52 Current Psychology / Spring 2004
Current Psychology: Developmental • Learning • Personality • Social
Spring 2004, Vol. 23, No. 1,
Counting Your Blessings: Positive
Memories Among Grateful Persons
PHILIP C. WATKINS, DEAN L. GRIMM, AND RUSSELL KOLTS
Eastern Washington University
We conducted two studies investigating the relationship of gratitude to autobiographi-
cal memory of positive and negative life events. Gratitude was assessed with an
attitudinal measure and college students were asked to recall both positive and nega-
tive events from their past. In both studies, a significant positive relationship was
found between trait gratitude and a positive memory bias. In Study 2 it was found that
gratitude still reliably predicted positive memory bias after controlling for depression.
Further, it was found that a positive intrusive memory bias was associated with grati-
tude in both studies. Thus, an important component of gratitude may be an enhanced
tendency to recall positive events from one’s life.
To “Count your blessings” has long been considered an effective way of increasing
one’s happiness. Are some individuals more likely to “count their blessings” than
others? More specifically, are grateful people more likely to reflect back on pleasant
memories? The purpose of this project was to examine the association of gratitude
with positive memory bias of life events. Recently there have been notable calls for a
more “positive psychology” (e.g., Seligman, 1998). Despite these calls and new re-
search activity, emotion research is still more generally oriented toward unpleasant
emotions and their disorders (Fredrickson, 1998). As Seligman (1998) has noted,
“Sadly, while plumbing the depths of what is worst in life, psychology lost its connection
to the positive side of life—the knowledge about what makes human life most worth
living, most fulfilling, most enjoyable and most productive” (p. 4). In this context, it is
clear that positive emotional states also deserve study, not merely because of the
imbalance in the literature, but because positive states are an essential aspect of a
fulfilled life and have been shown to promote adaptive behavior (Fredrickson, 1998).
Recently, more research has been directed at subjective well-being or happiness (Di-
ener, Suh, Lucas, & Smith, 1999). Most of this research has investigated factors that
might contribute to happiness. Although demographic variables do not seem to be
strongly related to happiness, a few personality traits (e.g., extraversion and neuroti-
cism) have consistently been found to be associated with subjective well-being (for a
review, see Myers & Diener, 1995). Thus, further research into correlates of happiness
appears to be needed.
Frijda’s emotional law of habituation states that with time we adapt or habituate to
our happiness or satisfaction (Frijda, 1988). Thus, even though one’s circumstances
53Watkins, Grimm, and Kolts
may be favorable, this law suggests that individuals habituate to their circumstances
and may not feel the same degree of life satisfaction as they had previously experi-
enced. Hence, an important goal of happiness research is to determine how one’s
subjective well-being may be prolonged without constantly needing an upgrade in
one’s circumstances (Diener et al., 1999). Frijda suggested that “Adaptation to satis-
faction can be counteracted by constantly being aware of how fortunate one’s condi-
tion is” (Frijda, 1988, p. 354). In this statement, we believe Frijda has essentially
defined an important component of trait gratitude. We have defined gratitude as the
attitude of appreciating life as a gift and recognizing the importance of expressing that
appreciation (Watkins, Porter, & Curtis, 1996; Watkins, Woodward, Stone, & Kolts, in
press). It would seem that trait gratitude would counteract the aforementioned adapta-
tion process, enabling the grateful person to maintain her or his level of happiness in
the face of unchanging, or even challenging, life circumstances. Thus, we propose that
gratitude should have a strong association with subjective well-being. Past research
has supported this proposition, showing that gratitude is as strongly related to happi-
ness as other personality variables that have been consistently shown to be related to
well-being (McCullough, Emmons, & Tsang, 2002; Watkins, Porter, & Miller, 1997;
Watkins et al., in press; for a review, see Watkins, in press). Thus, gratitude research
may provide important clues into what constitutes a “positive psychology.”
What characterizes a grateful person? We have argued that three aspects character-
ize dispositional gratitude: an appreciation for others, an appreciation for simple plea-
sures, and a sense of abundance (Watkins et al., in press). Factor analysis of the
Gratitude, Resentment, and Appreciation Test (GRAT), a trait gratitude measure with
good psychometric properties, has supported this three-factor structure of gratitude
(Watkins et al., in press). A sense of abundance (or, stated negatively, a lack of a sense
of deprivation), is important to pervasive gratitude because if one feels deprived in
life, or if one feels they are entitled to more benefits than they have received, he or she
is not likely to feel grateful. An important aspect regarding a sense of abundance is the
ability to recall those benefits that one has experienced in the past. If positive events
come easily to mind, then one is likely to judge his or her life as abundant, and
consequently should be more likely to report a pervasive sense of gratitude. Thus, we
predicted that an essential quality of gratitude itself would be the tendency to recall
and relish positive life events. From another perspective, Watkins (in press) has argued
that grateful affect should promote the encoding and retrieval of positive memories.
Research has shown that grateful emotion involves the perception of the value of the
gift and the goodness of the benefactor (for reviews see McCullough, Kilpatrick,
Emmons, & Larson, 2001; Watkins, 2001). Thus, if one feels grateful in response to a
benefit, this additional cognitive elaboration should enhance encoding of the event in
memory, and should also improve its retrievability. Second, grateful affect should
promote reflecting back on memories of positive benevolence. Memory research has
conclusively shown that retrieving a representation has great mnemonic advantages
(Payne, 1987; Roediger & Challis, 1989). Hence, if gratitude promotes reflecting back
on pleasant experiences, this should enhance their retrievability. Third, if one is feeling
grateful at retrieval, this affect should promote the recollection of positive events via
54 Current Psychology / Spring 2004
mood-congruent memory (Bower, 1981; for a review see Blaney, 1986). Affective
network theory (Bower, 1981) states that emotional memories are organized by emo-
tion nodes, and when a node is activated this increases the activation of relevant
emotional memories. In several studies gratitude has been clearly shown to be a
positive affect (e.g., Watkins, Scheer, & Coletti, 2002; Watkins et al., in press), hence
if one feels grateful this should activate positive emotion nodes, thus increasing the
accessibility of positive life event memories. While intuitively appealing, this mood-
congruent retrieval mechanism is likely the weakest of our suggestions presented here.
Although there is some evidence that retrieval mood can bias retrieval, a number of
studies have failed to find this effect and it appears that mood-congruent memory is
largely due to mood-congruent encoding processes (Blaney, 1986). Dispositional grati-
tude implies that grateful individuals should have a lower threshold for grateful affect
(McCullough, Emmons, & Tsang, 2002). In other words, individuals high in trait
gratitude should be more likely to feel grateful in response to benefits. In several
studies we have shown that indeed, GRAT scores predict grateful affective responses
(Watkins et al., 2002; Watkins et al., in press, Study 4). Thus, grateful individuals
should be more likely to respond to positive events with gratitude, to reflect back on
life benefits, and to be feeling grateful when recollecting past events. All of these
aspects should promote enhanced recall of positive life events.
In two studies, we sought to investigate whether gratitude is associated with a
positive life event recall bias. In both studies, a measure of gratitude was administered
(the GRAT), followed by an autobiographical recall task. We used a listing method
(e.g., Seidlitz & Deiner, 1993), in which subjects listed positive and negative events
from their life for three minutes each. We predicted that gratitude would be positively
associated with a positive autobiographical memory bias.
Sixty-six undergraduate students completed this study, receiving partial course credit
for their participation. Student participation was in a group format and was anony-
Questionnaires. The GRAT (1996, see also Watkins et al., in press) was used to
assess trait gratitude. This appears to be a reliable instrument (coefficient alpha = .92,
Spearman-Brown = .89, two week test-retest rs = .79 and .90). Research has also
supported the validity of the GRAT (Watkins et al., in press). The GRAT was shown
to be positively related to satisfaction with life (rs between = .49 and .68 in five popu-
lations) and positive affectivity (r = .36). Also as predicted, the GRAT was inversely
associated with depression as measured by the Beck Depression Inventory (r =-.34
55Watkins, Grimm, and Kolts
to-.72), aggression/hostility (r =-.30), and narcissism (r =-.49). Subsequent research
has also shown that clinically depressed individuals and non-depressed individuals
with a history of depression have reliably lower GRAT scores than controls (Wood-
ward, Moua, & Watkins, 1998).
Life events recall sheets. Participants recalled life events by listing memories on
prepared recall sheets following the approach of Seidlitz and Diener (1993). Each
participant was provided with a recall sheet for positive events and a sheet for negative
life events. Instructions for the recall of pleasant events directed participants to “briefly
list as many positive events as you can from the past 3 years of your life.” The recall
sheet included a table with a number in one column (numbering 1 through 32), “Event”
heading the second column, and finally a column titled “Rating.” Participants were
instructed that each number (each row) represented a “different life event.” Under the
event column participants were instructed to write a brief sentence for each life event
they recalled. At the bottom of the recall sheets we included 40 boxes that participants
could check if an intrusive memory happened to come to mind during the recall task.
An intrusive memory was defined as a memory opposite that of the valence that they
were instructed to remember. For example, on the positive recall sheet participants
were told “If, while trying to think of positive events in your life, you actually remem-
ber some negative events, check one of the boxes provided at the bottom of the page
for each negative event that you think of.” We used the simple procedure of checking
boxes for recording intrusive memories because we felt that having participants actu-
ally write down their intrusive memories would provide too much interference to the
primary task (intentional recall of valenced memories), and this could activate more
intrusive memories. Instructions for the negative event recall sheets were identical to
the positive recall sheets, with the exception of the valence of the events to be remem-
We also used Seidlitz and Diener’s (1993) life events checklist, which contains
various life events that participants could endorse. The events on this questionnaire are
divided up into positive and negative objective events and positive and negative sub-
jective events. Objective events are those which were salient, unlikely to be forgotten,
and could in principle be verified by an outside observer (e.g., “Death of parent or
sibling”). Subjective events, on the other hand, required considerable interpretation by
the participant (e.g., “Haven’t received enough support and understanding from my
All participants were tested in an anonymous group format. Participants received a
packet with the questionnaires and the life event recall sheets. Initially, participants
completed the GRAT. After all participants had completed the GRAT, they were
introduced to the life event recall sheets. Instructions were read out loud by a research
assistant, and participants were then given three minutes to complete each recall sheet
(positive and negative). Order of valence of life events to be recalled was randomly
assigned. After completion of the recall tasks, participants were instructed to go back
56 Current Psychology / Spring 2004
to their recorded memories and rate the “favorableness” of each event they recalled
using a 7-point Likert type scale. This scale was anchored as follows: 0 = “Very
Unfavorable Event,” 1 = “Unfavorable Event,” 2 = “Slightly Unfavorable Event,” 3 =
“Neither Favorable or Unfavorable,” 4 = “Slightly Favorable Event,” 5 = “Favorable
Event”, 6 = “Very Favorable Event.” After participants completed the rating task they
once again completed the GRAT. Finally participants completed the life events check-
list. Participants were then debriefed and were thanked for their participation.
Because it is possible that order of valenced recall could influence the results, we
analyzed the number of positive and negative memories recalled by order. We con-
ducted this analysis for both intentional and intrusive memories. Thus we conducted a
2 (Order) x 2 (Valence) x 2 (Memory Type) repeated measures ANOVA. In this mixed
design Valence and Memory Type were repeated measures. No main effect for Order
was found, F (1, 64) = 0.65, ns. There was a large main effect for Memory Type, F (1,
64) = 235.01, p<.05. As might be expected, this effect was due to all participants
retrieving many more intentional than intrusive memories. There was also a large main
effect for Valence, F(1, 64) = 65.81, p<.05, because participants tended to recall more
positive than negative memories. However, this main effect was modified by a Memory
Type x Valence interaction, F(1, 64) = 20.37, p<.05. As shown in Table 1, this
interaction reflected that positive memories were more likely to be recalled intention-
ally, but intrusive memories were not as biased toward the positive. Although we
found a Valence x Order interaction (F [1, 64] = 29.67, p<.05), we also found a three
way interaction between Valence, Order, and Memory Type, F (1, 64) = 4.61 p<.05.
To evaluate this interaction, we conducted a two way Order x Valence ANOVA for
each memory type.
Recall Means by Order, Valence, and Type of Memory
Recall Type Valence Positive/Negative Negative/Positive Totals
Positive 10.16 (3.96) 11.62 (4.46) 10.91 (4.26)
Negative 7.50 (3.55) 5.94 (3.20) 6.70 (3.44)
Positive 2.53 (2.23) 6.09 (3.85) 4.36 (3.62)
Negative 3.19 (2.33) 1.59 (1.89) 2.36 (2.25)
Note. SDs are in parentheses. Note that when positive recall instructions were given first, negative
intrusions would have occurred before positive intrusions in the recall task (and vice versa).
57Watkins, Grimm, and Kolts
Although the Valence x Order interaction was reliable for both memory types
(intentional memories: F [1, 64] = 10.23, p<.05; intrusive memories: F[1, 64] = 36.44,
p<.05), it is apparent that the three way interaction described earlier was due to a
stronger trend for intrusive memories. The general trend regarding Order appears to be
a practice effect in that participants did better in the second recall task than the first.
Regardless of Valence, it is apparent that participants generally reported more inten-
tional memories and less intrusive memories in the second recall trial (see Table 1).
Thus, although participants generally exhibited a bias towards positive events, this bias
was stronger when the positive events were retrieved in the second recall trial. The
exception to the tendency for all subjects to exhibit a positive recall bias is that if
positive events were retrieved first, a positive bias was not observed with intrusive
memories. Although we find these results interesting, because order was randomly
assigned it does not represent a confound. To confirm this we divided the subjects into
two groups using a median split with the GRAT and conducted a four way ANOVA
with Group (grateful and ungrateful participants) Order, Valence, and Memory Type.
In this analysis no interactions involving Group and Order were significant.
We also felt that order of recall could have an impact on participant’s scores on the
GRAT. Because participants took the GRAT both before and after the recall task, we
could evaluate the impact of order of recall as well as the impact of the recall task
itself on GRAT scores. Although GRAT scores were found to be very stable for the
two administrations (r=.94), we felt that it was important to evaluate the possibility
that the valence of events recalled immediately before the second administration of the
GRAT could impact report of gratitude. Thus we conducted a 2 (Order) x 2 (Time: pre
and post recall task) ANOVA for GRAT scores with Time as a repeated measure.
Neither the main effects of Order or Time were reliable, but a significant Order x Time
interaction emerged, F (1, 64) = 4.41, p<.05. As seen in Table 2, when positive events
were recalled first followed by negative events, pre and post GRAT scores were
virtually identical. However, when positive events were retrieved after negative events,
GRAT scores appeared to increase. Thus, when an individual recalled positive events
immediately before retaking the GRAT this appeared to increase reports of pervasive
GRAT Scores by Order of Recall
GRAT Administration Positive/Negative Negative/Positive
Pre Recall GRAT 174.75 (3.76) 179.35 (3.65)
Post Recall GRAT 174.06 (4.07) 182.76 (3.95)
Note. Standard errors are in parentheses.
58 Current Psychology / Spring 2004
Association between Memory and Gratitude
The primary purpose of this study was to evaluate the relationship between the
disposition of gratitude as measured by the GRAT and life event recall bias. We first
formed a recall bias score by subtracting the number of negative events recalled from
the number of positive events reported for each participant. This calculation is neces-
sary in order to remove general memory individual difference effects (recall of posi-
tive and negative events were positively correlated, r=.45 , p<.05). We found that the
pre GRAT was reliably related to both intentional recall bias (r=.26, p<.05), and
intrusive memory bias (r=.32, p<.05). Post recall administration of the GRAT was also
significantly associated with intentional memory bias (r=.31, p<.05) and intrusive
memory bias (r=.38, p<.05). Interestingly, mean event favorability rating was not
related to pre recall GRAT (r=.13, ns) or post recall GRAT (r=.06, ns) scores. Thus,
although grateful individuals tended to recall more positive memories, they did not
generally tend to rate their life events more positively. To investigate this further, we
divided the participants into two groups using a median split with the GRAT and
conducted a 2 (Group: grateful and ungrateful) x 2 (Valence) x 2 (Order) mixed
ANOVA for mean favorability ratings. In this analysis Valence was the repeated
measure. Here we found a two way interaction between Group and Valence, F (1, 61)
= 7.93, p<.05. Observation of the group means showed that grateful individuals tended
to rate positive events as more favorable (5.37 [sd=0.44] vs. 4.87 [sd=1.04]). How-
ever, grateful participants tended to rate negative memories as less favorable (0.93
[sd=0.59] vs. 1.44 [sd=1.25]) than did less grateful individuals. We explore this find-
ing further in Study 2.
We also conducted correlational analyses with objective and subjective event memo-
ries from the life events checklist. As might be expected, number of positive events
listed during our recall task was positively associated with number of positive objec-
tive life events (POLE) endorsed from the life events checklist (r=.39, p<.05). Simi-
larly, number of negative events recalled was reliably related to the number of nega-
tive objective life events (NOLE) endorsed (r=.26, p<.05). Thus, the actual frequency
of positive and negative events experienced likely contributed to one’s recall “bias.”
However, this relationship was far from veridical. Interestingly, the number of positive
subjective life events (PSLE) from the checklist was not reliably associated with
positive events recall (r=.22, p=.07), but the relationship between negative subjective
checklist events endorsed (NSLE) and number of negative events recalled (r=.32,
p<.05) was stronger than the relationship between NOLEs and negative recall. Number
of positive intrusive memories recalled was significantly correlated to both POLEs
(r=.30, p<.05), and PSLEs (r=.24, p<.05). However, negative intrusions were not
significantly related to either NOLEs (r=.06) or NSLEs (r=.19) events from the check-
list. Number of objective and subjective life events endorsed were also found to be
related to gratitude. Although number of PSLEs endorsed correlated significantly with
pre GRAT scores (r=.39, p<.05), number of POLEs endorsed did not (r=.14, ns). On
the other hand, GRAT scores were reliably associated with both NOLEs (r=-.28,
p<.05) and NSLEs (r=-.27, p<.05).
59Watkins, Grimm, and Kolts
As predicted, in Study 1 we found that trait gratitude was positively associated with
a positive memory bias. Gratitude was associated with both intentional and intrusive
positive memory biases. These results suggest that pleasant events come to mind more
easily for grateful than for less grateful individuals. This may be one way that grati-
tude promotes subjective well-being; grateful individuals are more likely to recall
pleasant life events, and hence are more likely to judge their lives as satisfactory.
Although grateful individuals tended to show a positive memory bias, gratitude was
not associated with favorability ratings of the events recalled. However, the favorability
rating that we used did not clearly identify the time perspective of the event. In other
words, we cannot definitively infer whether this rating reflected our participants’ origi-
nal experience of the event, or how they felt about the event at the time of recollection.
This is an issue that we attempted to clarify in Study 2.
We also found that gratitude was inversely associated with the number of negative
objective (events not likely to be mistakenly recalled) and negative subjective life
events endorsed from the life events checklist. However, in terms of positive events,
only subjective positive life events were correlated with gratitude. Taken together, this
suggests that one’s subjective impression of life events contributes to one’s sense of
gratitude. Alternatively, it may be that a pervasive sense of gratitude contributes to
positive interpretations of life events. This latter interpretive approach supports more
of an encoding bias theory of gratitude, i.e., that grateful individuals tend to encode
life events in a positive fashion. Findings from the life events checklist support the
idea that while actual negative life events contribute to one’s sense of gratitude, fre-
quency of objective positive events do not. It seems to be more important that indi-
viduals interpret life events as positive in order to feel grateful. These findings are
quite preliminary, and researchers may want to investigate further how objective life
events and one’s subjective interpretation of these events actually contribute to trait
gratitude. Results from the life events checklist raise another explanation of our pri-
mary finding. Our positive memory bias/gratitude association could have been due to
the fact that grateful people have actually had more positive events happen to them.
However, the subjective events were more strongly correlated with gratitude than the
objective events. In fact, positive objective events were not correlated with gratitude.
Taken together, this suggests that although grateful people actually have less unpleas-
ant lives in an objective sense, they do not seem to have a more positive objective past.
These results suggest that grateful individuals tend to encode life events in a more
positive fashion (Seidlitz & Diener, 1993). It is also important to note that negative
intrusive memories were not reliably correlated with the events checklist. Thus, while
one may argue that our intentional memory bias was simply a reflection of a positive
past, this argument is less tenable with intrusive memory bias. Whatever the case, it is
clear that positive events are more likely to come to mind for grateful than for less
One final caveat to these results should be noted. As discussed earlier, in past
studies we have found that gratitude is negatively correlated with depression (for a
60 Current Psychology / Spring 2004
review, see Watkins & Ola, 2001). It is also known that depression is associated with a
negative memory bias in various types of memory tests (for a review of explicit
memory bias see Blaney, 1986; for a review of implicit memory bias see Watkins,
2002). Thus, it is possible that the observed memory bias/gratitude association was
largely an artifact of a depression memory bias. This is another concern we attempted
to address in Study 2.
In Study 1 we found that gratitude was associated with a positive memory bias.
This supported the idea that positive events come easily to mind for grateful individu-
als. If recalling pleasant events has a positive emotional impact on grateful people,
they should be more likely to judge their lives as satisfactory. However, in Study 1 we
did not specifically investigate the emotional impact of the memories that were re-
called. In Study 2 we asked our participants to report the emotional impact of the
recalled event at the time it occurred (“then” ratings), as well as the impact recalling
the event has on them now (“now” ratings). We predicted that gratitude would not be
related to “then” emotional impact ratings, but would be positively related to “now”
emotional impact ratings. In other words, we predicted that recalling past events has a
more positive impact on current emotional state for grateful than for less grateful
individuals. Similarly, we also predicted that negative memories would show more
emotional improvement from “then” to “now” ratings for grateful than for less grateful
individuals. Second, in Study 2 we wanted to replicate our findings from Study 1
while controlling for the possible impact of depression. We predicted that gratitude
would predict positive memory bias independently from depression.
One hundred and twenty-two students participated in this study, and received their
choice of class credit or financial remuneration ($10, U.S.). Different from the group
format of Study 1, in this study students completed the experimental protocol individu-
As in Study 1, we used the GRAT to measure trait gratitude. The recall sheets used
in this study were also similar to Study 1, with the exception that instead of providing
spaces for indicating favorability ratings for the recalled events, we provided two
columns for “then” and “now” emotional impact ratings (described below). Our mea-
sure of depression was the Beck Depression Inventory (BDI; Beck, 1978), one of the
most frequently used and most well-validated self-report measures of depression. We
61Watkins, Grimm, and Kolts
also assessed depression history with the Past Accounts of Sadness Test (PAST; Watkins
& Curtis, 1994).
After signing the informed consent, participants completed the BDI, the PAST, and
the GRAT, followed by the life events recall task. Order of recall of valenced events
was randomly assigned. After recalling positive and negative events for three minutes
each, participants rated their memories for emotional impact. For each memory re-
corded, participants recorded the emotional impact of the event when it originally
occurred, and the emotional impact produced by the current recollection of the event.
Emotional impact ratings were assessed with a 9-point Likert type scale, where 1=very
negative effect, 2=strong negative effect, 3=moderate negative effect, 4=slight nega-
tive effect, 5=no effect at all, 6=slight positive effect, 7=moderate positive effect,
8=strong positive effect, and 9=very positive effect. Participants were then debriefed
and thanked for their participation.
RESULTS AND DISCUSSION
Results from our order analysis replicated our findings from Study 1, and we will
not take the space to reiterate these findings. In autobiographical recall tests such as
these, it appears that participants get better with practice, i.e., they are able to recall
more desired events and experience less intrusive memories in the second recall trial.
Memory Bias Analyses
As in Study 1 we created positive recall bias scores and correlated them with GRAT
scores. Again we found that gratitude was correlated with positive intentional memory
bias (r=.39, p<.05) and positive intrusive memory bias (r=.22, p<.05). Thus, the events
that grateful individuals recall from their past appear to be more pleasant than the
memories of less grateful individuals. However, this relationship could be an artifact
of the negative memory bias that has been well documented in depression (Blaney,
1986), because the GRAT has been shown to have reliable inverse associations with
depression (Watkins & Ola, 2001). Indeed, in this study the BDI was significantly
correlated with the GRAT (r=-.51, p<.05). Thus we conducted a partial correlation
between the GRAT and positive memory bias, while holding BDI scores constant.
This analysis revealed that the GRAT predicted positive intentional memory bias
independently from depression (r=.21, p<.05). Thus, grateful individuals tend to recall
more positive events from their past, and this is not simply because grateful individu-
als tend to be less depressed than their less grateful counterparts.
A second purpose of this study was to investigate how gratitude affects the emo-
62 Current Psychology / Spring 2004
tional impact of positive and negative memories over time. After recalling emotional
events our participants examined their recollections and rated the emotional impact of
each memory for how it affected them at the time the event occurred (“then” ratings),
and how recalling the event affects them now (“now” ratings). First, it is important to
note that GRAT scores did not correlate significantly with “then” ratings (positive
events: r=.14; negative events: r=-.10). Thus, it appears that when asked to recall
positive and negative life events, grateful and less grateful individuals recall very
similar events (at least according to their own retrospective evaluation). Our primary
question, however, was whether grateful individuals would experience these memories
more positively at the time of recollection. We felt that the clearest and most informa-
tive way to investigate this question was to use an ANOVA approach. However,
parallel correlational analyses indicated essentially identical conclusions to the ANOVA
results we report below. We divided our participants into two gratitude groups using a
Emotional Impact Ratings by Gratitude Group, Memory Valence, and Time Perspective
63Watkins, Grimm, and Kolts
median split of the GRAT scores of this population (median=175.50). We conducted a
2 (Gratitude Group) x 2 (Valence: positive and negative) x 2 (Time Perspective: then
and now) mixed ANOVA with Valence and Time Perspective as repeated measures.
As expected, we found a large effect for Valence, F(1,120)=1500.198, p<.05, indicat-
ing that positive memories were rated much more positively than were negative memo-
ries. We also found a main effect for Time Perspective, F(1,120)=86.415, p<.05,
however this main effect was modified by a significant Valence x Time Perspective
interaction, F(1,120)=214.294, p<.05. As seen in Figure 1, this interaction was due to
negative memories showing more change over time than positive memories, with
negative memories improving over time, and positive memories showing some de-
crease in positive impact. These emotional impact findings appear to be consistent
with similar studies from the autobiographical memory literature; both positive and
negative memories appear to lose their intensity with time, but negative memories tend
to lose more of their bite than positive memories lose of their joy (e.g., Holmes, 1970;
Walker, Vogl, & Thompson, 1997).
The three-way interaction between Gratitude Group, Valence, and Time perspective
was not significant, F(1,120)=0.31, ns. However, supporting our predictions, the two-
way interaction between Gratitude Group and Time Perspective was reliable,
F(1,120)=9.20, p<.05). As shown by Figure 1, this interaction reflected that grateful
individuals showed more improvement in terms of emotional impact of negative memo-
ries, but showed less decrement with time with positive memories. Stated differently,
although the “then” emotional impact ratings of positive and negative events did not
differ for grateful and less grateful participants, ratings of current emotional impact for
both positive and negative memories were more positive for grateful than for less
Are grateful individuals more likely to retrieve blessings from their past? The
results of the two studies reported here suggest that indeed, they are. In both studies,
we found that trait gratitude was associated with a positive memory bias. We found
this relationship in intentional memories, as well as in intrusive memories. Not only do
grateful individuals tend to recall more positive memories when instructed to do so,
they also tend to have more positive memories come to mind even when they are
attempting to remember negative life events. Thus, one component of gratitude ap-
pears to be the ability to recall more positive relative to negative events from one’s
past. Our second major finding was that the memories of grateful individuals had more
positive emotional impact on them than did the recollections of their less grateful
Several interesting implications emerge from these findings. First, consider the
finding that gratitude is associated with a positive memory bias. Recalling positive
events is likely to enhance one’s judgment of their satisfaction with life (Schwartz &
Clore, 1983; Strack, Schwartz, & Gschneidinger, 1985), and this is one way gratitude
might promote subjective well-being (Watkins, in press). Some experimental work
64 Current Psychology / Spring 2004
supports this formulation. In three studies, Emmons and McCullough (2003) found
that a simple practice of recollecting things one was grateful for produced a wide
variety of emotional benefits, including life satisfaction and optimism. Second, the
ability to bring more pleasant events to mind might assist one in coping with stressful
or challenging events. For example, if one receives a poor test grade, the ability to
remember more positive experiences with past tests will probably enhance one’s abil-
ity to cope with the disappointing result, and give one more optimism for dealing with
Implications from our emotional impact findings also suggest ways in which grati-
tude might enhance subjective well-being. Generally, for more grateful individuals, we
found that positive events had more positive impact and that recalling negative events
produced less unpleasant emotion. Because research has shown that recollections that
enhance mood also enhance satisfaction with life judgments (e.g., Strack et al., 1985),
this implies that grateful individuals should receive more emotional benefits from
remembering benefits from their past. We found the results from emotional impact
ratings of negative events particularly interesting. Grateful individuals showed more
emotional recovery from “then” to “now” ratings than did their less grateful counter-
parts. This suggests that gratitude may assist one in processing unpleasant emotional
memories (Rachman, 1980). Although the direction of causation cannot be determined
from this study, our results provide support for the notion that gratitude may help one
reframe negative emotional memories in such a way as to decrease their aversive
impact on the individual. In another study (Masingale et al., 2001), grateful victims of
trauma reported significantly lower levels of post-traumatic symptoms than moder-
ately, or low grateful victims of trauma. We propose that gratitude might assist indi-
viduals in finding positive consequences from negative events in their lives.
Our emotional impact findings also provide interesting suggestions regarding per-
vasive gratitude and redemptive sequences in life stories. McAdams and colleagues
(McAdams, Reynolds, Lewis, Patten, & Bowman, 2001) have proposed that redemp-
tive sequences in one’s life story are important to a life narrative that supports satisfac-
tion with life. Briefly, a redemptive sequence is a bad life event that eventually turns
good. McAdams et al. (2001) found that redemptive sequences in one’s life story were
more predictive of happiness than the overall positivity of their life stories. Our results
regarding the emotional recovery of negative events suggest that grateful individuals
may have more of these redemptive sequences than those who are less grateful. In fact,
while only 36% of our less grateful participants had negative life events that improved
by 2 points or more on our emotional impact scale, 62% of our grateful participants
improved in this way. Similarly, while only 13% of our less grateful participants had
“now” emotional impact ratings of negative events of five or more (five being the
neutral point on the scale), 24% of our grateful participants had mean “now” ratings of
five or greater. It is possible that grateful people are more able to see how bad events
might turn out for the good. However, we were not specifically investigating this
hypothesis when the study was designed, thus future research should more directly test
the theory that grateful people are more likely to tell redemptive life stories. Another
interesting hypothesis would be to test whether one is more likely to feel grateful for
65Watkins, Grimm, and Kolts
redemptive stories than for those that are merely positive. In this context, narrative
research should provide results informative to the nature of gratitude.
As with most investigations, the design and results of our studies leave many
questions unanswered. For example, are grateful people more likely to count their
blessings, or do they simply have more blessings to count? Findings from the life
events check list that included positive and negative “objective” events (events not
likely to be misrecalled), indicated that grateful individuals actually have had better
lives. In fact, there is some evidence that expressing gratitude can produce more social
benefits for the grateful person (for a review, see McCullough et al., 2001). However,
the number of positive and negative subjective events endorsed from our checklist was
also correlated with gratitude. Thus, it is likely that both actual events, and one’s
interpretation of life events, contribute to a pervasive sense of gratitude (Seidlitz &
Further, our design and results leave open the question as to whether we were really
investigating a “memory bias.” In fact, objective events from our checklist did corre-
late with our recall bias measures, but the strongest association only accounted for
15% of the variance of the recall bias. Thus, the “memory bias” measure we used as
our primary memory variable was far from veridical to the actual occurrence of posi-
tive and negative life events. However, as in most autobiographical memory studies, in
no way did we control the encoding of positive and negative events. In order to
investigate a true memory bias, one would need to control the encoding of valenced
information, and future studies should attempt this in a quasi-experimental design. In
this regard, information-processing methodologies should be helpful to the investiga-
tion of cognitive processes in gratitude. If there is a positive memory bias associated
with gratitude, what are the cognitive mechanisms that underlie this bias? Encoding,
elaborative, and/or retrieval processes could conceivably contribute to this bias (Watkins,
in press), and the information processing approach would likely be helpful in this
Another limitation of our study is that we asked our participants to recall positive
and negative events. Although our grateful participants recalled relatively more posi-
tive to negative events than did less grateful individuals, this does not necessarily
indicate that they would exhibit a similar bias if not instructed to do so. Of course, this
is a weakness in many memory studies, but it would be informative to know if grateful
individuals really are more likely to count their blessings even if not instructed to do
so. Two approaches to this issue seem feasible. One could utilize a think-aloud proce-
dure to investigate involuntary autobiographical recollections (e.g., Grimm, Watkins,
Brown, & Whitney, 2002). A second but perhaps less desirable approach would be to
ask individuals to simply list emotional memories from the past without reference to
valence. Although this approach would not be tapping spontaneous recollections, it
would avoid the constraint of requiring participants to specifically list positive and
negative events in separate trials, and thus it might be a better reflection of one’s
autobiographical memory bias.
Finally, our results leave one more mystery unresolved. Does pervasive gratitude
contribute to the recollection of positive events, or does the remembrance of positive
66 Current Psychology / Spring 2004
events make one more pervasively grateful? The design of our studies does not allow
us to resolve this dilemma. In Study 1, we found that if participants recalled positive
memories after negative memories, GRAT scores were found to increase. This finding
suggests that when positive events from the past are brought to mind one is more
likely to feel grateful for their life. However, our results do not rule out the possibility
that trait gratitude contributes to the recall of past benefits. We suspect that the most
likely answer to this conundrum is “yes”; gratitude promotes the recollection of posi-
tive events, but recalling positive events enhances one’s feeling of pervasive gratitude.
In sum, across two studies we found that trait gratitude was positively correlated
with a positive autobiographical memory bias, and in our second study we demon-
strated that this relationship was independent from depression. Although our results
leave many questions unanswered as to the mechanisms of this bias, it is clear that in
memory, grateful individuals have more accessible blessings to count.
Accepted for Publication: August 4, 2002.
Address correspondence to: Philip Watkins, Department of Psychology, 151 Martin Hall, Eastern
Washington University, Cheney, WA 99004. E-mail: philip.Watkins@mail.ewu.edu.
Beck, A. T. (1978). Depression inventory. Philadelphia: Center for Cognitive Therapy.
Blaney, P. (1986). Affect and memory: A review. Psychological Bulletin, 99, 229–246.
Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129–148.
Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective well-being: Three decades of
progress. Psychological Bulletin, 125, 276–302.
Emmons, R. A., & McCullough, M. E. (2003). Counting blessings versus burdens: An experimental investi-
gation of gratitude and subjective well-being in daily life. Journal of Personality and Social Psychology.
Fredrickson, B. F. (1998). What good are positive emotions? Review of General Psychology, 2, 300–319.
Frijda, N. H. (1988). The laws of emotion. American Psychologist, 23, 349–358.
Grim, D. L., Watkins, P. C., Brown, A., & Whitney, A. A. (2002, June). Involuntary memory bias in
depression. Presentation to the Annual Convention of the American Psychological Society, New Or-
leans, LA, June, 2002.
Holmes, D. S. (1980). Differential change in affective intensity and the forgetting of unpleasant personal
experiences. Journal of Personality and Social Psychology, 15, 234–239.
Masingale, A. M., Schoonover, S., Kraft, S., Burton, R., Waring, S., Fouad, B., Tracy, J., Phillips. S.,
Kolts, R. L., & Watkins, P. (2001, December). Gratitude and post-traumatic symptomatology in a
college sample. Presentation at the Annual Convention of the International Society for Traumatic Stress
Studies, New Orleans, LA.
McAdams, D. P., Reynolds, J., Lewis, M., Patten, A. H., & Bowman, P. J. (2001). When bad things turn
good and good things turn bad: Sequences of redemption and contamination in life narrative and their
relation to psychosocial adaptation in midlife adults and in students. Personality and Social Psychology
Bulletin, 27, 474–485
McCullough, M. E., Kilpatrick, S. D., Emmons, R. A., & Larson, D. B. (2001). Is gratitude a moral affect?
Psychological Bulletin. 127, 249–266.
Myers, D., & Diener, E. (1995). Who is happy? Psychological Science, 6, 10–19.
Payne, D. G. (1987). Hypermnesia and reminiscence in recall: A historical and empirical review. Psycho-
logical Bulletin, 101, 5–27.
Rachman, S. J. (1980). Emotional processing. Behaviour Research and Therapy, 18, 51–60.
Roediger, H. L., & Challis, B. H. (1989). Hymermnesia: Improvements in recall with repeated testing. In
C. Izawa (Ed.), Current issues in cognitive processes: The Tulane Flowerree Symposium on Cognition
(p. 175–199). Hillsdale, NJ: Lawrence Erlbaum.
Schwartz, N., & Clore, G. (1983). Mood, misattribution, and judgments of well-being: Informative and
directive functions of affective states. Journal of Personality and Social Psychology, 45, 513–523.
67Watkins, Grimm, and Kolts
Seidlitz, L., & Diener, E. (1993). Memory for positive versus negative life events: Theories for the
differences between happy and unhappy persons. Journal of Personality and Social Psychology, 64,
Seligman, M. E. P. (1998). Positive social science. APA Monitor, 29, 4.
Strack, F., Schwartz, N., & Gschneidinger, E. (1985). Happiness and reminiscing: The role of time per-
spective, affect, and mode of thinking. Journal of Personality and Social Psychology, 49, 1460–1469.
Walker, W. R., Vogl., R. J., & Thompson, C. P. (1997). Autobiographical memory: Unpleasantness fades
faster than pleasantness over time. Applied Cognitive Psychology, 11, 399–413.
Watkins, P. C. (2001). Gratitude: The benefits of an emotional state and trait. Spirituality and Medicine
Connection, 5, 1, 6–7.
Watkins, P. C. (in press). Gratitude and subjective well-being. In R. A. Emmons & M. E. McCullough
(Eds.), The psychology of gratitude. New York: Oxford University Press.
Watkins, P. C. & Curtis, N. (1994, April). Brief assessment of depression history. Presentation at the 74th
Annual Convention of the Western Psychological Association, Kona, HI.
Watkins, P. C., & Ola, D. (2001, August). Gratitude and depression: How a human strength might mitigate
human adversity. In R. A. Emmons (Chair), Gratitude and positive emotionality as links between social
and clinical science. Symposium presented at the 109th Annual Convention of the American Psychologi-
cal Association, San Francisco, CA., August, 2001.
Watkins, P. C., Porter, W. T., & Curtis, N. (1996, April). The attitude of gratitude: Development of a new
measure. Presentation to the 1996 Annual Convention of the Western Psychological Association, San
Watkins, P. C., Porter, W. T., and Miller, C. (1997, April). Gratitude and Subjective Well Being. Paper
presented to the Annual Convention of the Western Psychological Association, Seattle, WA.
Watkins, P. C., Scheer, J., Coletti, T., & Whitney, A. A. (2002, June). The debt of gratitude: Dissociating
gratitude and indebtedness. Paper presented to the Annual Convention of the American Psychological
Society, New Orleans, LA, June 2002.
Watkins, P. C., Woodward, K., Stone, T., & Kolts, R. (in press). Gratitude and happiness: Development of
a measure of gratitude, and relationships with subjective well-being. Social Behavior and Personality.
Woodward, K. M., Moua, G. K., & Watkins, P. C. (1998, April). Depressed individuals show less grati-
tude. Presentation at the 1998 joint convention of the Western Psychological Association and the Rocky
Mountain Psychological Association, Albuquerque, NM.