661/1157 (80-., pii).
Singer, T.,Seymour, B., O’Doherty, J.P., Stephan, K.E., Dolan, R.J., Frith, C.D., 2006. Empathic
neural responses are modulated by the perceived fairness of oth ers. Nature 439,
466–469. http://dx.doi.org/10.1038/nature04271.
Singer, T., Critchley, H.D., Preuschoff, K., 2009. A common role of insula in feelings, empa-
thy and uncertainty. Trends Cogn. Sci. 13, 334–340. http://dx.doi.org/10.1016/j.tics.
2009.05.001 (doi:S1364-6613(09)00135-1 [pii]).
Tangney, J.P., Stuewig, J., Mashek, D.J., 2007. Moral emotions and moral behavior. Annu. Rev.
Psychol. 58, 345–372. http://dx.doi.org/10.1146/annurev.psych.56.091103.070145.
Tankersley, D., Stowe, C.J., Huettel, S.A., 2007. Altruism is associated with an increased
neural response to agency. Nat. Neurosci. 10, 150–151. http://dx.doi.org/10.1038/
nn1833.
Tomlin, D., Kayali, M.A.,King-Casas, B., Anen,C., Camerer, C.F., Quartz, S.R., Montague, P.R.,
2006. Agent-specific responses in the cingulate cortex during economic exchanges.
Science 312, 1047–1050. http://dx.doi.org/10.1126/science.1125596.
Vollm, B.A., Taylor, A.N., Richardson, P., Corcoran, R., Stirling, J., McKie, S., Deakin, J.F.,
Elliott, R., 2006. Neuronal correlates of theory of mind and empathy: a functional
magnetic resona nce imaging study in a nonverbal task. NeuroImage 29, 90–98
(doi:S1053-8119(05)00511-2 [pii]\r10.1016/j.neuroimage.2005.07.022).
Watkins, P.C., Woodward, K., Stone, T., Kolts, R.L., 2003. Gratitude and happiness: devel-
opment of a measure of gratitude, and relationships with subjective well-being.
Soc. Behav. Personal. Int. J. http://dx.doi.org/10.2224/sbp.2003.31.5.431.
Wong, J., Owen, J., Gabana,N., Brown, J., McInnis, S., Toth, P.,Gilman, L., n.d. Doesgratitude
writing improve the mental health of psychotherapy clients? Evidence from a Ran-
domized Controlled Study.
Wood, A.M., Maltby, J., Stewart, N., Linley, P.A., Joseph, S., 2008. A social-cognitive model
of trait and state levels of gratitude. Emotion 8, 281–290. http://dx.doi.org/10.1037/
1528-3542.8.2.281.
Zahn, R., Moll, J., Paiva,M., Garrido, G., Krueger, F., Huey,E.D., Grafman, J., 2009. The neural
basis of human social values: evidence from functional MRI. Cereb. Cortex 19,
276–283. http://dx.doi.org/10.1093/cercor/bhn080.
Zahn, R., Garrido, G., Moll, J., Grafman, J., 2014. Individual differences in posterior cortical
volume correlate withproneness to prideand gratitude. Soc.Cogn. Affect. Neurosci. 9,
1676–1683. http://dx.doi.org/10.1093/scan/nst158.
10 P. Kini et al. / NeuroImage 128 (2016) 1–10