NIH Public Access

Author Manuscript
J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
Published in final edited form as:
NIH-PA Author Manuscript

J Pers Soc Psychol. 2008 January ; 94(1): 158–167. doi:10.1037/0022-3514.94.1.158.

Poignancy: Mixed Emotional Experience in the Face of Meaningful

Endings

Hal Ersner-Hershfield,
Department of Psychology, Stanford University
Joseph A. Mikels,
Department of Psychology, Stanford University, now at the Department of Human Development,
Cornell University
Sarah J. Sullivan, and
Department of Psychology, Stanford University, now at the Department of Psychology, University
of California, Los Angeles
NIH-PA Author Manuscript

Laura L. Carstensen
Department of Psychology, Stanford University

Abstract
The experience of mixed emotions increases with age. Socioemotional selectivity theory suggests
that mixed emotions are associated with shifting time horizons. Theoretically, perceived constraints
on future time increase appreciation for life, which, in turn, elicits positive emotions such as
happiness. Yet, the very same temporal constraints heighten awareness that these positive
experiences come to an end, thus yielding mixed emotional states. In 2 studies, the authors examined
the link between the awareness of anticipated endings and mixed emotional experience. In Study 1,
participants repeatedly imagined being in a meaningful location. Participants in the experimental
condition imagined being in the meaningful location for the final time. Only participants who
imagined “last times” at meaningful locations experienced more mixed emotions. In Study 2, college
seniors reported their emotions on graduation day. Mixed emotions were higher when participants
were reminded of the ending that they were experiencing. Findings suggest that poignancy is an
emotional experience associated with meaningful endings.
NIH-PA Author Manuscript

Keywords
mixed emotions; poignancy; endings; aging; socioemotional selectivity theory

When John Updike described baseball legend Ted Williams' last time at bat, he wrote that the
experience “crowd[ed] the throat with joy” (Updike, 1960, p. 112). In doing so, Updike
eloquently captured the phenomenon of poignancy: Watching his baseball hero's final swing
simultaneously evoked positive and negative emotions. Recent research has shown that in
everyday life, older people experience mixed emotions, such as happiness and sadness, more
so than their younger counterparts and that this co-occurrence of positive and negative emotion
becomes more frequent with age (Carstensen, Pasupathi, Mayr, & Nesselroade, 2000; Ong &
Bergeman, 2004). In socioemotional selectivity theory (SST), Carstensen and colleagues
(2000) argued that this occurs because older adults are more aware of limited time than younger

Correspondence concerning this article should be addressed to Hal Ersner-Hershfield, Department of Psychology, Stanford University,
450 Serra Mall, Stanford, CA 94305. [email protected].
 Ersner-Hershfield et al. Page 2

adults and that the experience of “last times” increases with age (Carstensen, 2006; Carstensen,
Fung, & Charles, 2003; Carstensen, Isaacowitz, & Charles, 1999). Consequently, even
relatively positive everyday events become tinged with mixed emotions. However, a causal
NIH-PA Author Manuscript

link between limited time and emotional experience has not been documented. The present
studies offer evidence that limited time in the context of meaningful experience gives rise to
poignancy.

Aging, Emotion, and Limited Time

Over the past two decades, research on emotional functioning and emotional experience reveal
more consistency than change in aging individuals. Most aspects of emotional experience are
quite stable. Older adults do not differ from younger adults in the subjective intensities of self-
reported emotions (see Levenson, 2000, for a review) or in the profiles of physiological
reactivity (Levenson, Carstensen, Friesen, & Ekman, 1991; Levenson, Carstensen, & Gottman,
1994). The spontaneous production of emotional faces and other aspects of emotional
expressive behavior (Levenson et al., 1991; Tsai, Levenson, & Carstensen, 2000) also appear
to change little.1 However, despite stability in the experience of elicited emotions, everyday
emotional experience does change with age. Most notably, relative to their younger
counterparts, older adults report fewer negative emotional experiences (Carstensen et al.,
2000; Gross et al., 1997; Mroczek & Kolarz, 1998), especially reductions in anger (Lawton,
Kleban, & Dean, 1993; Magai, 1999), coupled with increases in positive affect (Mroczek &
NIH-PA Author Manuscript

Kolarz, 1998) or sustained levels of positive emotions (Carstensen et al., 2000). Furthermore,
there is some evidence that older adults experience more complex emotional experiences. For
example, among the young–old, written descriptions of emotion are more complex (Labouvie-
Vief, DeVoe, & Bulka, 1989), and positive and negative emotions co-occur more frequently
(Carstensen et al., 2000; Ong & Bergeman, 2004).

SST is a life-span theory of motivation that maintains that as time passes and people approach
endings, activities that are devoid of meaning are less interesting and desirable. Interest in novel
information—because it is closely linked to future needs—declines. Instead, the increasing
awareness of time constraints focuses attention on the present, and this temporal shift increases
the value that people place on the most important aspects of life. Thus, SST posits that when
the future is seen as expansive and endings are not acutely anticipated, greater focus is placed
on knowledge-related goals and information seeking. By contrast, when individuals approach
endings, they are motivated to pursue emotionally meaningful goals and focus more on the
here and now (Carstensen, 1995; Carstensen et al., 1999). This increased focus on pursuing
emotionally meaningful goals in the present leads to an intensified desire for and ultimate
experience of more positive emotion and less negative emotion over time, as is observed with
older adults. However, when time is constrained and people become progressively aware of
NIH-PA Author Manuscript

the finitude of life, emotional experiences become increasingly more complex. Such states are
positive given the pursuit of emotional goals but come to entail a mix of negative emotions
because endings, by their very nature, also increase the anticipation of last times.

Mixed Emotions
The concept of poignancy presumes the experience of mixed emotions. Evidence exists that
positive and negative emotions do occur closely together, if not simultaneously.2 At an
empirical level, Williams and Aaker (2002), for instance, found that participants felt both happy

1Note, however, that Malatesta-Magai, Jonas, Shepard, and Culvert (1992) found older adults to be more expressive.
2Although it remains contested whether differently valenced emotions can co-occur (Cacioppo & Berntson, 1994; Russell & Carroll,
1999), it should be noted that the plausibility of mixed emotional experiences does not play a role in this debate. Rather, the central
question in this dispute concerns whether individual positive and negative emotions can co-occur at the exact same moment in time
(Feldman Barrett & Russell, 1998).

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 3

and sad after viewing certain advertisements. Additionally, Larsen, McGraw, and Cacioppo
(2001) demonstrated that participants felt both happy and sad on move-out day, on graduation
day, and after watching the movie Life is Beautiful. Thus, although most emotional experiences
NIH-PA Author Manuscript

are either positive or negative, there appear to be circumstances that give rise to mixed
emotional experiences.

Defining Poignancy
Mixed emotions can take on many different forms, be it a combination of anger and joy, disgust
and satisfaction, or embarrassment and sadness. But unlike these other instances of mixed
emotions, poignancy seems to be particular to the experience of endings, of no longer having
something that one once had. And, we maintain that whereas the anticipation of an ending is
not a necessary component of other mixed emotions, it is integral to the experience of
poignancy. For example, nostalgia (Wildschut, Sedikides, Arndt, & Routledge, 2006), which
although also composed of mixed emotions, involves thinking back on momentous experiences
that have already occurred, rather than experiencing or anticipating endings. Our view is
consistent with that of philosopher Karl Duncker, who argued that poignancy results from
recognizing that something once possessed is or will no longer be present: “A feeling of not
having takes on a greater poignancy if it is a no-longer-having, a loss (whether of something
actually or almost possessed)” (Duncker, 1941, p. 418). In Duncker's framework, then, it is the
awareness or anticipation of not having something that one once had that breeds poignancy.
NIH-PA Author Manuscript

Duncker's theorizing speaks directly to the intimate relationship between poignancy and
endings in life that mark the passage of time. The heightened sense of mortality that occurs
naturally as individuals progress through life gives rise to an appreciation of life's fragility
along with an awareness that the most cherished aspects of life are fleeting. Lazarus's (1991)
appraisal theory suggests that such a sense of loss is normally associated with sadness.
However, because feelings of poignancy also involve the knowledge that one is progressing
through life, a sense of happiness is evoked as well. Accordingly, we postulate that poignancy
comprises a mixture of happiness and sadness that occurs when one faces meaningful endings.

To date, causal relationships among aging, anticipated endings, and mixed emotions remain
theoretical. In particular, SST posits that the associations between aging and poignancy are
accounted for by perceived constraints on time. If the sense of limited time rather than changes
in processes associated with aging causes changes in emotional experience, then younger adults
in limited-time situations should also, theoretically, experience comparable emotional states
as older adults under experimental conditions that constrain time.

Although previous research has demonstrated that older adults experience poignancy more
frequently than younger adults (Carstensen et al., 2000; Ong & Bergeman, 2004), there is no
NIH-PA Author Manuscript

empirical evidence to suggest that older adults experience poignancy in an ever-present

ubiquitous fashion. Mailing a letter or brushing one's teeth may be no more likely to elicit
poignancy in an older person than a younger person. Seeing an old friend, however, may
generate different experiences for the old and the young. In theoretical terms, poignancy occurs
in the face of significant endings that signify the passage of time. SST suggests that a sense of
limited time causes an individual to focus on emotionally meaningful aspects of life.
Poignancy, then, may arise not only when time is limited but also when individuals are faced
with the potential loss of something emotionally meaningful.

Overview of the Present Research

In the present studies, we tested the link between limited-time horizons and the experience of
mixed emotions. In Study 1, participants in one of two conditions were induced through
systematic imagery to imagine that they were in a meaningful location. Limited time was
manipulated by instructing participants in the experimental condition to imagine that they were

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 4

in the location for the final time. In Study 2, participants were asked to report their emotions
during the real-life ending of college graduation.
NIH-PA Author Manuscript

We sought to test two hypotheses: First, under controlled conditions in a laboratory setting,
does limited time give rise to the experience of poignancy? If poignancy results from
appreciating the fragility of life through the lens of limited time, then imagining doing
something meaningful for the final time, such as spending time at a favorite location, should
produce poignancy, regardless of age. Accordingly, in Study 1, we expected that participants
in the experimental condition would experience more poignancy during the limited-time guided
imagery induction than participants in the control condition.

Second, we examined whether a real-life ending could produce the experience of poignancy
and whether placing emphasis on this ending could heighten the experience of poignancy. As
such, in Study 2, we hypothesized that college seniors would experience poignancy on their
graduation day and that drawing attention to the natural ending would increase this sense of
poignancy.

Study 1
Method
Research Participants—Sixty younger participants (25 men, 35 women; M = 19.62 years;
NIH-PA Author Manuscript

50% Caucasian, 25% African American, 25% “other”) and 60 older participants (26 men, 34
women; M = 77.47 years; 81.67% Caucasian, 16.67% African American, 1.66% “other”) took
part in the study. Younger participants received course credit or were paid $25 for their
participation. Older participants were paid $25 for their participation. There were no significant
differences between age groups in household income, t(107) = −0.02, p = .98. Table 1 depicts
the descriptive statistics for participant information from Study 1.

Materials—An IBM R-40 Thinkpad with E-Prime software (Schneider, Eschman, &
Zuccolotto, 2002) was used for stimulus presentation and data acquisition. Participants
completed a number of measures to assess demographic background variables, physical health,
and cognitive abilities. Together, these measures provide descriptive information.

Wahler Physical Health Inventory (Wahler, 1973): The Wahler Physical Health Inventory
measures each participant's subjective interpretation of his or her physical health. Participants
are asked to rate the extent to which they are bothered by 42 different physical health symptoms.

Wechsler Digit Span Test (Wechsler, 1997): The Wechsler Digit Span Test requires
participants to repeat a string of numbers forward and backward. Considered a test of short-
NIH-PA Author Manuscript

term memory, it has been normed for older adults and correlates well with general intelligence.

F, A, S, and Category Naming Task (Spreen & Benton, 1977): Verbal fluency was assessed
by asking participants to name as many “F,” “A,” and “S” words as they could in 60 s.
Participants were then asked to name as many members of the category “animal” as they could
in 60 s.

Experimental Procedures—Participants first read and signed a consent form and then
completed a demographics questionnaire. To eliminate variation in the experimental induction,
instructions for each scenario were recorded using iTunes software and were played for
participants. Participants were seated in front of the computer and were instructed to do the
following: “Think of a place that has personal significance to you. Please think of a specific,
meaningful location that you go to with people whom you care about.” After the participant
had selected a meaningful location, the experimenter recorded the location.

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 5

Participants were then taken through three guided imagery induction trials. In the first trial,
participants were given the following instructions:
Please close your eyes. Now please take a moment to imagine being at the location
NIH-PA Author Manuscript

that you just described. As best you can, place yourself in the location. Notice your
surroundings. Notice any people whom you are with, their faces, and voices. Take in
everything that you see. Listen carefully to the sounds of your surroundings. Take a
deep breath and notice the smells. Notice the air on your skin. Now take whatever
time you need to fully experience the sights, sounds, and smells of the environment
and the overall experience of the location. When you are ready, please open your eyes.
Upon completing this induction, all participants completed an emotion questionnaire on the
computer in which they rated the degree to which they were experiencing each of 19 different
emotions (positive: accomplishment, amusement, contentment, excitement, happiness,
interest, joy, and pride; negative: anger, anxiety, boredom, disgust, embarrassment, fear,
frustration, guilt, irritation, sadness, and shame) on a 7-point scale ranging from 1 (not at all)
to 7 (extremely).

The second guided imagery induction trial immediately followed this emotion questionnaire
and was identical to the first trial in all but one respect: In this trial, participants were asked to
imagine the experience of being at their favorite location as if they were there in 2 months'
time. After this guided imagery induction, all participants completed the same emotion
NIH-PA Author Manuscript

questionnaire again.

In the third guided imagery induction, participants in the control condition were asked to
imagine the experience of being at their favorite location as if they were there in 4 months'
time. Control participants then completed the emotion questionnaire for the final time. In the
experimental limited-time condition, however, participants were asked to imagine the
experience of being at their favorite location as if they were there in 4 months' time and,
importantly, as though this would be the last time that they would be able to visit the meaningful
location. After finishing the third trial, participants in the experimental condition then
completed the emotion questionnaire for the final time. Thus, the experimental manipulation
occurred during the third trial.

All participants then completed a series of tasks and questionnaires, composed of the Digit
Span Test (Wechsler, 1997), the Verbal Fluency task (Spreen & Benton, 1977), and the Wahler
Physical Health Inventory (Wahler, 1973).

Results and Discussion

Preliminary Analyses—On the Wahler Physical Health Inventory, younger and older adults
NIH-PA Author Manuscript

were not significantly different in their self-reported physical health, t(118) = 0.16, p = .88.
On the cognitive tasks, younger adults generated more words in the “F, A, S naming” verbal
fluency task, t(118) = 4.82, p < .001, and named significantly more animals in the “animal”
verbal fluency task, t(118) = 7.98, p < .001. Younger adults also performed significantly better
in the forward Digit Span Test, t(118) = 4.75, p < .001, and the backward Digit Span Test, t
(118) = 5.46, p < .001. Table 1 depicts the descriptive statistics for these measures.

Operationalizing Poignancy—Previous research has operationalized poignancy generally

as the co-occurrence of positive and negative emotion (Carstensen et al., 2000). To measure
co-occurrence in their experience-sampling procedure study, Carstensen et al. (2000)
computed, for each participant, a Pearson's r correlation between positive and negative emotion
across 35 sampling occasions. Accordingly, r values near zero were associated with a greater
degree of co-occurrence, whereas highly negative r values indicated that positive and negative
emotion were not often present during the same sampling occasion. However, because the

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 6

present investigation was concerned with one emotional reaction as a function of a specific
time-limiting manipulation, a correlational analysis was not feasible. Rather, a more immediate
measure of mixed emotions was needed.
NIH-PA Author Manuscript

In the literature on attitudinal ambivalence, Kaplan (1972) proposed a metric that could be used
to assess the degree to which attitudes or feelings were mixed about an object or event.
Recently, Kaplan's (1972) metric has been used to capture mixed emotions (Wildschut et al.,
2006). Using Kaplan's (1972) formula, mixed emotions were defined as:

In this equation, ME refers to mixed emotions, PA to average positive affect, and NA to average
negative affect. However, as noted by Priester and Petty (1996), Kaplan's (1972) formula can
be reduced to simply two times the minimum amount of PA and NA. Thus, it is easier to
conceptualize mixed emotions as the minimum of PA and NA. Such an approach has been
taken by both Schimmack (2001) and Larsen, McGraw, Mellers, and Cacioppo (2004).

More important, the MIN formula assigns a high mixed emotion score to experiences in which
positive and negative emotions are both relatively high. For example, if PA = 3 and NA = 2,
then the mixed emotion score would equal 2; if PA = 4 and NA = 3, then the mixed emotion
NIH-PA Author Manuscript

score would equal 3. Indeed, we have argued that a poignant experience represents such a mix
of emotions, indicating that the MIN formula is appropriate for our purposes. Emotional
experiences that are unidirectional in nature (i.e., cases in which the score of one emotion is
very low and the other is moderate or high) receive the lowest possible mixed emotion score.
So, for a case in which PA = 3 and NA = 1, the mixed emotion score would equal 1; likewise,
for a case in which PA = 5 and NA = 1, the mixed emotion score would again equal 1. Both
experiences are similarly unidirectional and, as such, receive low mixed emotion scores.

However, because our conceptualization of poignancy involves the positive emotion of

happiness3 and the negative emotion of sadness, the mixed emotions measure that we use here
is defined as:

where ME refers to mixed emotions and MIN refers to minimum.

NIH-PA Author Manuscript

Tests of Main Hypotheses

Changes in happiness and sadness—Before exploring the mixed emotion scores, we
examined changes in happiness and sadness independently. A repeated measures analysis of
variance (ANOVA) with two between-subjects factors (age group: young, old; condition:
control, experimental) and two within-subjects factors (time: first imagery induction time trial,
second imagery induction time trial, third imagery induction time trial; valence: positive,
negative) was used to examine the effect of the experimental limited-time manipulation on
happiness and sadness. Results indicated that there was a main effect of valence, such that all
participants experienced a greater degree of happiness relative to sadness, F(1, 116) = 218.35,
p < .001. There was no Age Group × Valence interaction, indicating that older adults and

3Although other emotions, such as joy, are closely related to happiness, for the sake of parsimony, we opted to use a single indicator of
positive emotion (happiness) and a single indicator of negative emotion (sadness).

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 7

younger adults experienced similar amounts of sadness and happiness. Table 2 depicts the
happiness and sadness emotion mean scores for all three time trials for both conditions and age
groups.
NIH-PA Author Manuscript

Moreover, results indicated that the limited-time manipulation significantly affected

participants' emotional experience. Specifically, a Time × Valence interaction, F(2, 115) =
48.43, p < .001, was qualified by a significant Time × Valence × Condition three-way
interaction, F(2, 115) = 30.17, p < .001. To examine the nature of this interaction, we ran paired
samples t tests so that happiness and sadness before the experimental limited-time manipulation
could be directly compared with happiness and sadness after the experimental limited-time
manipulation. Specifically, participants in the control condition exhibited a significant decrease
in happiness from the second imagery induction time trial to the third imagery induction time
trial, t(59) = 2.73, p < .01. It is possible that this decrease in positive emotion was a function
of simple habituation to the initial happiness experienced (Frijda, 1988). Tellingly, control
condition participants showed no comparable change in terms of sadness, indicating that in the
absence of a limited-time manipulation, the experience of this emotion does not change from
the second imagery induction time trial to the third imagery induction time trial, t(59) = −0.38,
p = .70. By contrast, the participants in the experimental condition showed changes in both
happiness and sadness. These participants showed a significant decrease in happiness from the
second imagery trial to the third imagery trial, t(59) = 4.81, p < .001, and a significant increase
in sadness from the second imagery trial to the third imagery trial, t(59) = −10.20, p < .001.
NIH-PA Author Manuscript

As such, the paired samples t tests revealed that the limited-time manipulation caused
participants to experience more sadness and less happiness. Note, however, that although
experimental condition participants showed a trend toward experiencing more sadness than
happiness, t(29) = −1.99, p = .052, these participants were nevertheless endorsing happiness
at a level that was above the midpoint of the scale. Put another way, participants in the
experimental condition maintained levels of happiness above the midpoint of the scale despite
a spike in sadness. More important, there were no age group interactions, indicating that both
older and younger adults were affected by the limited-time manipulation in similar ways.
Again, see Table 2 for the happiness and sadness mean scores and standard deviations for all
three time trials for both conditions and age groups.

Changes in mixed emotions—The decrease in happiness and increase in sadness suggest

a more mixed emotional experience as a function of the limited-time manipulation. But because
these changes in emotional experience correspond to mean-level changes across participants,
they do not speak to changes in emotional experience within individuals. To examine changes
in emotional experience within individuals, we calculated mixed emotion scores for each
participant, as described above, using the MIN measure. We ran a repeated measures ANOVA
on these scores with two between-subjects factors (age group: young, old; condition: control,
NIH-PA Author Manuscript

experimental) and one within-subjects factor (time: first imagery induction time trial, second
imagery induction time trial, third imagery induction time trial). Results indicated that there
was a main effect of time, such that mixed emotions increased over the course of the three
imagery inductions, F(2, 115) = 8.68, p < .001. This main effect of time was qualified by a
significant Time × Condition interaction, F(2, 115) = 12.33, p < .001. To elucidate this Time
× Condition interaction, we ran paired samples t tests so that the experience of mixed emotions
before the limited-time induction could be directly compared with the experience of mixed
emotions after the limited-time induction. Participants in the control condition showed no
significant change in mixed emotions from the second imagery induction trial to the third
imagery induction trial, t(59) = 0.64, p = .53. Participants in the experimental condition, in
contrast, showed a significant increase in mixed emotions from the second imagery induction
trial to the third imagery induction trial, t(59) = −5.03, p < .001. This increase in mixed
emotions, then, can be attributed to the limited-time manipulation that occurred between the
second guided imagery induction trial and the third guided imagery induction trial. Figure 1

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 8

depicts the mixed emotion scores for all three guided imagery induction time trials for both
conditions. Finally, there was no significant Age Group × Condition interaction, nor was there
a Time × Condition × Age Group three-way interaction, indicating that both age groups
NIH-PA Author Manuscript

responded to the limited-time manipulation in similar ways.

Examining the relationship between happiness and sadness: As noted above, the limited-
time manipulation caused participants' happiness to decrease and their sadness to increase,
leading to a mixed emotional state. However, was this emotional experience reflective of a
qualitatively distinct, “mixed” state? Or, was this result just a reflection of an inverse
relationship between happiness and sadness? For participants in both conditions, there was no
relationship between happiness and sadness on the first two guided imagery induction trials,
control participants: r(58) = −.14, p = .29 and r(58) = −.22, p = .09; experimental participants:
r(58) = −.23, p = .08 and r(58) = −.16, p = .23. However, for the third imagery induction,
because of the changing nature of happiness and sadness, participants in both conditions
exhibited moderate negative correlations between happiness and sadness, control participants:
r(58) = −.36, p < .005; experimental participants: r(58) = −.39, p < .005.

In summary, we examined whether limited time resulted in the mixed emotional experience
of poignancy. Results indicated that the limited-time manipulation gave rise to a significant
increase in the experience of sadness and a significant decrease in the experience of happiness.
Furthermore, upon examining the mixed emotions metric, we found that those participants in
NIH-PA Author Manuscript

the experimental condition experienced an increase in mixed emotions as a function of the

limited-time manipulation. More important, envisioning endings did not turn a positive
experience into a negative experience; rather, it added a component of sadness to an already
happy experience. These findings held true for both younger and older age groups, suggesting
that a meaningful limited-time situation4 such as an ending can produce poignancy regardless
of age. However, this study was conducted in a laboratory setting using hypothetical endings.
To what extent, then, does poignancy arise during an actual ending in the real world? To address
this issue, we conducted a second study in which emotions were assessed during a real ending.

Study 2
In Study 1, we demonstrated that a meaningful ending gave rise to the experience of poignancy.
In that study, we used a guided imagery induction in a laboratory setting to evoke poignancy.
Although past research has demonstrated that guided imagery inductions can be used as an
effective tool to generate an emotional experience (Gerrardshesse, Spies, & Hesse, 1994), a
question remains: Do endings in a natural setting also elicit poignancy? Furthermore, can the
salience of an ending be emphasized to amplify the feeling of poignancy? To answer these
questions, in Study 2, we surveyed graduating college students about their emotional
NIH-PA Author Manuscript

experiences on graduation day. For some participants, we emphasized the ending that they
were experiencing (limited-time condition), whereas for other participants, we did not
emphasize the ending that they were experiencing (control condition). It is important to note
that because all participants were experiencing an ending, we expected feelings of poignancy
to be relatively high in both conditions. However, because we placed emphasis on the ending
in the limited-time condition, we expected poignancy to be even higher in that condition than
in the control condition.

4In Study 1, the endings that participants imagined were also meaningful. SST suggests that the sense of limited time causes an individual
to focus on things in life that are emotionally meaningful. Do even mundane everyday experiences become meaningful under limited-
time conditions? Alternatively, are mixed emotions restricted to emotionally meaningful situations? In order to disentangle these factors,
we conducted a side study in which meaning and time limitations were experimentally manipulated. Results indicated that for participants
imagining the experience of being at a nonmeaningful location, very little emotion was experienced, and, accordingly, a limited-time
manipulation did not produce significant changes in emotional experience.

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 9

Method
Participants—One hundred ten graduating seniors at Stanford University (49 men, M = 22.06
years; 56 women, M = 21.80 years, and 5 participants who failed to report their age or gender)
NIH-PA Author Manuscript

participated in the study. Seniors were approached on graduation day, June 18, 2006, and asked
whether they would like to participate in a brief survey about their experiences.

Experimental Procedure—Study 2 was composed of two between-subjects conditions. In

each condition, participants were asked to rate the degree that they were experiencing each of
19 different emotions on a 7-point scale (positive: accomplishment, amusement, contentment,
excitement, happiness, interest, joy, and pride; negative: anger, anxiety, boredom, disgust,
embarrassment, fear, frustration, guilt, irritation, sadness, and shame). In the control condition
(n = 59), participants were given the following instructions: “Keeping in mind your current
experiences, please rate the degree to which you feel each of the following emotions,” and
were then presented with the list of 19 emotions. In the limited-time condition (n = 51), in
which emphasis was placed on the ending that they were experiencing, participants were given
the following instructions: “As a graduating senior, today is the last day that you will be a
student at Stanford. Keeping that in mind, please rate the degree to which you feel each of the
following emotions,” and were then presented with the list of 19 emotions.

Results and Discussion

NIH-PA Author Manuscript

Differences in Happiness and Sadness—As in Study 1, we first sought to examine

differences in happiness and sadness as a function of the limited-time manipulation.
Independent samples t tests were used to examine the effect of the limited-time manipulation
on happiness and sadness. Participants in the two conditions showed no difference in positive
emotion, t(108) = 0.08, p = .93. However, as a function of the limited-time manipulation, there
was a significant difference in sadness, t(108) = −2.34, p < .05, such that participants in the
limited-time condition experienced more sadness than participants in the control condition.
More important, similar to Study 1, participants in the experimental condition were still
experiencing happiness at a level that was greater than the midpoint of the scale. In fact, in this
naturalistic examination of poignancy, despite the ending they were facing, participants in both
conditions exhibited more happiness than sadness, control condition: t(58) = −9.68, p < .001;
experimental condition: t(50) = −6.28, p < .001. Table 3 depicts the happiness and sadness
mean scores for participants in both conditions.

Differences in Mixed Emotions—Just as in Study 1, mixed emotion scores were calculated

for each participant. An independent samples t test was used to examine whether these scores
differed as a function of condition. As hypothesized, making the ending of graduation day more
salient caused participants in the limited-time condition to show higher mixed emotion scores
NIH-PA Author Manuscript

than participants in the control condition, t(108) = −2.34, p < .05. Figure 2 depicts the means
of the mixed emotion scores in both conditions from Study 2.

It is important to note that although participants in the control condition were not reminded of
the ending that they were experiencing, they were experiencing an ending just the same, as
evidenced by their relatively high mixed emotion scores. In fact, mixed emotion scores from
control condition participants in this study were significantly higher than the mixed emotion
scores from control condition participants in Study 1, t(117) = −3.50, p < .005. These results
indicate that a natural ending can produce the experience of poignancy. Furthermore, when
participants were reminded of the ending that they were experiencing, feelings of poignancy
were heightened.

Examining the Relationship Between Happiness and Sadness—In Study 1, we

found a negative relationship between happiness and sadness for participants in both the control

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 10

and experimental conditions. Because participants repeatedly imagined being at a meaningful

location and were then asked to imagine a final time there, it is possible that a bias in affective
forecasting might have led to this negative correlation. In previous studies, Gilbert, Wilson,
NIH-PA Author Manuscript

and colleagues have demonstrated that when thinking about future events, people tend to
overestimate how negative they will actually feel and underestimate how positive they will
actually feel (Gilbert, Pinel, Wilson, Blumberg, & Wheatley, 1998; Wilson & Gilbert, 2003,
2005). It is possible, then, that because participants in Study 1 were asked to make judgments
regarding feelings they would experience during a hypothetical ending, they reported feeling
more sadness and less happiness than they would when facing an actual ending in the real
world. Such an inflation of sadness and deflation of happiness perhaps led to the negative
correlation between the two emotions. Because our measure of emotion in Study 2 involved a
real-world examination of poignancy, we did not expect there to be a strong negative
relationship between happiness and sadness. Indeed, in Study 2, the correlations between
happiness and sadness were nonsignificant, control condition: r(57) = .17, p = .21; experimental
condition: r(49) = .26, p = .07. Unlike our hypothetical exercise in the laboratory setting, in
the case of a real-life ending, it seems that both emotions related to each other in a slightly
positive direction.

General Discussion
We maintain that the experience of poignancy can be considered a mixed emotional experience
NIH-PA Author Manuscript

that occurs when one is reminded of the passing of time during a meaningful experience. To
test the causal link between endings and the experience of mixed emotions, we measured
participants' emotional responses while imagining a visit to a meaningful location for the last
time. Results from Study 1 indicated that the limited-time manipulation gave rise to a
significant decrease in the experience of happiness and a significant increase in the experience
of sadness from baseline. Moreover, upon examining the mixed emotions scores, we found
that a sense of limited time did indeed produce a mixed emotional experience. It should be
noted that this mixed emotional experience was not unidirectionally negative in nature but
rather remained positive as well. Hence, one critical component to the experience of poignancy
appears to be a sense of limited time. Results from Study 2 showed that real-life endings outside
of the laboratory also produce a sense of poignancy. As such, we provide evidence that the
experience of poignancy involves mixed emotions in the face of a meaningful ending.

Our conceptualization of poignancy arises out of the observation that chronological age is
correlated with mortality. Subsequently, in daily life, we postulate that older people are more
aware of endings than younger people are. We hypothesized that endings provoke a mixed
emotional experience because a sense of limited time focuses individuals on emotionally
meaningful goals. Such goals tend to be positive in nature, yet, because an ending highlights
NIH-PA Author Manuscript

the passing of time, it also gives rise to an increase in negative emotion. Previous research has
demonstrated that older adults experience such poignant states more often than younger adults
(Carstensen et al., 2000; Ong & Bergeman, 2004), but the extent to which this change in
emotional experience is due to a sense of limited time and not simply the result of changes in
other processes associated with aging has not been explored. More important, in both studies
we demonstrated that the experience of poignancy occurs equally in younger and older adults.
Thus, here we provide the first evidence that some of the changes in emotional experience
associated with aging (viz. poignancy) are a result of facing a meaningful ending rather than
a result of chronological age per se.

Although these results provide experimental support for the assertion that certain age-related
changes in emotional experience are due to a shift in time perspective, alternative possibilities
have been offered to address mixed emotions in old age. Williams and Aaker (2002), for
instance, have proposed that older adults experience co-occurring positive and negative

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 11

emotions because they are more likely than younger adults to accept the duality of emotional
experience. In other words, compared with younger adults, older adults are better able to
recognize that emotions do not have to be experienced in a unidirectional way. Such an
NIH-PA Author Manuscript

explanation for poignancy is not necessarily at odds with SST's account that poignancy arises
as a result of limited time. It is possible, for example, that limited time can also cause an
individual to be more accustomed to accepting duality. Just the same, future research should
attempt to disentangle these two explanations of poignancy.

Our findings lend support to Cacioppo and colleagues' contention that mixed emotional
experiences can and do occur (Cacioppo & Berntson, 1994; Cacioppo, Gardner, & Berntson,
1999). As demonstrated in both studies, when individuals face endings, emotional experience
includes a mixture of happiness and sadness. But although emotional experience is mixed,
these results do not speak to the nature of the behavioral output, or action tendency (Lazarus,
1991) that would arise from the experience of poignancy. Previous research makes us doubtful
that the action tendency of poignancy—or the action tendency of any emotion for that matter
—would be mixed (Feldman Barrett & Russell, 1998). Lazarus (1991), for example, has noted
that the action tendency of happiness is to approach others and share with them one's positive
outcomes. On the contrary, sadness leads to a state of withdrawal and avoidance of others
(Frijda, 1988). In theory, then, the action tendency of poignancy would most likely be
singularly valenced. Future studies are necessary, however, to discern the nature of behavioral
responses to mixed emotional experiences.
NIH-PA Author Manuscript

We found that poignant experiences entailed a mix of happiness and sadness. Was this
experience truly mixed? Or, were our results a reflection of an inverse relationship between
the two emotions? Although it is true that there was a significant negative correlation between
happiness and sadness for participants experiencing poignancy in the experimental condition
of Study 1, r(58) = −.39, p < .005, it should be noted that this correlation is moderate in
magnitude (Cohen, 1988) and not large, as might be expected if positive emotions simply
replaced negative emotions. Furthermore, we found a trend toward a positive relationship
between happiness and sadness in our real-world examination of poignancy. As noted above,
the discrepancy between the correlations from Study 1 and Study 2 might be a function of a
bias in affective forecasting that occurred during Study 1 but not Study 2. Unlike Study 1, in
Study 2, participants were asked to report how they felt during a real-life ending.
Understandably, we did not obtain a similar negative correlation. Nevertheless, future research
should be conducted to further examine the relationship between happiness and sadness during
mixed emotional experiences. Furthermore, in both Study 1 and Study 2, we found that the
mix of happiness and sadness that participants felt was characterized by the addition of sadness
to an already happy experience. Put another way, the endings did not in themselves produce
the experience of happiness. However, it could be the case that some special cases of endings
NIH-PA Author Manuscript

will actually give rise to both happiness and sadness. Future research should attempt to
document various types of endings and the effects that they have on poignancy.

Lastly, researchers such as Larsen, Hemenover, Norris, and Cacioppo (2003) and Spiegel
(1998) have noted that the coactiva-tion of positive and negative emotion can often be an
important step in the coping process, for the ability to face and accept negative life events is a
crucial aspect of the healing course. Spiegel, Bloom, Kraemer, and Gottheil (1989) found, for
example, that breast cancer patients who underwent group therapy and expressed the full range
of their emotions lived twice as long as control patients who were in no such therapy group.
Similarly, Bower, Kemeny, Taylor, and Fahey (1998) demonstrated that bereaved HIV-
seropositive men showed positive health outcomes when they both confronted stressors in their
lives and found a sense of meaning in the trauma that they had undergone. Finally, Carstensen
et al. (2000) found that emotional complexity in everyday life is associated with less
neuroticism and better emotional control. Thus, to the degree that mixed emotional experiences

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 12

can aid in the coping process, poignancy may be a special case of a mixed emotional experience
that may help individuals in dealing with anticipated loss.
NIH-PA Author Manuscript

In summary, findings from the present studies lend further support to the notion that certain
age-related changes in emotional experience are due to the experience of limited time and are
not necessarily a result of organic processes associated with aging, such as the degradation of
physical or cognitive abilities. As SST predicts, a sense of limited time, when coupled with
the loss of something emotionally meaningful, produces the experience of poignancy.

Acknowledgments
This research was supported in part by National Institute on Aging Grant RO1-8816 awarded to Laura L. Carstensen
and by a grant from the Norman H. Anderson fund. We thank Sam Maglio and Gina Schiel for their assistance in
collecting data as well as Jennifer Heil, Anthony Ong, Nilam Ram, Rick Robins, Ewart Thomas, Jeanne Tsai, and the
members of the Bay Area Affective Science Training Program for helpful discussions of statistical and conceptual
issues.

References
Bower JE, Kemeny ME, Taylor SE, Fahey JL. Cognitive processing, discovery of meaning, CD4 decline,
and AIDS-related mortality among bereaved HIV-seropositive men. Journal of Personality and Social
Psychology 1998;66:979–986.
NIH-PA Author Manuscript

Cacioppo JT, Berntson GG. Relationship between attitudes and evaluative space: A critical review, with
emphasis on the separability of positive and negative substrates. Psychological Bulletin 1994;115:401–
423.
Cacioppo JT, Gardner WL, Berntson GG. The affect system has parallel and integrative processing
components: Form follows function. Journal of Personality and Social Psychology 1999;76:839–855.
Carstensen LL. Evidence for a life-span theory of socioemotional selectivity. Current Directions in
Psychological Science 1995;4:151–156.
Carstensen LL. The influence of a sense of time on human development. Science 2006 June 30;312(5782):
1913–1915. [PubMed: 16809530]
Carstensen LL, Fung HL, Charles ST. Socioemotional selectivity theory and emotion regulation in the
second half of life. Motivation and Emotion 2003;27:103–123.
Carstensen LL, Isaacowitz DM, Charles ST. Taking time seriously: A theory of socioemotional
selectivity. American Psychologist 1999;54:165–181. [PubMed: 10199217]
Carstensen LL, Pasupathi M, Mayr U, Nesselroade J. Emotional experience in everyday life across the
adult life span. Journal of Personality and Social Psychology 2000;79:644–655. [PubMed: 11045744]
Cohen, J. Statistical power analysis for the behavioral sciences. Vol. 2nd. Hillsdale, NJ: Erlbaum; 1988.
Duncker K. On pleasure, emotion, and striving. Philosophy and Phenomenological Research 1941;1:391–
430.
NIH-PA Author Manuscript

Feldman Barrett L, Russell JA. Independence and bipolarity in the structure of current affect. Journal of
Personality and Social Psychology 1998;74:967–984.
Frijda NH. The laws of emotion. American Psychologist 1988;43:349–358. [PubMed: 3389582]
Gerrardshesse A, Spies K, Hesse FW. Experimental inductions of emotional states and their effectiveness:
A review. British Journal of Psychology 1994;85:55–78.
Gilbert DT, Pinel EC, Wilson TD, Blumberg SJ, Wheatley TP. Immune neglect: A source of durability
bias in affective forecasting. Journal of Personality and Social Psychology 1998;75:617–638.
[PubMed: 9781405]
Gross JJ, Carstensen LL, Pasupathi M, Tsai J, Skorpen CG, Hsu AYC. Emotion and aging: Experience,
expression, and control. Psychology and Aging 1997;12:590–599. [PubMed: 9416628]
Kaplan KJ. On the ambivalence-indifference problem in attitude theory and measurement: A suggested
modification of the semantic differential technique. Psychological Bulletin 1972;77:361–372.
Labouvie-Vief G, DeVoe M, Bulka D. Speaking about feelings: Conceptions of emotion across the life
span. Psychology and Aging 1989;4:425–437. [PubMed: 2619949]

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 13

Larsen, JT.; Hemenover, SH.; Norris, CJ.; Cacioppo, JT. Turning adversity to advantage: On the virtues
of the coactivation of positive and negative emotions. In: Aspinwall, LG.; Staudinger, UM., editors.
A psychology of human strengths: Perspectives on an emerging field. Washington, DC: American
NIH-PA Author Manuscript

Psychological Association; 2003. p. 211-216.

Larsen JT, McGraw AP, Cacioppo JT. Can people feel happy and sad at the same time? Journal of
Personality and Social Psychology 2001;81:684–696. [PubMed: 11642354]
Larsen JT, McGraw AP, Mellers BA, Cacioppo JT. The agony of victory and thrill of defeat: Mixed
emotional reactions to disappointing wins and relieving losses. Psychological Science 2004;15(5):
325–330. [PubMed: 15102142]
Lawton MP, Kleban MH, Dean J. Affect and age: Cross-sectional comparisons of structure and
prevalence. Psychology and Aging 1993;8:165–175. [PubMed: 8323721]
Lazarus, RS. Emotion and adaptation. New York: Oxford University Press; 1991.
Levenson, RW. Expressive, physiological, and subjective changes in emotion across adulthood. In:
Qualls, SH.; Abeles, N., editors. Psychology and the aging revolution: How we adapt to longer life.
Washington, DC: American Psychological Association; 2000. p. 123-140.
Levenson RW, Carstensen LL, Friesen WV, Ekman P. Emotion, physiology, and expression in old age.
Psychology and Aging 1991;6:28–35. [PubMed: 2029364]
Levenson RW, Carstensen LL, Gottman JM. Influence of age and gender on affect, physiology, and their
interrelations: A study of long-term marriages. Journal of Personality and Social Psychology
1994;67:56–68. [PubMed: 8046584]
Magai, C. Affect, imagery, attachment: Working models of interpersonal affect and the socialization of
NIH-PA Author Manuscript

emotion. In: Cassidy, J.; Shaver, P., editors. Handbook of attachment theory and research. New York:
Guilford Press; 1999. p. 787-802.
Malatesta-Magai CZ, Jonas R, Shepard B, Culvert C. Type A personality and emotional expressivity in
younger and older adults. Psychology and Aging 1992;7:551–561. [PubMed: 1466824]
Mroczek DK, Kolarz CM. The effect of age on positive and negative affect: A developmental perspective
on happiness. Journal of Personality and Social Psychology 1998;75:1333–1349. [PubMed:
9866191]
Ong AD, Bergeman CS. The complexity of emotions in later life. Journal of Gerontology: Psychological
Sciences 2004;59B:117–122.
Priester JR, Petty RE. The gradual threshold model of ambivalence: Relating the positive and negative
bases of attitudes to subjective ambivalence. Journal of Personality and Social Psychology
1996;71:431–449. [PubMed: 8831157]
Russell JA, Carroll JM. On the bipolarity of positive and negative affect. Psychological Bulletin
1999;125:3–30. [PubMed: 9990843]
Schimmack U. Pleasure, displeasure, and mixed feelings? Are semantic opposites mutually exclusive?
Cognition & Emotion 2001;15:81–97.
Schneider, W.; Eschman, A.; Zuccolotto, A. E-Prime user's guide [Computer software and manual].
Pittsburgh, PA: Psychology Software Tools; 2002.
NIH-PA Author Manuscript

Spiegel D. Getting there is half the fun: Relating happiness to health. Psychological Inquiry 1998;9:66–
68.
Spiegel D, Bloom JR, Kraemer HC, Gottheil E. Effects of psychosocial treatment on survival of patients
with metastatic breast cancer. Lancet 1989;2:888–891. [PubMed: 2571815]
Spreen, O.; Benton, AL. Neurosensory Center Comprehensive Examination for Aphasia. Victoria, BC:
Neuropsychology Laboratory, University of Victoria; 1977.
Tsai JL, Levenson RW, Carstensen LL. Autonomic, subjective, and expressive responses to emotional
films in older and younger Chinese Americans and European Americans. Psychology and Aging
2000;15:684–693. [PubMed: 11144327]
Updike J. Hub fans bid kid adieu. The New Yorker 1960 October 22;:109–131.
Wahler, HJ. Wahler Physical Symptoms Inventory. Los Angeles: Western Psychological Services; 1973.
Wechsler, D. WAIS-III Administration and scoring manual. San Antonio, TX: Psychological
Corporation; 1997.

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 14

Wildschut T, Sedikides C, Arndt J, Routledge C. Nostalgia: Content, triggers, functions. Journal of

Personality and Social Psychology 2006;91:975–993. [PubMed: 17059314]
Williams P, Aaker JL. Can mixed emotions peacefully coexist? Journal of Consumer Research
NIH-PA Author Manuscript

2002;28:636–649.
Wilson TD, Gilbert DT. Affective forecasting. Advances in Experimental Social Psychology
2003;35:345–411.
Wilson TD, Gilbert DT. Affective forecasting: Knowing what to want. Current Directions in
Psychological Science 2005;14(3):131–134.
NIH-PA Author Manuscript
NIH-PA Author Manuscript

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 15
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Figure 1.
Mixed emotions scores for control and experimental conditions in all three time trials in Study
1. Shading for Times 1 and 2 are similar in order to represent the contrast of interest with Time
3. Error bars represent standard error of the mean.
NIH-PA Author Manuscript

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 16
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Figure 2.
Mixed emotions scores for control and limited-time conditions in Study 2. Error bars represent
standard error of the mean.
NIH-PA Author Manuscript

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 Ersner-Hershfield et al. Page 17

Table 1
Participant Characteristic Information in Study 1
NIH-PA Author Manuscript

Younger Older
(n = 60) (n = 60)

Variable M SD M SD

Age 19.63 1.40 77.47 6.88

Household income ($ in K) 63.39 59.77 63.58 47.80
Digit Span
 Forward 9.88 2.42 7.88 2.19
 Backward 8.72 2.73 6.15 2.41
Verbal Fluency
 Animal 21.92 4.73 14.95 4.83
 F, A, S naming 14.88 3.45 11.78 3.60
Wahler Physical Symptoms 44.10 22.33 37.60 22.22

Note. K = thousands.
NIH-PA Author Manuscript
NIH-PA Author Manuscript

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Table 2
Happiness and Sadness Means and Standard Deviations by Time Trial and Condition for Both Age Groups in Study 1

Happiness Sadness

Time 1 Time 2 Time 3 Time 1 Time 2 Time 3

Ersner-Hershfield et al.

Variable M SD M SD M SD M SD M SD M SD

Younger
 Control 6.13 0.90 6.10 0.96 5.47 1.70 2.03 1.75 1.93 1.67 2.13 1.66
 Exp. 6.00 1.05 5.87 1.50 4.60 1.55 2.77 1.63 2.63 1.71 5.43 1.72
Older
 Control 6.00 1.41 5.97 1.22 5.63 1.50 2.47 2.17 2.50 2.06 2.40 2.19
 Exp. 6.23 1.04 5.73 1.34 4.47 2.39 1.87 1.41 2.03 1.47 5.33 2.25
Total
 Control 6.07 1.18 6.03 1.08 5.55 1.59 2.25 1.98 2.22 1.88 2.27 1.93
 Exp. 6.12 1.04 5.80 1.41 4.53 2.00 2.32 1.58 2.33 1.61 5.38 1.94

Note. For the Younger group, n = 60, with n = 30 in each condition. For the Older group, n = 60, with n = 30 in each condition. Exp. = Experimental.

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.
Page 18
 Ersner-Hershfield et al. Page 19

Table 3
Happiness and Sadness Means and Standard Deviations by Condition in Study 2
NIH-PA Author Manuscript

Happiness Sadness

Condition M SD M SD

Control 5.49 1.22 3.07 1.70

Limited time 5.47 1.39 3.82 1.67

Note. For the control group, n = 59. For the limited-time group, n = 51.
NIH-PA Author Manuscript
NIH-PA Author Manuscript

J Pers Soc Psychol. Author manuscript; available in PMC 2010 January 18.