See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/247497268

The Debt of Gratitude: Dissociating Gratitude and Indebtedness

Article  in  Cognition and Emotion · February 2006

DOI: 10.1080/02699930500172291

CITATIONS READS

152 3,169

4 authors, including:

Philip Watkins Melinda Ovnicek

Eastern Washington University Eastern Washington University
45 PUBLICATIONS   2,211 CITATIONS    1 PUBLICATION   152 CITATIONS   

SEE PROFILE SEE PROFILE

Russell L Kolts
Eastern Washington University
34 PUBLICATIONS   2,170 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Youth Gratitude Project View project

Gender differences in gratitude and indebtedness. View project

All content following this page was uploaded by Philip Watkins on 20 March 2014.

The user has requested enhancement of the downloaded file.

COGNITION AND EMOTION
2006, 20 (2), 217±241

The debt of gratitude:

Dissociating gratitude and indebtedness
Philip C. Watkins, Jason Scheer, Melinda Ovnicek, and
Russell Kolts
Eastern Washington University, Cheney, WA, USA

We investigated the relationship between the emotional states of gratitude and

indebtedness in two studies. Although many have suggested that these affects are
essentially equivalent, we submit that they are distinct emotional states. Following
Heider (1958), we propose that with increasing expectations of return commu-
nicated with a gift by a benefactor, indebtedness should increase but gratitude
should decrease. The results of two vignette studies supported this hypothesis, and
patterns of thought/action tendencies showed these states to be distinct. In addition,
we found that with increasing expectations communicated by a benefactor, ben-
eficiaries reported that they would be less likely to help the benefactor in the
future. Taken together, we argue that the debt of gratitude is internally generated,
and is not analogous to an economic form of indebtedness.

Gratitude is a duty which ought to be paid, but which none have a right to expect.
Jean Jacques Rousseau

Gratitude has been identified as ``one of the most neglected emotions and one of
the most underestimated of the virtues'' (Solomon, 2004, p. v). Although gra-
titude appears to be one of the most understudied emotions in psychological
science (Emmons, 2004), it is an emotion evident in all cultures and is a virtue
encouraged by all of the major religions (Emmons & Crumpler, 2000). In
addition, recent analyses and research have supported the idea that gratitude is
an important component of the good life (Emmons & McCullough, 2003, 2004;
McCullough, Emmons, & Tsang, 2002; McCullough, Kilpatrick, Emmons, &
Larson, 2001; Watkins, 2004; Watkins, Woodward, Stone, & Kolts, 2003).
Given the importance of this social emotion and its apparent neglect, there is a
need for research on the cognitive structure of grateful emotion.

Correspondence should be addressed to Philip Watkins, Eastern Washington University,

Psychology Department, 151 Martin Hall, Cheney, Washington, WA99004, USA; e-mail:
Philip.Watkins@mail.ewu.edu
This research was supported by a generous grant from the John Templeton Foundation, awarded
to the first author.

# 2006 Psychology Press Ltd

http://www.psypress.com/cogemotion DOI:10.1080/02699930500172291
218 WATKINS ET AL.

Although the benefits of gratitude have also been lauded in the popular press,
not all have been as enthusiastic about gratitude. Stalin for example, concluded
that: ``Gratitude is a sickness suffered by dogs''. In somewhat more diplomatic
prose, Hobbes also expressed his reservations about gratefully receiving sig-
nificant benefits (cited in Greenberg, 1980, p. 17).

To have received from one, to whom we think ourselves equal, greater benefits
than there is hope to requite, disposeth to counterfeit love; but really secret hatred;
and puts a man into the estate of desperate debtor. . . .For benefits oblige; and
obligation is thraldom; and unrequitable obligation, perceptual thraldom; which is
to ones equal, hateful.

In this passage, Hobbes appears to equate gratitude with indebtedness. If gra-

titude and indebtedness are essentially equivalent emotions, gratitude may not
be as pleasant an experience as is commonly assumed. Indeed, a number of
scholars in the social sciences have equated gratitude and indebtedness (e.g.,
Komter, 2004). In the studies reported here, we sought to study the relationship
between gratitude and indebtedness, specifically to investigate whether gratitude
and indebtedness are distinct emotional states.
The reduction of gratitude and indebtedness into a single construct may
be due to the influential work of Marcel Mauss (1925/2002), who argued
that in pre-capitalist societies the community was held together by recipro-
city. Thus, expressions of gratitude are viewed as simple exchanges for bene-
fits to restore social balance in obedience to the norm of reciprocity. In the
past, psychology appears to have followed his lead. For example, in his ear-
lier writings Greenberg treats gratitude and indebtedness as synonymous
(Greenberg, 1980). Similarly, in an important seminal study of gratitude, Tes-
ser and colleagues combined gratitude and indebtedness into a composite
dependent variable because they were significantly correlated (Tesser, Gate-
wood, & Driver, 1968). As a psychological construct, indebtedness has
been defined as ``a state of obligation to repay another'' in the context of
the receipt of a benefit from another (Greenberg, 1980, p. 4). Greenberg
goes on to argue that indebtedness is an emotional state of ``arousal and dis-
comfort'' (p. 4), and that when one is in this state one is alert to opportu-
nities to reduce this discomfort. Gratitude on the other hand, has been
defined as ``a feeling of thankful appreciation for favors received'' (Gur-
alnik, 1971, p. 327). Are gratitude and indebtedness essentially equivalent
emotional states?
Some data suggest that these two constructs may not be equivalent. For
example, Greenberg, Bar-Tal, Mowrey, and Steinberg (1982) found that 92% of
their research participants said that being ``indebted'' to others was an
unpleasant state, but a Gallup poll indicated that the vast majority of people feel
that gratitude is a happy state (Gallup, 1998). At least in the minds of most
people, gratitude and indebtedness are experienced very differently (see also
Gray, Emmons, & Morrison, 2001).
 THE DEBT OF GRATITUDE 219

Although psychology has largely neglected the study of gratitude (Emmons

& Crumpler, 2000; McCullough et al., 2001), gratitude may be an important
component of happiness (McCullough et al., 2001; Watkins, 2004; Watkins et
al., 2003). Heider (1958) is a notable exception to this neglect in that he felt that
gratitude was an important interpersonal emotion, and therefore should be
explicated. However, even Heider's treatment of gratitude is quite short, and
gratitude is discussed largely as the counterpart to revenge in the context of
retributive emotions. But Heider's theory of gratitude may be informative as to
how gratitude and indebtedness might differ. Heider argued that in situations
where the benefactor obliges the beneficiary to be grateful, the beneficiary
would actually experience less gratitude. This is because the beneficiary would
prefer to infer that their gratitude is self-motivated, rather than controlled by
external sources. Thus, when there is an increasing external expectation of return
associated with a benefit, Heider predicted a decreasing experience of gratitude.
Although Heider did not explicitly discuss indebtedness, from his theory one
might deduce a contrasting scenario for this emotional state. It seems plausible
that with increasing expectations from a benefactor, the beneficiary should feel
an increasing sense of indebtedness. Thus, Heider's approach to gratitude
appears to imply that gratitude and indebtedness will show different patterns in
response to increasing expectations of return communicated by the benefactor.
Distinct emotional states should have distinct action tendencies (Frijda, 1986,
1988). If gratitude is distinct from indebtedness it should show different action
tendencies. Recently, Fredrickson has applied her broaden and build theory of
positive emotions to gratitude (Fredrickson, 2004). In her theory, she argues that
gratitude should have broad and creative thought/action tendencies similar to
other positive emotions, whereas indebtedness should be associated with a more
narrowed tit-for-tat type of reciprocity (p. 160). We sought to evaluate whether
gratitude and indebtedness are associated with distinct action tendencies.
In two vignette studies we investigated patterns of gratitude and indebtedness
responses to favours received. In each study, participants read a vignette
describing the receipt of a benefit, and were asked how they would respond to
this benefit on a number of emotion scales, including gratitude and indebtedness.
In the stories we manipulated the benefactor's expectations of return associated
with the gift. We predicted that with increased benefactor expectation of return,
indebtedness would increase but gratitude would decrease. Participants also
completed scales assessing thought/action readiness. Here, we predicted that
gratitude and indebtedness would be associated with distinct thought/action
readiness states.

STUDY 1
In study 1 we investigated the relationship between gratitude and indebtedness
with a vignette that described a person helping the protagonist move. We created
a story that we felt would ensure responses of gratitude by describing a valuable
220 WATKINS ET AL.

favour that was not expected (McCullough et al., 2001). Our stories were
identical, with the exception of what kind of response was expected by the
benefactor. We created three levels of expectation: one where no return favour
was expected (no expectation condition); one where the benefactor expected
some return expression of gratitude (moderate expectation condition); and
finally a condition where the benefactor expected a similar return favour in
addition to an expression of gratitude for the move assistance (high expectation
condition).

Method
Participants
A total of 107 students were involved in this study and received partial course
credit for their participation. Participants were run anonymously in a group
format and were randomly assigned to experimental condition (no expectation: n
= 35, moderate expectation: n = 37, high expectation: n = 35).

Materials
Questionnaires. It is possible that our experimental manipulation could
affect mood state. In order to assess emotional state both before and after the
primary experimental task, we administered the Positive and Negative Affective
States scales (PANAS; Watson, Clark, & Tellegen, 1988). We instructed our
participants to respond with regard to their immediate affective state. We also
used two measures of trait gratitude to investigate relationships of dispositional
gratitude to emotional responses to the scenarios. The GQ-6 (McCullough,
Emmons, & Tsang, 2002), and the GRAT (Watkins et al., 2003), are developed
trait gratitude measures that appear to have good psychometric properties.
The following base vignette was administered to all participants:

You have met someone in one of your classes, have become fairly well acquainted,
and have now known them for three months. You have studied several times
together with your new friend, have had coffee on several occasions, and you have
enjoyed your conversations. You have found a new apartment which is a better
deal for you, and so you decide that you will move to the new apartment. You
decide that it's best to complete the move in one day so you rent a truck to move on
Saturday. During the week your friend asks you what you're doing on Saturday,
and you explain that you'll be spending the day moving. When Saturday rolls
around you rent the truck and drive to your old apartment to begin your move.
When you arrive at your apartment you see your new friend waiting for you, ready
to help you move. Your friend helps you for most of the day until all your boxes
are transported to your new apartment. The move is completed by 2 pm, when you
originally thought that you would finish in the evening.
 THE DEBT OF GRATITUDE 221

We accomplished our manipulation by adding several sentences to the end of the

base vignette. In the no expectation condition the vignette concluded with: ``As
you consider your friend's help, you know them well enough that you feel your
friend does not expect any kind of return favour. Your friend's help was offered
simply as a gift without any future expectations on you''. The sentence: ``As
you consider your friend's help, you remember that others have told you that
when this friend helps anyone, he or she expects a clear expression of thanks,
usually in person and in the form of a card or written note'', was added to the
vignette for the moderate expectation condition. In the high expectation con-
dition the following explanation was added to the core scenario: ``As you
consider your friend's help, you remember that others have told you that when
this friend helps anyone, he or she expects a clear expression of thanks, usually
in person and in the form of a card or written note, and they also expect a return
favour. You happen to know that your friend is moving next Saturday''.

Procedure

After completing an immediate state PANAS, participants were randomly

assigned to read one of the three scenarios. Before reading the vignette, we
instructed our participants to: ``Please try to put yourself in the situation below
and imagine how you would respond''. After reading the vignette we asked our
participants to respond to various emotion scales on what amounted to a 5-point
Likert-type scale. Subjects were probed about seven different emotional states:
gratitude, indebtedness, gladness, resentment, guilt, irritation/annoyance, and
pride. Half of the subjects responded to the gratitude probe first and the
indebtedness probe third, and the remaining subjects did the converse. Subjects
were first asked to determine if they in fact would experience the emotion
(responding in a yes/no format), and if they indicated ``yes'' they were to rate
the extent of their emotional response ranging from 1 (``mild'') to 4
(``extreme''). Because we felt that the term indebtedness would not be familiar
to at least some of our subjects, we included a brief definition (following
Greenberg, 1980) in parentheses after the term (``feeling obligated to repay'').
After estimating their emotional responses to the scenario, subjects answered
a number of questions related to thought/action tendencies (cf. Frijda, 1986). As
with the emotions, we first asked our subjects to indicate whether they would
have any ``urge to do or think'' in the way indicated, and if so, we asked them to
indicate the extent of their inclination on a 5-point Likert-type scale (ranging
from ``slight urge'' to ``very strong urge''). These items generally followed
Frijda's approach (1986), and are listed in Table 1. This was followed by a
question regarding the participants' inclination to help the benefactor if they saw
them in need (on a 5-point Likert scale). Subjects then recorded their age and
gender and the gender of the benefactor that they imagined in the vignette. They
222 WATKINS ET AL.

TABLE 1
Study 1: Thought/Action readiness items

Thought/Action readiness item Thought/Action

readiness factor

I would feel like approaching my friend Approach

I would feel like avoiding my friend Active Against
I would feel like being with my friend Approach
I would notice my friend more and be more interested in them Adoration
I would feel like rejecting my friend Active Against
I would feel like ignoring my friend Active Against
I would feel like acting against my friend ±
I would feel like trying to regain control over my friend Active Against
I would feel like stopping whatever I was doing, and pay attention Yielding
only to my friend
I would feel like yielding to my friend Yielding
I would feel like maintaining control over my friend Active Against
I would feel like insulting my friend to their face Active Against
I would feel like insulting or deriding my friend to others when my Passive Against
friend was not present
I would feel like fighting with my friend ±
I would feel like running from my friend ±
I would feel like helping my friend Adoration
I would feel like expressing my happiness to my friend Adoration
I would feel like giving my friend a gift Adoration
I would feel like doing something for my friend Adoration
I would feel like complimenting my friend Adoration
I would feel like praising my friend to others when my friend was Adoration
not present
I would feel like thinking positive thoughts or happy memories about Adoration
my friend
I would feel like thinking negative thoughts or unpleasant memories Passive Negative
about my friend

Note: Items listed with no factor title showed no variance and thus were eliminated from the
analysis.

then completed a second administration of the PANAS. Finally, all participants

completed the GRAT and the GQ-6.

Results and discussion

Our primary analysis was conducted with a 3 (benefactor expectation condition)
6 2 (emotion: gratitude and indebtedness) mixed GLM. Emotion was analysed
as a repeated measure. The dependent variable was extent of emotional
response. Our primary predictions were supported; we found an interaction
between level of benefactor expectation and emotion type, such that as
 THE DEBT OF GRATITUDE 223

Figure 1. Gratitude and indebtedness by benefactor expectation condition

benefactor expectations increased, gratitude decreased but indebtedness

increased, F(2, 104) = 4.05, p < .05. Figure 1 illustrates this result. Also, note the
strong main effect of emotion type indicating that regardless of expectation
condition, subjects responded with more gratitude than indebtedness, F(1, 104) =
119.91, p < .05. No main effect was found for the expectation condition,
F(2, 104) = 0.14, ns.
Correlational analyses of emotional responses to the scenarios (see Table 2)
showed that gratitude was strongly associated with gladness (r = .46) and
moderately with pride (r = .19), but negatively correlated with resentment (r =
7.24) and annoyance (r = 7.29). Gratitude and indebtedness were not found to
be reliably associated (r = .16, p = .097). Only guilt was significantly correlated
with indebtedness (r = .25, p < .05). This corroborates past research showing that
gratitude tends to covary with positive affect while indebtedness covaries with
some negative affects.
We reduced our 24 thought/action tendency items through factor analysis to 6
factors and analysed responses by condition. Four thought/action tendency items
224 WATKINS ET AL.

TABLE 2
Study 1: Correlations of emotional states

Emotion 1. 2. 3. 4. 5. 6. 7.

1. Gratitude ± .16 .46*** 7.24* 7.00 7.29** .19*

2. Indebtedness ± .13 .03 .25* .08 .09
3. Gladness ± 7.07 7.11 7.31** .24*
4. Resentment ± .24* .37*** .11
5. Guilt ± .20* 7.04
6. Irritation/ Annoyance ± 7.08
7. Pride ±

* p < .05; ** p < .01; *** p < .001.

showed no variance and thus were eliminated from the analysis. The factors are
listed with the relevant items in Table 1. Three factors were significantly
affected by condition: Adoration, F(2, 104) = 8.47, p < .05; Approach, F(2, 104)
= 10.45, p < .05; and Yielding, F(2, 104) = 8.91, p < .05Ðall decreasing with
increased expectation of return (see Figure 2). Expectation condition did not
reveal a significant impact on the remaining factors (the pattern of the means
with each of the antisocial factors showed essentially no differences between
low and moderate obligation conditions, and increased action tendencies for the
high obligation condition). Thus, as return expectations from the benefactor
increased, prosocial action tendencies decreased. We also correlated gratitude
and indebtedness responses to the scenarios with our 6 action tendency factors.
Gratitude responses were positively and reliably associated with prosocial
thought/action tendencies (Adoration, Approach, Yielding), and negatively
associated with antisocial action tendencies (Active Against, Passive Negative,
and Passive Against). Table 3 contains the relevant Pearson r correlation
coefficients for both gratitude and indebtedness. Indebtedness was not reliably
correlated with any of the action tendency factors. As thought/action tendency is
an important attribute of distinct emotions (Frijda, 1986, 1988), the pattern of
correlations supports the hypothesis that gratitude and indebtedness are different
emotional states.
To evaluate Fredrickson's (2004) broaden and build theory with respect to
gratitude, we created a variable that reflects the total number of endorsed pro-
social action tendencies, and correlated this with gratitude responses. If a subject
endorsed at least a ``1'' on any prosocial action tendency item, this was con-
sidered to indicate some inclination on this item, and the total number of pro-
social items endorsed with one or more was computed for each subject.
Gratitude was positively associated with total number of prosocial action ten-
dencies endorsed (r = .44, p < .05), but indebtedness was not (r = .08, ns). This
finding supports Fredrickson's theory that gratitude is associated with a broader
array of responses to a benefit than is indebtedness. We also created a
 Figure 2. Thought/Action tendency by benefactor expectation condition

TABLE 3
Study 1: Thought/Action tendency correlations with gratitude and indebtedness

Thought/Action tendency factor

Emotion Adoration Approach Yielding Active Passive Passive

Against Negative Against

Gratitude .57 .40 .30 7.21 7.37 7.27

Indebtedness .16 .04 .10 .17 .16 .13

Note: All correlations with gratitude were statistically significant, but none of the above
correlations with indebtedness reached significance.

225
226 WATKINS ET AL.

complimentary variable tapping total number of antisocial action tendencies

endorsed. Gratitude was inversely associated with the total number of antisocial
action tendency items endorsed (r = 7.25, p < .05), but indebtedness was
positively associated with the antisocial total (r = .20, p < .05).
Is trait gratitude predictive of subjects' responses to the scenarios? We
conducted correlational analyses with our trait gratitude measures and emotional
response variables to investigate this question. With the emotional responses to
the scenario, the GRAT was significantly associated with gratitude (r = .20), and
gladness (r = .34), and negatively associated with guilt (r = 7.31), and
annoyance (r = 7.27). Generally speaking, correlations involving the GQ-6
showed the same pattern of relationships, but the correlation between the GQ-6
and gratitude response did not reach significance (r = .14). This pattern of results
suggests that grateful individuals are more likely to respond positively to a gift
or favour. In addition, we found that the GRAT and GQ-6 correlated highly (r =
.77), supporting the concurrent validity of these instruments.
We also found interesting results with respect to affect change over the
course of the experiment. We created affect change scores by subtracting pre-
treatment from posttreatment PANAS scores. As subjects reported more grati-
tude in response to the vignettes, positive affect increased as measured by the
PANAS (r = .25, p < .05), and negative affect decreased (r = 7.17, p < .05).
Indebtedness was not reliably associated with affect change (positive affect: r =
7.10, negative affect: r = 7.01). Thus, merely imagining grateful responses to
a benefit is related to improvements in mood.
Finally, we found that expectation condition significantly affected a subject's
reported inclination to help the benefactor in the future, F(2, 104) = 3.86, p < .05.
As expectation from the benefactor increased, inclination to help decreased.
To summarise the important findings of Study 1, we found that when a favour
is given, with increasing expectations of return from the benefactor the bene-
ficiary feels less gratitude but more indebtedness. Second, we found that indi-
viduals were more likely to respond with gratitude than indebtedness to a favour.
Also, we found that gratitude was positively associated with positive emotional
states and negatively correlated with unpleasant states, but indebtedness was
only reliably associated with guilt. Fourth, we found that the thought/action
tendency patterns of gratitude were distinct from those of indebtedness.

STUDY 2
In Study 1 we used a costly gift from the benefactor in order to ensure gratitude
responses. However, observation of means indicated that our results may have
suffered from a ceiling effect. In an attempt to rectify this problem, in Study 2
we elected to use a scenario with a benefit that was less costly to the benefactor,
and somewhat less valuable to the beneficiary. We hoped that this benefit would
produce less gratitude (Tesser et al., 1968), and so be less vulnerable to a ceiling
 THE DEBT OF GRATITUDE 227

effect. In this study our vignette described a benefactor photocopying notes for
the beneficiary for a day they missed class. This less costly favour also provides
for a situation that is likely to be more common for our respondents. We hoped
that these scenarios would provide us with information relevant to more
everyday gratitude responses. We also felt that our 5-point response scales may
have been somewhat limited in scale, and thus we expanded our response scales
to 7 alternatives. In Study 1, we clearly indicated in our vignette that a positive
relationship existed between the benefactor and the beneficiary before the favour
took place. However, it is possible that a different pattern of results might ensue
if the benefactor was initially disliked. In fact, Heider (1958) suggests that
favours from disliked benefactors will result in less gratitude on the part of the
beneficiary. Thus, in Study 2 we not only manipulated sense of expectation as in
Study 1, but we also attempted to manipulate the valence of the benefactor.

Method
Overview and design
In this study we used a 3 (benefactor expectation condition) 6 2 (valence of
benefactor: liked or disliked) 6 2 (emotional state) mixed design. Expectation
condition and benefactor valence were between-subjects factors, and emotional
state was a repeated measure. Thus, participants were randomly assigned to read
one of six different vignettes, and were asked how they felt they would respond
emotionally to the situation.

Participants
As in Study 1, participants were run in a group format, and their participation
was anonymous. A total of 152 respondents were involved in this study, and
received extra credit for their participation.

Materials
Questionnaires. As before, we administered the PANAS both before and
after our experimental treatment, and participants took the GRAT and the GQ-6
as well.
The following base vignette was administered to all participants:

You have met someone in one of your classes, have become fairly well acquainted,
and have now known them for three months. You have had class projects with this
person and like (dislike) them. You are unable to make it to class on a day when
very important information was being discussed. The following week this same
classmate comes to class with a photocopy of their notes from the class you
missed. They explain that they had noticed you were missing from class and so
took notes for you.
228 WATKINS ET AL.

We attempted to manipulate the valence of the benefactor with the ``like'' and
``dislike'' statements as shown above. As before, we manipulated obligation by
adding a sentence to the end of the vignette. For the no expectation condition we
added: ``Based on what others have told you, you realise that your classmate's
help was offered with no expectation of return''. For the moderate expectation
condition participants read the following sentence at the conclusion of the
vignette: ``As you consider your classmate's help, based on what others have
told you, you realise that when your classmate helps anyone they expect a clear
expression of thanks, usually in person and in the form of a card or written
note''. Finally, in the high expectation condition we added the sentences: ``As
you consider your classmates help, based on what others have told you, you
realise that when this classmate helps anyone he or she expects a clear
expression of thanks, usually in person and in the form of a card or written note
and they also expect a return favour. Your classmate has also mentioned to you
that they have to miss class next week''.

Procedure
The procedure was essentially identical to Study 1 with the following excep-
tions. After reading the vignette participants were asked to indicate how they
would feel in response to the situation described in the vignette. They were
asked to respond to 11 different emotional states: ``grateful'', ``happy'',
``indebted'', ``contended'', ``guilty'', ``wonder'', ``fortunate'', ``flustered'',
``obligated'', ``uneasy'', and ``thankful''. Participants responded to these
emotion queries on a 7-point scale. The emotion probes were presented in the
above order, with the exception that half of the participants responded to the
indebtedness query first and gratitude third. Also, we defined indebtedness
slightly differently than in Study 1. In this case the definition ``feeling like you
owe'' was included in parentheses after the word ``indebted''. Following their
estimations of their emotional responses, participants responded to various
thought/action tendency questions as in Study 1. Although our list of thought/
action tendency items was very similar to Study 1, in this study the items we
used were from Frijda's research (Frijda, Kuipers, & ter Schure, 1989), as well
as several items we felt to be relevant to gratitude (Table 4 contains a list of
these items). We also utilized a 7-point Likert type scale (0 = ``no urge at all''
and 6 = ``very strong urge'').
As in Study 1, the thought/action queries were followed with a question
regarding the participant's inclination to help the benefactor in the future, and a
question about the gender of the imagined benefactor. We then asked partici-
pants to indicate their like/dislike of the benefactor on a 9-point scale ranging
from ``I disliked them very much'' to ``I liked them very much''. This item
would serve as a manipulation check for our benefactor valence manipulation.
We then asked: ``How important do you think it is to express gratitude to
 THE DEBT OF GRATITUDE 229
TABLE 4
Study 2: Thought/Action readiness items

Thought/Action readiness item Thought/Action readiness factor

I would feel like approaching or making contact with this person Positive Response/Approach
I would want to have nothing to do with this person or would Rejecting/Avoiding
want to stay away from this person
I would feel like staying close or being receptive to this person Positive Response/Approach
I would want to pay more attention to this person Positive Response/Approach
I would feel like rejecting this person Rejecting/Avoiding
I would feel like ignoring my friend Rejecting/Avoiding
I would feel like opposing or insulting this person Rejecting/Avoiding
I would feel like trying to regain control over this person Inhibition
I would feel like stopping whatever I was doing, and pay Positive Response/Approach
attention only to this person
I would feel like yielding (or not opposing) the wishes of this Positive Response/Approach
person
I would feel like I had command of this situation Rejecting/Avoiding
I would boil inside Rejecting/Avoiding
I would feel like helping Positive Response/Approach
I would feel like keeping this person at a distance Rejecting/Avoiding
I would feel like giving this person a gift Positive Response/Approach
I would feel like doing something for this person Positive Response/Approach
I would feel like complimenting this person Positive Response/Approach
I would feel like praising this person to others when my friend Positive Response/Approach
was not present
I would feel like thinking positive thoughts or recalling happy Positive Response/Approach
memories about my friend
I would feel like thinking negative thoughts or unpleasant Rejecting/Avoiding
memories about my friend
I would wish that they hadn't given me their notes in the first Rejecting/Avoiding
place
I would not be able to concentrate or order my thoughts Inhibition
I would feel like disappearing, so I wouldn't be noticed by Apathy
anyone
I would feel inhibited, paralyzed, or frozen Inhibition
I would blush or be afraid of blushing Inhibition
I would not feel like doing anything, I would feel apathetic Apathy
I would feel like quitting or giving up Inhibition
I would feel like shutting myself off from my surroundings Inhibition
I would want to do something, but would not know what to do Inhibition
I would cry or feel like crying Inhibition
I would feel excited and restless, I could not sit still Inhibition
I would feel exuberant, would feel like singing, jumping, or Play/Leisure
undertaking things
I would laugh or feel like laughing Play/Leisure
I would feel at rest, that everything was OK, that there was no Play/Leisure
need to do anything

229
230 WATKINS ET AL.

others?'' (9-point Likert scale), followed by queries about their gender and age.
Participants then completed another immediate state PANAS, followed by the
GRAT and GQ-6. Finally, we asked our participants two religiosity questions
not relevant to the primary purpose of this study. After completing the packet,
students were awarded class credit and thanked for their participation.

Results and discussion

Manipulation check
We first conducted a manipulation check on likeability of the benefactor with
a 2 (like and dislike) 6 3 (benefactor expectation condition) ANOVA on the
likeability of the benefactor. We found a large main effect for our likeability
manipulation, F(1, 145) = 54.68, p < .0001, indicating that our manipulation was
successful. Expectation condition did not show a significant main effect on
likeability, F(2, 145) = 2.14, p = .12, nor did it interact with like/dislike of the
benefactor, F(2, 145) = 0.36, ns.

Primary analyses
For our primary analysis we conducted a 3 (benefactor expectation) 6 2
(like/dislike of benefactor) 6 2 (type of emotion: gratitude vs. indebtedness)
mixed GLM on extent of emotion. Type of emotion was a repeated measure. As
in Study 1, a large main effect was found for emotion, F(1, 146) = 77.05,
p < .0001, indicating that regardless of expectation condition, participants
responded with more gratitude than indebtedness (see Figure 3). We also found a
main effect for likeability of the benefactor, F(1, 146) = 8.11, p < .004. This
effect resulted from the fact that if subjects liked their benefactor, they reported
more gratitude and more indebtedness. Likeability of the benefactor did not
interact with any other variables (all Fs < 1). Supporting our primary predic-
tions, and replicating the results of Study 1, we found a significant two-way
interaction between benefactor expectation and type of emotion, F(2, 146) =
8.95, p < .0001. Again, as expectation of return from the benefactor increased,
gratitude decreased but indebtedness increased (see Figure 3). The three-way
interaction between benefactor expectation, benefactor likeability, and emotion
type was not significant, F(2, 146) = 0.19, ns. Thus, as in Study 1, we were able
to show that gratitude and indebtedness show opposite patterns in response to
increasing expectations of return from a benefactor. This offers further support
for the conclusion that gratitude and indebtedness are indeed distinct emotional
states.
As noted earlier, we gave our subjects a slightly different definition of
indebtedness in this study (``feeling that you owe''). This definition is different
from that used in Study 1, and different from the original definition provided by
Greenberg (1980, ``feeling obligated to repay''). One could argue that our
 THE DEBT OF GRATITUDE 231

Figure 3. Gratitude, indebtedness, and obligation by benefactor expectation condition

``obligation'' emotional response might correspond more closely to the form of

indebtedness that we measured in Study 1. Thus, we also conducted a 3
(benefactor expectation) 6 2 (like/dislike of benefactor) 6 2 (type of emotion:
gratitude vs. obligation) mixed ANOVA on extent of emotion, with type of
emotion as a repeated measure. Again, we found a large main effect for type of
emotion, F(1, 146) = 74.72, p < .0001, indicating that regardless of expectation
condition, respondents reported more gratitude than obligation. In contrast to our
previous finding, we found no main effect due to likeability of the benefactor,
F(1, 146) = 0.72, ns. Likeability of the benefactor did not show any other main
effects. However, a marginal interaction was found between likeability of the
benefactor and emotion type, F(1, 146) = 3.81, p = .053. When participants liked
the benefactor they showed more gratitude than when they disliked their giver,
but they showed slightly less obligation when their benefactor was liked than
when disliked. As with our previous analysis, the two-way interaction between
benefactor expectation and emotion type was significant, F(2, 146) = 15.31,
p < .0001. Although this interaction revealed a similar pattern as our initial
analysis, obligation showed a slightly different pattern from indebtedness. While
232 WATKINS ET AL.

indebtedness showed a clear linear increase with benefactor expectation, obli-

gation showed no increase from the no expectation to moderate expectation
condition, but a large increase in the high expectation condition (see Figure 3).
As before, the three-way interaction was not significant, F(2, 146) = 0.27, ns.

Correlations of emotional responses to the scenarios

Table 5 shows the correlations between the emotions we measured in

response to the various scenarios. As in Study 1, gratitude tended to correlate
positively with positive affective states. In this study, gratitude did correlate
significantly with indebtedness (r = .25). Although the correlation between
gratitude and indebtedness did not reach significance in Study 1, the correlations
were not significantly different (.16 vs. .25; z = .74, ns). This statistically
nonsignificant difference may reflect our different definitions of indebtedness,
but whatever the case, gratitude and indebtedness shared only 6.2% of variance
in this study, far from what one would expect if these were essentially equivalent
emotional states. Indebtedness was most strongly correlated with obligation (r =
.64), showing the similarity of these feeling states. Although indebtedness was
correlated with some positive affective states, it was also significantly correlated
with several negative states (guilt and flustered), whereas gratitude was not. In
this study, indebtedness appeared to be a somewhat mixed affective state, but
obligation was not. The feeling of being obligated was primarily correlated with
negative affective states. These results are largely consistent with those of Study
1, and support the conclusion that gratitude and indebtedness are distinct
affective states.

TABLE 5
Study 2: Correlations among emotional states

Emotion 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

1. Grateful ± .25 .12 .73 .61 .49 .12 .04 .58 7.09 7.18
2. Indebted ± .64 .20 .27 .23 .25 .09 .19 .26 .14
3. Obligated ± .13 .08 .10 .30 .22 .09 .37 .37
4. Thankful ± .60 .53 .08 .11 .65 7.12 7.16
5. Happy ± .69 .11 .14 .60 7.23 7.27
6. Contented ± .03 .06 .52 7.22 7.32
7. Guilty ± .41 .16 .54 .56
8. Wonder ± .14 .28 .37
9. Fortunate ± 7.10 7.12
10. Flustered ± .67
11. Uneasy ±

Note: All correlations > |.15|, p < .05, correlations > |.20|, p < .01, correlations > |.27|, p < .001.
 THE DEBT OF GRATITUDE 233

Analyses of Thought/Action tendencies

As in Study 1, we reduced our 34 action tendency items through factor

analysis. Our initial factor analysis indicated that five factors had eigenvalues
close to or above two, so we retained five factors. We titled these factors
Rejecting/Avoiding, Positive Response/Approach, Inhibited, Apathy, and Play/
Leisure. Action tendency items and the factor they most strongly loaded on are
found in Table 4.
We first conducted separate ANOVAs to evaluate the impact of our inde-
pendent variables on each action tendency factor. Benefactor expectation
showed a significant main effect on the Positive Response/Approach factor,
F(2, 144) = 15.62, p < .0001, such that as benefactor expectation increased,
positive response and approach to the benefactor decreased. Not surprisingly,
likeability of the benefactor also impacted this factor, F(1, 144) = 12.77,
p < .0001, with subjects showing more positive responses and approach ten-
dencies if they liked the benefactor. However, no significant interaction was
found (F < 1).
When analysing the effects of our independent variables on the Rejection/
Avoidance action tendency factor, complementary results were found. Again,
we found a main effect of benefactor expectation on rejection and avoidance,
F(2, 145) = 9.32, p < .0001. The more the benefactor expected in return, the
more rejection and avoidance action tendencies were revealed. Likeability of the
benefactor also significantly affected this factor, F(1, 145) = 7.53, p < .01. When
the benefactor was disliked subjects were more likely to show increased
rejection and avoidance action tendencies. With this factor the benefactor
expectation 6 benefactor likeability interaction approached significance,
F(2, 145) = 2.43, p = .09. This trend appeared to be due to the fact that with
increased benefactor expectation, Rejection/Avoidance action tendency
increased much more if the benefactor was disliked.
Benefactor expectation was also found to have a significant effect on our
Play/Leisure action tendency factor, F(2, 146) = 4.16, p < .05. In this case, as
benefactor expectation of return increased, exuberance, laughter, and rest ten-
dencies decreased. No other significant effects were found with this factor.
Finally, no significant effects of our manipulations were found with the Inhib-
ited or the Apathy action tendency factors.
We also conducted correlation analyses between the emotional responses to
the vignettes and the reported action tendencies. We first correlated gratitude,
indebtedness, and obligation with our five action tendency factors (see Table 6).
Observation of these correlations also supports the distinctiveness of gratitude
and indebtedness. Gratitude is strongly correlated with prosocial tendencies, but
appears to be associated with inhibited antisocial tendencies (indicated by the
significant negative correlation with the Rejection/Avoidance factor). Although
indebtedness shows some positive correlation with prosocial thought/action
234 WATKINS ET AL.

TABLE 6
Study 2: Correlations between emotional responses and action tendency factors

Action tendency factor

Emotion Positive Response/ Rejection/ Inhibition Apathy Play/Leisure

Approach Avoidance

Gratitude .44*** 7.42*** 7.06 .01 .14

Indebtedness .29*** 7.07 .06 .08 7.07
Obligation .07 .19* .28** .01 7.11

* p < .05; ** p < .01; *** p < .001.

tendencies, it does not appear to inhibit antisocial tendencies. Feelings of

obligation, on the other hand, are not associated with prosocial action tenden-
cies, and in fact are positively correlated with antisocial tendencies (as indicated
by the significant positive correlation with the Rejection/Avoidance factor).
Interestingly, obligation was positively correlated with the Inhibition factor.
This suggests that when people feel obligated to return a favour, they tend to feel
upset and are somewhat confused about how to respond to the situation.
To evaluate Fredrickson's (2004) ideas regarding the diversity of prosocial
action tendencies related to gratitude, we created a variable that indicated the
total number of prosocial action tendencies endorsed as in Study 1. Consistent
with Fredrickson's suggestions, gratitude was more strongly associated with
the total number of endorsed prosocial action tendencies (r = .38, p < .05) than
was indebtedness (r = .30, p < .05), although this difference was not statisti-
cally significant (t = .87, ns). Additionally, Gratitude was significantly more
strongly related to prosocial tendencies than was obligation (r = .11, ns; t =
2.89, p < .01). Recall that in this experiment ``feeling obligated'' is probably
closer to our original conception of indebtedness in Study 1. These results
support the theory that gratitude is associated with a more broad array of pro-
social responses than is indebtedness. Total number of antisocial action ten-
dencies endorsed was inversely associated with gratitude (r = 7.36, p < .001),
not correlated with indebtedness (r = .00), and positively related to obligation
(r = .25, p < .05).
The general picture one gleans from the action tendency results is that gra-
titude tends to be associated with broad prosocial thought/action tendencies, and
appears to inhibit antisocial tendencies. Action tendencies associated with
indebtedness and obligation on the other hand, appear to be more narrow and
mixed. Although some prosocial action tendencies are associated with indebt-
edness/obligation, they tend to be much weaker than with gratitude, and obli-
gation is actually associated with an increased likelihood of experiencing some
antisocial tendencies.
 THE DEBT OF GRATITUDE 235

Altruism analyses
We conducted a 3 (benefactor expectation) 6 2 (likeability of the benefactor)
analysis on one's reported inclination to help. Recall that we actually had two
items that tapped this tendency, one's reported likelihood of help if their
benefactor was in need, and the help item in the action tendency list (``I would
feel like helping''). No significant effects were found with the help when the
benefactor was in need item, although means were in the expected direction. It
appeared that results were limited in this analysis because of a ceiling effect
(i.e., virtually all subjects reported that they would be very likely to help their
benefactor if they were in need). However, results did reach significance with
the help action tendency item. Here we found a main effect for benefactor
expectation condition, F(2, 146) = 7.46, p < .001, with inclination to help
decreasing as benefactor expectation increased. We also found a main effect for
likeability of the benefactor, F(1, 146) = 4.91, p < .05, with subjects reporting
more inclination to help if they liked their benefactor. The interaction between
benefactor expectation and likeability of benefactor was not significant,
F(2, 146) = 0.35, ns. Thus, as in Study 1, we are left with an interesting paradox
of altruism; as expectations of return favors from a benefactor increase, the
beneficiary is less inclined to help.
We also conducted correlational analyses between emotional responses to the
scenarios, and inclination to help. Gratitude was significantly correlated with
both help variables (r = .34 and .30), as was indebtedness (r = .26 and .20).
Obligation on the other hand, was only weakly associated with the inclination to
help the benefactor in need (r = .18), and was not associated with the action
tendency to help (r = .04).

Emotion change analyses

Using the PANAS we again evaluated the impact of our manipulation on
emotion change. Thus, we conducted a 3 (benefactor expectation) 6 2 (bene-
factor likeability) 6 2 (time: pretest and posttest) mixed GLM with time as our
repeated measure. We found a main effect for time, F(1, 142) = 17.54, p < .001.
In general, subjects reported less positive emotion after the treatments than
before. However, this main effect was modified by a time 6 benefactor
expectation interaction, F(2, 142) = 3.31, p < .05. This interaction was due to the
fact that in the condition with no benefactor expectation, positive affect
remained constant. However, in the moderate and high expectation conditions
positive affect decreased from pre to posttest. No other interactions reached
significance.
The only significant effect involving negative affect was with the time 6
benefactor expectation interaction, F(2, 142) = 3.05, p = .05. Negative affect
decreased most in the no expectation condition (71.24), decreased somewhat
less in the moderate expectation condition (70.77), but actually increased in the
236 WATKINS ET AL.

high expectation condition (0.73). As in Study 1, our vignette manipulation

actually impacted our subject's mood.
To summarise the findings of Study 2, once again we found that as the return
expectations of the benefactor increased, the gratitude of the beneficiary
decreased, but indebtedness and feelings of obligation increased. Thus, we again
were able to dissociate gratitude and indebtedness. Although likeability of the
benefactor did not interact with benefactor expectation, if subjects liked their
benefactor they reported more gratitude and more indebtedness. Also, gratitude
and indebtedness appeared to show distinct action tendencies. Taken together,
these results support the proposition that gratitude and indebtedness should be
seen as distinct emotional states.

GENERAL DISCUSSION
In both studies, we showed that gratitude and indebtedness can be dissociated. In
this sense, gratitude and indebtedness may best be viewed as distinct emotional
states. Supporting Heider's predictions, we found that with increasing expec-
tations from a benefactor, indebtedness increases but gratitude decreases. Cor-
relational analyses within emotional responses and with our thought/action
tendency factors supported the theory that gratitude should be viewed as a
positive affect but indebtedness has mixed affective associations. Finally, we
found that gratitude responses were more strongly associated with inclination for
future altruism than indebtedness and feelings of obligation. In their review of
gratitude as a moral affect, McCullough et al. (2001) argue that gratitude
functions as a moral barometer, a moral motivator, and a moral reinforcer. The
moral motivator hypothesis submits that grateful affect motivates individuals to
prosocial behaviour. Although a few studies offer support for this hypothesis,
direct evidence supporting this notion is scant. Our results appear to provide
direct support for the moral motivator hypothesis.
Taken together, the results of this study leave us with a curious paradox of
giving and gratitude. If gifts are given for the purpose of receiving return favours
from the beneficiary, the beneficiary is less likely to feel grateful, and is less
likely to feel like returning the favour. The more a benefit is received as a gift of
grace, the more likely there will be a return of gratitude.
We believe our results have notable implications for reciprocity theory
(Gouldner, 1960). First, it is interesting that the more a benefactor made the
norm of reciprocity salient, the less grateful the beneficiary felt, and they
apparently felt less motivated to fulfil the requirements of the norm of reci-
procity. Our results seem to suggest that when exchange aspects of a relationship
are emphasised by the benefactor, the beneficiary is less likely to feel grateful.
Perhaps one important distinction between indebtedness and gratitude is that
indebtedness is an emotion of exchange, whereas gratitude is not. Although
there is often some kind of reciprocity that is often motivated by gratitude, the
 THE DEBT OF GRATITUDE 237

evidence presented here suggests that prosocial returns of gratitude are not
viewed by the actor in terms of exchange. Stated differently, when returning a
favor motivated by gratitude, the actor does not consider their behaviour as some
kind of exchange for past favours.
In this matter it is interesting to consider our results in the context of the
distinction between exchange and communal relationships. Clark and Mills
(1979) defined communal relationships as those in which the partners give to
each other primarily as a response to perceived need, while in exchange rela-
tionships people give in response to perceived benefits from their partner. It is
possible that our participants assumed a communal relationship with their
benefactor, and when the benefactor communicated relationship rules of
exchange, this change in relationship caused the observed changes in gratitude
and indebtedness. However, communal relationships as described by Clark and
Mills (1979), tend to be intimate attachment relationships that do not fit with the
``acquaintances'' we described in our vignettes. Nonetheless, the communal/
exchange relationship distinction is likely to be of importance when considering
gratitude and indebtedness responses to benefits. Future research may wish to
manipulate relationship status in this manner.
Another alternative explanation to our results might be that our participants
were surprised with the return expectations of the benefactor, and this is what
caused the decrement in gratitude. If, when receiving a benefit, beneficiaries
assume that benefactors do not expect a return favour, then when the benefactor
communicates expectations of return, this results in surprise and hence results in
less gratitude as well. On the other hand, the norm of reciprocity might lead us to
predict that individuals usually expect the benefactor to desire a return favour.
Of course, it is an empirical question as to whether beneficiaries typically
believe their benefactor expects a return favour or not, and individual differ-
ences are likely to play an important role here. Whatever the case, these
considerations provide interesting questions for future research.
An obvious limitation of these studies is that we used vignettes rather than
actual benefits given to our subjects. It is possible that our findings reflect
subject's lay theories of gratitude, not their actual experience of response to
benefits. In this way, future studies that use actual benefits would extend our
findings. However, there are significant advantages to vignette studies. In
vignette studies such as these, one can carefully control the perceived expec-
tations of the benefactor. With the actual conferral of a benefit, it would seem
much more difficult to control the perceived expectations of the benefactor.
Also, it is difficult to actually provide subjects with meaningful benefits in
experimental settings. In general, vignette studies offer the experimenter control
over conditions that cannot be easily achieved with other laboratory manip-
ulations. Also, it should be emphasised that in both experiments our vignette
manipulations impacted subject's mood. In general, as expectations from the
benefactor increased in the scenarios, mood deteriorated. This suggests that our
238 WATKINS ET AL.

subjects actually did try to imagine themselves in the scenario, and the fact that
the scenarios impacted mood suggests that the emotional responses subjects
reported to the vignettes are reflective of how they would have responded in
actual life settings.
Another limitation of this study is common in most psychology experi-
ments; our population was culturally limited. Would these results extend to
individuals of other cultures? This would be an interesting question for future
research. It is possible that in other cultural settings, gratitude and indebtedness
are seen as much more synonymous than they are in the United States. For
example, a number of researchers have noted that emotions of dependence and
consequently indebtedness are more acceptable in eastern cultures such as
Japan. Although there are likely to be important cultural differences in how
individuals respond to benefits, a recent study suggested that indebtedness is
still an undesirable state to individuals in eastern cultures. Naito, Wangwan,
and Tani (2005) investigated feelings evoked by a favour and found that for
both Japanese and Thai students, indebtedness loaded strongly on their nega-
tive affect factor (along with ``shame'', ``regret'', and ``uneasiness'').
``Thankfulness'' loaded strongly on the positive affect factor. Although grati-
tude appears to be quite cross-cultural, expressions of gratitude vary con-
siderably between cultures, and cross-cultural investigations of gratitude and
indebtedness would be informative.
In this study we were able to show that benefactor expectations impact
gratitude and indebtedness in very different ways. Are there other variables
that might dissociate gratitude and indebtedness? One candidate might be
the locus of cause of the benefit. Some studies have shown that a bene-
ficiary experiences more indebtedness when they ask for a benefit than
when a benefactor provides unprompted help (e.g., Greenberg & Saxe,
1975). Would gratitude responses reveal a different pattern? Gratitude and
indebtedness might also show different patterns of response following recipro-
city behaviour. Suppose a benefit is given to a beneficiary. If gratitude and
indebtedness are distinct affects as we propose, when the beneficiary pro-
vides a return favor one might expect indebtedness to dissipate but grati-
tude to remain. Future investigations that demonstrate dissociations between
gratitude and indebtedness would provide significant contributions to our
understanding of gratitude.
In these studies we found that gratitude is more clearly an enjoyable state
than is indebtedness, and that gratitude was more predictive of an inclination for
future altruism than was indebtedness. These findings have interesting impli-
cations for altruism that is motivated by gratitude vs. helping behaviour more
motivated by indebtedness. If gratitude is indeed a more enjoyable state than
indebtedness, might favours motivated by gratitude be more enjoyable than
those motivated by indebtedness? If this is true, one might also propose that acts
of ``thanks giving'' are more likely to continue than acts of indebtedness. These
 THE DEBT OF GRATITUDE 239

suggestions are highly speculative at this point, but would be a fruitful avenue
for future research. If ``thanks giving'' is indeed more enjoyable and enduring
than altruism motivated by indebtedness, gratitude may well be an impetus for
the ``upward spirals'' or ``cycles of virtue'' that have been recently proposed
(Fredrickson, 2004; Watkins, 2004).
Our findings from Study 2 suggest that there may be different forms of
indebtedness. Indebtedness and feelings of obligation appeared to show some-
what different patterns in this study, and our finding does not appear to be
unique (Greenberg & Frisch, 1972). This points to the need to carefully define
indebtedness for subjects. For the sake of consistency, we recommend Green-
berg's (1980) definition that we used in Study 1 (``feeling obligated to repay'').
However, is it possible that one may feel as though they are in someone's debt,
but do not feel obligated to repay? It seems worthwhile to investigate this form
of indebtedness as well, although some may argue that this is the essence of
gratitude (e.g., Roberts, 2004).
Is there a debt of gratitude? Although we dissociated gratitude and indebt-
edness in these studies and these states are clearly experienced differently, in
Study 2 we found that gratitude and indebtedness were positively correlated.
Tesser and colleagues also found that gratitude and indebtedness were sig-
nificantly correlated (Tesser et al., 1968). Also, gratitude was correlated with a
desire to help their benefactor in the future. However, the debt of gratitude
does not appear to be analogous to an economic debt. An economic debt is one
where the debtor must recompense the lender, the debt is established by the
lender, and it is a state the debtor desires to be free of (cf. Berger, 1975; Card,
1988; Wellman, 1999). Thus, although gratitude appears to be associated with
a desire to recompense the benefactor, participants report enjoying this feeling,
and if the obligation is externally imposed the desire to return the favour
decreases. Roberts' approach to gratitude might shed some light on this
dilemma (1991a, 1991b, 2004). He defines gratitude as a glad acceptance of
our debt to one who has acted for our benefit. In this view, gratitude is
expressed as a token of appreciation for the benefit and for the beneficiary's
relationship with the benefactor. If this is the case, then the grateful person
may respond with grateful recompense, not necessarily to discharge the debt,
but rather to establish and encourage the interdependent relationship with his/
her benefactor. Perhaps Milton (1667/1999, p. 238) has best summarised this
paradox of gratitude:

A grateful mind
By owing owes not, but still pays, at once
Indebted and discharged.

Manuscript received 10 May 2004

Revised manuscript received 4 May 2005
240 WATKINS ET AL.

REFERENCES
Berger, F. (1975). Gratitude. Ethics, 85, 298±309.
Card, C. (1988). Gratitude and obligation. American Philosophical Quarterly, 25, 115±127.
Clark, M. S., & Mills, J. (1979). Interpersonal attraction in exchange and communal relationships.
Journal of Personality and Social Psychology, 37, 12±24.
Emmons, R. A. (2004). Introduction. In R. A. Emmons & M. E. McCullough (Eds.), The psychology
of gratitude (pp. 3±16). New York: Oxford University Press.
Emmons, R. A., & Crumpler, C. A. (2000). Gratitude as human strength: Appraising the evidence.
Journal of Social and Clinical Psychology, 19, 56±69.
Emmons, R. A., & McCullough, M. E. (2003). Counting blessings versus burdens: An empirical
investigation of gratitude and subjective well-being in daily life. Journal of Personality and
Social Psychology, 84, 377±389.
Emmons, R. A., & McCullough, M. E. (2004). The psychology of gratitude. New York: Oxford
University Press.
Fredrickson, B. L. (2004). Gratitude, like other positive emotions, broadens and builds. In R. A.
Emmons & M. E. McCullough (Eds.), The psychology of gratitude (pp. 145±166). New York:
Oxford University Press.
Frijda, N. (1986). The emotions. Cambridge, UK: Cambridge University Press.
Frijda, N. (1988). The laws of emotion. American Psychologist, 43, 349±358.
Frijda, N. H., Kuipers, P., & ter Schure, E. (1989). Relations among emotion, appraisal, and emo-
tional action readiness. Journal of Personality and Social Psychology, 57, 212±228.
Gallup survey results on ``gratitude'', adults and teenagers (1998). Emerging Trends, 20, 4±5, 9.
Gouldner, A. (1960). The Norm of reciprocity: A preliminary statement. American Sociological
Review, 25, 161±178.
Gray, S. A., Emmons, R. A., & Morrison, A. (2001, August). Distinguishing gratitude from
indebtedness in affect and action tendencies. Poster presented to the Annual Convention of the
American Psychological Association, San Francisco, CA.
Greenberg, M. S. (1980). A theory of indebtedness. In K. J. Gergen, M. S. Greenberg, & R. H. Wills
(Eds.), Social exchange: Advances in theory and research. New York: Plenum.
Greenberg, M. S., Bar-Tal, D., Mowrey, J., & Steinberg, R. (1982). Cross-cultural comparisons in
the perception of indebtedness. Unpublished manuscript, University of Pittsburgh.
Greenberg, M. S., & Frisch, D. M. (1972). Effect of intentionality on willingness to reciprocate a
favor. Journal of Experimental Social Psychology, 8, 99±111.
Greenberg, M. S., & Saxe, L. (1975). Importance of the locus of help initiation and type of outcome
as determinants of reactions to another's help attempt. Social Behavior and Personality, 3,
101±110.
Guralnik, D. B. (1971). Webster's new world dictionary. Nashville, TN: Southwestern.
Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.
Komter, A. (2004). Gratitude and gift exchange. In R. A. Emmons & M. E. McCullough (Eds.), The
psychology of gratitude (pp. 195±212). Oxford, NY/UK: Oxford University Press.
Mauss, M. (2002). The gift: The form and reason for exchange in archaic societies. New York:
Routledge. (Original work published 1925)
McCullough, M. E., Emmons, R. A., & Tsang, J. (2002). The grateful disposition: A conceptual and
empirical topography. Journal of Personality and Social Psychology, 82, 112±127.
McCullough, M. E., Kilpatrick, S. D., Emmons, R. A., & Larson, D. B. (2001). Is gratitude a moral
affect? Psychological Bulletin, 127, 249±266.
Milton, J. (1999). Paradise Lost. In B. Raffel (Ed.), The annotated Milton: Complete English poems
(pp. 133±536). New York: Bantam. (Original work published 1667)
Naito, T., Wangwan, J., & Tani, M. (2005). Gratitude in university students in Japan and Thailand.
Journal of Cross-Cultural Psychology, 36, 247±263.
 THE DEBT OF GRATITUDE 241
Roberts, R. C. (1991a). Mental health and the virtues of community: Christian reflections on con-
textual therapy. Journal of Psychology and Theology, 19, 319±333.
Roberts, R. C. (1991b). Virtues and rules. Philosophy and Phenomenological Research, 51, 325±343.
Roberts, R. C. (2004). The blessings of gratitude: A conceptual analysis. In R. A. Emmons & M. E.
McCullough (Eds.), The psychology of gratitude (pp. 58±78). New York: Oxford University
Press.
Solomon, R. C. (2004). Forward. In R A. Emmons & M. E. McCullough (Eds.), The psychology of
gratitude (pp. v±xi). New York: Oxford University Press.
Tesser, A., Gatewood, R., & Driver, M. (1968). Some determinants of gratitude. Journal of Per-
sonality and Social Psychology, 9, 233±236.
Watkins, P. C. (2004). Gratitude and subjective well-being. In R. A. Emmons & M. E. McCullough
(Eds.), The Psychology of Gratitude (pp. 167±192). New York: Oxford University Press.
Watkins, P. C., Woodward, K., Stone, T., & Kolts, R. D. (2003). Gratitude and happiness: The
development of a measure of gratitude and its relationship with subjective well-being. Social
Behavior and Personality, 31, 431±452.
Watson, D., Clark, L.A., & Tellegen, A. (1988). Development and validation of brief measures of
positive and negative affect: The PANAS Scales. Journal of Personality and Social Psychology,
54, 1063±1070.
Wellman, C. H. (1999). Gratitude as virtue. Pacific Philosophical Quarterly, 80, 284±300.

View publication stats