ArticlePDF AvailableLiterature Review

Taking time seriously: A theory of socioemotional selectivity

Authors:

Abstract

Socioemotional selectivity theory claims that the perception of time plays a fundamental role in the selection and pursuit of social goals. According to the theory, social motives fall into 1 of 2 general categories--those related to the acquisition of knowledge and those related to the regulation of emotion. When time is perceived as open-ended, knowledge-related goals are prioritized. In contrast, when time is perceived as limited, emotional goals assume primacy. The inextricable association between time left in life and chronological age ensures age-related differences in social goals. Nonetheless, the authors show that the perception of time is malleable, and social goals change in both younger and older people when time constraints are imposed. The authors argue that time perception is integral to human motivation and suggest potential implications for multiple subdisciplines and research interests in social, developmental, cultural, cognitive, and clinical psychology.
SCIENCE WATCH
Taking Time Seriously
A Theory of Socioemotional Selectivity
Laura L. Carstensen
Derek M. Isaacowitz
Susan T. Charles
Stanford University
University of Pennsylvania
Stanford University
Socioemotional selectivity theory claims that the percep-
tion of time plays a fundamental role in the selection and
pursuit of social goals. According to the theory, social
motives fall into 1 of 2 general categoriesthose related to
the acquisition of knowledge and those related to the reg-
ulation of
emotion.
When time is perceived as open-ended,
knowledge-related goals are prioritized. In contrast, when
time is perceived as
limited,
emotional goals assume pri-
macy. The inextricable association between time left in life
and chronological age ensures age-related differences in
social goals. Nonetheless, the authors show that the per-
ception of time is malleable, and social goals change in
both younger and older people when time constraints are
imposed. The authors argue that time perception is integral
to human motivation and suggest potential implications
for multiple subdisciplines and research interests in
social, developmental, cultural, cognitive, and clinical
psychology.
I often feel that death is not the enemy of life, but its friend, for
it is the knowledge that our years are limited which makes them
so precious.—Rabbi Joshua L. Liebman
(1961,
p. 106)
T
he monitoring of time is so basic to human function-
ing that it was likely instrumental in the evolution of
human thought and cognition (Suddendorf & Cor-
ballis,
1997). Markings engraved in ancestral bones dating
back to the Ice Age reflect systematic recordings of a lunar
calendar (Marshack, 1972), and the sophistication of Aztec
sundials reveals that time has been interwoven into the
social and political fabrics of societies for centuries (Aveni,
1995).
Although cultures clearly differ in their treatment of
time,
such as the tempo with which life is lived (Levine,
1997),
a basic awareness of time is ubiquitous in all known
cultures and peoples.
Scholars of theoretical physics, anthropology, astron-
omy, and philosophy have written extensively about peo-
ple's perception of time; in contrast, psychologists have
remained conspicuously silent on the topic. This is not to
say that tacit conceptions of time have been absent in social
science. On the contrary, psychologists have studied the
influence of historical periods on human development (El-
der & Clipp, 1994; Elder, Pavalko, & Hastings, 1991),
life-stage effects on values and attitudes (Sears, 1981),
cultural differences in the social norms pertaining to time
(Jones, 1988), and individual differences in time orienta-
tion (Gonzalez & Zimbardo, 1985). To the extent that
chronological age is an index of the passage of time, the
entire subdiscipline of developmental psychology is inher-
ently organized around this concept. Yet, if one really takes
time seriously and acknowledges that time provides the
structure from which people plan and implement all short-
and long-term goals, the implications for psychology are
far-reaching and have been largely ignored (Birren & Cun-
ningham, 1985).
People are always aware of time—not only of clock
and calendar time, but of lifetime. Biologist John Medina
(1996) wrote,
When contemplating life we inevitably assume the presence of an
internal clock. Wound to zero at birth, it incessantly and inher-
ently ticks away during our entire terrestrial tenure. So solid are
these concepts in our mind that we have coined the term, "life
span" to denote its boundaries, (p. 9)
As people move through life they become increasingly
aware that time is in some sense "running out." More social
contacts feel superficial—trivial—in contrast to the ever-
deepening ties of existing close relationships. It becomes
increasingly important to make the "right" choice, not to
waste time on gradually diminishing future payoffs. In-
creasingly, emotionally meaningful goals are pursued.
In the following pages we argue that the perception of
time as constrained or limited as opposed to expansive or
open-ended has important implications for emotion, cogni-
tion, and motivation. In particular, we argue that the ap-
Editor's note. Denise C. Park served as action editor for this article.
Author's
note.
Laura L. Carstensen and Susan T. Charles, Department of
Psychology, Stanford University; Derek M. Isaacowitz, Department of
Psychology, University of Pennsylvania.
Work on this article was supported by National Institute on Aging
Grant RO1-8816. We are greatly indebted to Helene Fung and Eleanor
Maccoby for their comments on drafts of this article.
Correspondence concerning this article should be addressed to Laura
L. Carstensen, Department of Psychology, Stanford University, Stanford,
CA 94305-2130.
March 1999 American Psychologist
Copyright 1999 by the American Psychological Association, Inc. OO03-066X/99/$2.0O
Vol. 54, No. 3, 165-181
165
Laura L
Carstensen
Photo by
L. A. Cicero
proach of endings is associated with heightened emphasis
on feelings and emotion states. Activities that are unpleas-
ant or simply devoid of meaning are not compelling under
conditions in which time is perceived as limited. Interest in
novel information, because it is so closely intertwined with
future needs, is reduced. Instead, when endings are primed
people focus on the present rather than on the future or the
past, and this temporal shift leads to an emphasis on the
intuitive and subjective rather than the planfui and analyt-
ical. The argument we make herein is that a temporal
emphasis on the present increases the vaiue people place on
life and emotion, importantly influencing the decisions they
make.
Subsequently, we argue that ihe perception of time is
inevitably iinked to the selection and pursuit of social
goals.
Our arguments are grounded in socioemotional se-
lectivity theory (Carstensen, 1991, 1993, 1995, 1998;
Carstensen, Gross, & Fung, 1997), which is a life-span
theory of social motivation in which the perception of time
plays a central role in the prioritization of social goals and
subsequent preferences for social partners. Because chro-
nological age is inextricably and negatively associated with
the amount of time left in life, age-related patterns do
emerge, but even these age patterns can be altered when
individuals adopt a lime perspective different from what is
predicted by their place in the life cycle.
In the following pages, we overview socioemotional
selectivity theory and describe a program of empirical
research that tests its postulates. Because we believe that
time is fundamental to human motivation, we then consider
!he broader implications (hat boundaries on time may have
for theory building and research in psychology.
Socioemotional Selectivity Theory
General
Tenets
of the Theory
Socioemotional selectivity theory addresses the role of time
in predicting the goals that people pursue and the social
partners they seek to fulfill them. Three presumptions un-
derlie the theory. First, the theory adopts as axiomatic the
belief thai social interaction is core to survival, with pre-
dispositions toward social interest and social attachment
having evolved over the millennia. Second, it considers
humans to be inherently agentic and to engage in behaviors
guided by the anticipated realization of goals (Bandura,
1982,
1991, 1997). Third, it presumes that because people
simultaneously hold multiple—sometimes opposing
goals,
the selection of goals is a precursor to action. So-
cioemotional selectivity theory maintains that the view of
time as expansive or limited influences the appraisal pro-
cess that precedes goal selection.
Over the years, different motivation theorists have
posited different sets of "basic" human needs or goals that
instigate action (Deci & Ryan,
1991;
James, 1890; Maslow,
1968;
Ryan, 1991, 1993; White, 1959). Socioemotional
selectivity theory is less concerned with which goals are
essential than with how social goals function to direct
behavior. According to the theory, diverse social goals,
ranging from seeking the answer to a question about the
weather to seeking emotional comfort, can be classified
into one of two broad functional categories: those related to
the acquisition of knowledge and those related to the reg-
ulation of emotion.
A tremendous amount of social behavior is motivated
by the pursuit of information. Contact with other people
provides a primary source of knowledge. Observations of
others and direct instruction from them play a central role
in human survival. Indeed, the intergenerational transmis-
sion of language, values, and culturally shared mental
representations are accomplished largely through social
means (D'Andrade, 1981; Shweder & Sullivan, 1990).
Knowledge acquisition through social contact is typically
necessary to master even nonsocial skills. And familiariz-
ing oneself with a broad spectrum of people allows indi-
viduals to understand the social climate, come to know
their own likes and dislikes, and begin to make evaluative
comparisons of themselves in relation to others. Thus, the
category of knowledge-related goals refers to acquisitive
behavior geared toward learning about the social and phys-
ical world.
The category of emotion motives refers in its broadest
sense to the regulation of emotional states via contact with
others. As Rolhbart (1994) states "from the earliest days,
emotion is regulated by others, and many of our emotions
and cognitions about emotion [are] developmental!y
shaped in a social context" (p. 371). Along with attempts to
avoid negative states and experience positive ones (Hig-
gins,
1997; Tomktns, 1970), the category of emotion mo-
lives also encompasses the desire to rind meaning in life,
gain emotional intimacy, and establish feelings of social
embeddedness.
166
March 1999 American Psychologist
Derek M.
Isaacowitz
According to the theory, knowledge- and emotion-
related goals together comprise an essential constellation of
goals thai motivates social behavior throughout life. On a
day-to-day basis, social goals compete with one another,
and often emotional goals vie with know ledge-related ones.
Seeking information, for example, may entail emotional
risks.
A scientist interested in critical feedback from a
colleague may expect to feel disheartened by it but wil!
pursue the feedback nonetheless. In our culture, maintain-
ing a satisfying relationship with an intimate partner typi-
cally requires that one refrain from seeking novel intimate
experiences. In addition, although people are motivated in
certain circumstances to seek confirmatory evidence of
their self-views, the same people in other circumstances are
motivated to disconfirm self-relevant views in order to
stimulate growth. Even though the desire to experience
positive emotions clearly motivates much behavior (Hig-
gins,
1987, 1997), in some cases, social contact is pursued
precisely because it elicits aversive emotions that motivate
achievement in some other domain (Norem & Cantor,
1986).
When know ledge-related goals compete with goals
involving the regulation of emotions, the relative impor-
tance of the two goals is weighed, and action is taken or not
taken accordingly.
The cardinal tenet of socioemotional selectivity theory
is that the assessment of time plays a critical role in the
ranking and execution of behaviors geared toward specific
goals.
Cognitive appraisal of time assists people in balanc-
ing long- and short-term goals in order to adapt effectively
to their particular circumstances. An expansive future is
associated with the pursuit of knowledge-related goals. The
young boy talks to his older cousin about college, not
because the information is relevant to him at the moment,
but because it may become so at some point in the future.
The student arriving for her first year at college finds a wide
range of social partners appealing and invests much time
and energy in making new friends. The young newlywed
couple spends considerable time trying to discover ways to
solve problems in their relationship because solutions will
allow them to avoid future conflicts. The theory predicts
that future-oriented goals such as these will be adaptively
prioritized when the future is perceived as expansive, and
that this will be the case even when knowledge-related
goals entail the delay of emotional rewards or emotional
costs.
When the conclusion of the appraisal process is that
time is limited, the acquisitive mode associated with un-
limited time is transformed into a more present-oriented
state.
Present orientation is likely to involve goals related to
feeling states, deriving emotional meaning, and experienc-
ing emotional satisfaction. Relieved of concerns about the
future, attention shifts to experiences occurring in the mo-
ment. When emotion regulation is the primary goal, people
are highly selective in their choice of social partners, nearly
always preferring social partners who are familiar to them,
because with these partners emotions are predictable and
often quite positive. Moreover, when time is limited social
interactions are navigated carefully in order to ensure that
their emotional quality is high. In contrast to the young
couple described above, the elderly couple often decides to
accept their relationship as it is, to appreciate what is good,
and ignore what is troubling, rather than seek new solutions
to problems. The college senior approaching graduation is
uninterested in meeting new students and instead shows
strong preferences for spending time with her best friends.
And sadly, the young boy living in a crime-ridden neigh-
borhood who believes that he will not live past the age of
20 is decidedly uninterested in conversations about college.
According to the theory, he will pursue present-oriented
goals,
perhaps by establishing strong social bonds through
gang membership. Like die older person, he perceives his
future as largely irrelevant and focuses his attention on the
present.
Clearly, the two classes of social goals described in
socioemotional selectivity theory do not reflect absolute,
nonoveriapping categories. First, Ihere is an emotional
component to all goal-directed behavior (Zajonc, 1997);
even the accounting process by which informational goals
are selected involves valenced evaluations of prospective
targets. Therefore, any distinction among goals in which
some are classified as emotional and others are classified as
nonemotional is, in some ways, problematic. We do not
dispute that there is an emotional component to information
seeking or that, conversely, there are elements of informa-
tion seeking in Ihe pursuit of emotional goals. Clearly,
there are. Second, there is ample evidence that when infor-
mation holds relevance to the immediate situation, it will
be sought regardless of temporal orientation (e.g., Turk-
Charles, Meyerowitz, & Gatz, 1997). For example, a hun-
gry person who perceives the future as limited will none-
theless speak to a waiter in a restaurant.
Rather, the delineation of social goals suggested by
socioemotional selectivity theory concerns those that are
March 1999 American Psychologist 167
Susan T.
Charles
primarily oriented to gaining knowledge or preparing for
the future and those that are primarily aimed at satisfying
emotional needs. Another way to think about the distinc-
tion is thai one class of goals is related to preparedness and
one io satisfaction in the moment. Interest in an attractive
stranger, although likely to involve both positive and neg-
ative emotions (e.g., happiness and anxiety), is acted on
primarily because of future possibilities. The potential
emotional satisfaction that may result from contact remains
largely unknown. We expect that much heterosocial con-
tact in adolescence is governed more by excitement or me
thrill of novelty than by emotional satisfaction. Young
people may embark on an exploration of potential mates,
for example, in order to find out what other people are like
in this type of relationship. Such behaviors are far from
unemotional, but the core motive underlying action is not a
search for emotionally meaningful experience. Thus, al-
though "knowledge-related" and "emotional" may be im-
perfect labels for motivations, we argue that the underlying
heuristics point to coherent streams of behavior aimed at
realizing goals and, more important, allow for complex
human behavior to be distinguished on functional grounds
such that useful predictions can be made.
Theoretical
Relevance to Life Span
Development
Human aging, inherently chronicled by the passage of time,
provides an ideal ground for exploring differences in time
perspective. Empirical studies suggest that people do carry
with them diffuse expectations about the relatively expan-
sive or limited future that awaits them. In our laboratory we
have collected questionnaire data from highly diverse sam-
ples spanning adolescence through very old age that doc-
ument clear associations between age and perceived time
left in life (Carstensen & Lang, 1997a). Older peopie
relative to their younger counterparts describe their futures
as limited and recognize that they do not have "all the time
in the world" left to pursue their goals. We expect that the
monitoring of time occurs regularly at an unconscious level
and is also primed acutely on a periodic basis by discrete
events that mark time, such as a child's wedding or a
friend's death.
Although research shows that older people consider
the past as the time of greatest activity and potency in
contrast to younger people's anticipation of future devel-
opment (Shmotkin, 1991; see also Cross & Markus, 1991;
Heckhausen, Dixon, & Baltes, 1989), the primary age
difference in time orientation concerns not the past but the
present. Older people are mostly present-oriented, less con-
cerned than the young with the far distant future (Finger-
man & Perlmutier, 1995). They do not dwell on the past,
however, as popular stereotypes suggest.
1
Rather, more
than other age groups, they focus on the here and now,
Socioe motion a! selectivity theory suggests that age-
reiated differences in the anticipated future lead to devel-
opmental trends in the ranking of knowledge-related and
emotional goals. The knowledge trajectory starts high dur-
ing the early years of life and declines gradually over the
life course as knowledge accrues and the future for which
it is banked grows shorter. The emotion trajectory is high
during infancy
2
and early childhood, declines from middle
childhood throughout early adulthood, and rises from later
adulthood into old age as future-oriented strivings become
less relevant.
Because knowledge strivings are so important from
late adolescence to middle adulthood, they are pursued
relentlessly even at the cost of emotional satisfaction. By
late adolescence and early adulthood, the regulation of
feeling states is relegated lower status than acquiring
knowledge. During this period of life, the exploration of the
world demands emotional resilience in die face of failures
and social rejections. Later in life, however, goals that are
satisfied by the resulting "feeling" state are more likely to
be pursued because they are experienced in the here and
now, a valuable commodity in the face of limited time.
Figure 1 provides the idealized trajectories of knowledge-
related and emotional goal salience across the life span.
Finally, socioemotional selectivity theory predicts that
endings are associated with qualitative changes in emo-
tional experience. In part, this is the consequence of in-
creasingly selective social partner choices and engagement
in smaller, but more emotionally meaningful, social net-
works (Carstensen el al., 1997). By shaping the social
world, negative emotional responses can be avoided and
positive ones optimized. This form of emotion control,
referred to in the literature as "antecedent regulation of
Past orientation is associated wilh depressive symptoms (e.g.. Hol-
riijn & Silver, 1998), but age is no! Rli.ihU associated with past focus.
2
In very early childhood, limited cognitive capacity precludes the
appreciation of abstract concepts of time. Subsequently, (he type of goa!
competition predicted by the theory is minima]. Infants are highly moti-
vated by both know ledge -related and emotional goals.
168
March 1999 American Psychologist
Figure 1
Idealized Model of Socioemotional
Selectivity
Theory's
Conception of
the
Salience of
Two Classes
of Social Motives Across the
Life
Span
High
Salience of
Social Motives
Emotion
Trajectory
Knowledge
Trajectory
Low
Infancy Adolescence
Middle Age
Old Age
Note.
From "The Social Context of Emotion," by L. L. Carstensen, J. Gross,
& H. Fung, 1997, Annual Review of Geriatrics and Gerontology, 17,
p. 331. Copyright 1997 by Springer Publishing Company, Inc. Reprinted
with permission.
emotion" (see Gross, in press), is arguably the most effec-
tive way to manage emotional experience at any age, but
theoretically improves over time. Older people not only
interact with fewer people, they interact primarily with
people who are well-known to them (Field & Minkler,
1988).
In old age, people's inner social circles are com-
posed primarily of old friends and family members. Kahn
and Antonucci (1980) refer to the handful of significant
others who accompany individuals through life as "social
convoys." In old age, social convoys knit individuals into
kin and friendship networks with unmatched capabilities to
affirm the sense of self and provide support in times of need
(Antonucci, 1990, 1991; Antonucci & Akiyama, 1997;
Antonucci & Jackson, 1987). Moreover, the life-long his-
tory of support exchange in long-term relationships can
allow even the very frail older person to feel needed by
others (Carstensen & Lang, 1997b). In short, the predict-
ability of interactions with familiar social partners permits
people to better navigate difficult social transactions, to
more reliably elicit positive emotions, and to obtain a sense
of social embeddedness and meaning in life.
In addition, the theory suggests that the knowledge
that time is limited has direct effects on emotional experi-
ence.
Appreciation of the fragility of life, recognition that
the passage of time cannot be stopped, and heightened
awareness of one's immediate surroundings directly alters
the experience of emotion.
3
We also expect that, relieved of
concerns for the future, endings bring out the best qualities
in people; kindness becomes a more prominent feature of
social exchanges during graduations, funerals, or retire-
ments. As people approach the ultimate ending—death
lives are evaluated, and a search for existential meaning in
life places emotion at center stage.
We have conducted several lines of research and used a
variety of research methods to test postulates from the theory.
In keeping with theoretical and empirical links between time
perspective and chronological age, much of this research has
considered age a proxy for time left in life. Over the years,
however, we have made repeated attempts to decouple age
from time, by drawing on studies of naturally occurring sub-
groups (e.g., young people living with a terminal illness) and
by using experimental methods to tease apart the effects of
age and time. In the following section, we describe our pro-
gram of empirical research. It is organized around four general
research themes that reflect central postulates of socioemo-
tional selectivity theory: (a) life-cycle differences in the sa-
lience of emotion, (b) age differences in the regulation of
emotion, (c) age differences in social network composition,
and (d) social preferences under conditions characterized by
limited or expansive time.
Empirical Findings From
Socioemotional Selectivity Theory
The
Salience of
Emotion
as People Approach
the
End
of
Life
A central postulate of socioemotional theory is that the
salience of particular social goals is influenced by the
perception of time. According to the theory, open-ended
time is associated with the pursuit of knowledge, and
constraints on time are associated with the prioritization of
emotional goals. Such changes are presumed to be evident
in the ways that people think about social partners and the
relative attention paid to emotion in cognitive operations.
Below we summarize findings from three studies that ex-
amined the cognitive dimensions along which people men-
tally represent social partners and the relative weights
placed on these dimensions at different points in the life
cycle. At the close of this section, we also report findings
from a study of age differences in incidental memory for
social narrative.
Mental representations of social partners.
We recognized early on in our research that we needed to
obtain evidence that the goal dimensions posited in socio-
emotional selectivity theory are actually evident in people's
thinking about social partners (Fredrickson & Carstensen,
1990).
Because direct questions about social goals (e.g.,
Would you learn something new by interacting with your
mother?) pose serious concerns about demand characteris-
tics,
we developed an experimental procedure based on
similarity judgments. By asking people to classify various
social partners on the basis of perceived similarities, we
were able to explore, the cognitive dimensions that people
use to make such judgments.
This experimental approach also allowed us to exam-
ine the relative weights placed on specific cognitive dimen-
3
Some previous research on time perspective associates present
orientation, and the concomitant failure to delay gratification, with hedo-
nism (Gonzalez & Zimbardo, 1985). However, these studies measure time
orientation within a relatively narrow time period; for example, will a
person study today or tomorrow? In contrast, "time" as construed in
socioemotional selectivity theory spans the life course. Although an
emphasis on the present is common to both, present orientation activated
by awareness of mortality leads to mixed emotional reactions, such as
poignancy, as opposed to hedonism.
March 1999 American Psychologist
169
sions by different groups of people. In two studies we
examined age differences in samples that included partic-
ipants as young as adolescents and as old as octogenarians.
Samples in both studies spanned a similar age range, but
the second was far more representative than the first in that
it was constructed such that men and women, blue- and
white-collar workers, and African and European Ameri-
cans were evenly distributed across the targeted age range
(Carstensen & Fredrickson, 1998). Together, these first two
studies allowed us to examine both the reliability and
generalizability of cognitive dimensions and their salience
in different age groups.
The third sample differed importantly in that age was
held constant across subsamples. In each subsample, the
average age was 37 years, and all participants were male.
However, each subsample differed by HIV status
(Carstensen & Fredrickson, 1998). One subsample was
HIV-negative, another HIV-positive but asymptomatic,
and a third subsample was HIV-positive and actively ex-
periencing symptoms of
AIDS.
Therefore, each subsample
was comparably aged, but differed in life expectancy, al-
lowing us to examine closeness to the end of life indepen-
dent of the experience factor that typically confounds chro-
nological age with place in the life cycle. Put differently,
the first two samples allowed us to examine mental repre-
sentations as a function of time since birth. The third
sample allowed us to examine the same questions as a
function of time until death. In all three studies, we pre-
dicted that identified cognitive dimensions would reflect
emotional and knowledge-related qualities of others and
that closeness to the end of life would be associated with
greater emphasis on the emotional as opposed to the knowl-
edge-related dimension.
The experimental procedures were identical in each
study. Research participants were presented with a set of 18
cards,
each of which described a particular type of social
partner. The set of social partners was designed to span a
broad spectrum of people, some of whom are likely to
provide novel information in the course of social interac-
tion (e.g., the author of a book you've read) and others who
are more likely to yield emotional payoffs (e.g., a close
friend).
Participants were asked to sort the cards into as
many or as few piles as they wished according to how
similarly they would feel interacting with the person de-
scribed on the card. After participants classified prospec-
tive social partners on the basis of similarity judgments,
data were submitted to multidimensional scaling analysis,
which revealed the dimensions along which people sorted
the cards.
The same three dimensions accounted reliably for
most of the variance in the mathematical solution in each
study. The first dimension clearly represented a valenced
(i.e.,
good-bad) dimension, which we labeled the "affec-
tive potential" of the social partners described on the cards.
Both additional dimensions were consistent with knowl-
edge-related qualities of the social partners: One was in-
terpreted as "future contact" and the other as "information-
seeking." Thus, people do indeed appear to think about
others in terms of the trajectories posited in the theory.
More pertinent to the study hypotheses were group
differences in the degree to which these dimensions gov-
erned their classifications. We found that successive age
groups placed increasingly greater emphasis on the affec-
tive potential of social partners, whereas younger adults
weighted the three dimensions fairly evenly. Not only was
this true for the overall study samples, but within the
samples the patterns held for men and women, blue- and
white-collar workers, and African and European Ameri-
cans.
Moreover, in our study based on HIV status, the
profile of findings paralleled those from our age-based
samples. HIV-positive, symptomatic, male participants
represented prospective social partners nearly exclusively
along affective dimensions, just as our oldest participants
did in our previous studies. We drew two conclusions from
this series of studies. First, the goal categories posited in
socioemotional selectivity theory are reliably reflected in
people's thinking about others, and second, the importance
of emotion in these assessments is more central in people
nearing the end of life.
Memory for social narratives. If emotion is
more salient to older as compared with younger adults,
older adults may process emotional information more
deeply and subsequently remember it better than nonemo-
tional information. Using an incidental memory paradigm,
we explored the type of information older and younger
people recall about an emotionally charged social interac-
tion (Carstensen & Turk-Charles, 1994). In light of well-
documented age differences in memory performance
(Smith, 1996), we did not predict that older adults would
remember more emotional information than younger
adults. Rather we predicted that of the material remem-
bered, proportionately more would concern emotional as-
pects of the situation.
We recruited a sample of research participants aged 20
to 83 years and asked them to read a two-page selection
drawn from a popular novel. At the end of an experimental
hour in which participants completed other unrelated ques-
tionnaires, they were asked to recall all that they could
about the story. Responses were transcribed and classified
as emotional or nonemotional. We calculated the propor-
tion of emotional to nonemotional information and exam-
ined its relationship to age. As depicted in Figure 2, the
proportion of emotional material recalled increased with
age.
Each successive age group recalled proportionately
more emotional material than nonemotional information
from the narratives.
All told, the above findings suggest that emotion is an
important dimension along which people consider others
across adulthood. Findings also suggest that emotional
qualities of others assume greater importance in mental
representations about social partners and in memory about
social interactions among increasingly older age-cohorts. A
similar profile occurs among subsamples of younger people
constructed according to their probable life expectancies;
closeness to the end of life is related to the prominence of
emotion in mental representations. The striking similarity
between the profile of older adults and the profile of
younger adults approaching death challenges a purely de-
170 March 1999 American Psychologist
Figure 2
Mean Proportion of Emotional Material Recalled in
Four
Adult Age Groups
0.35
0.30
! 0.25
1
=
a o.2o
o
S 0.15
0.10
0.05
0.0
(0.03)
T
(0.05)
(0.03)
(0.04)
0.20
0.22
0.32
0.34
20
29 35
45 53
67 70
83
Age Groups (years)
Note.
Error bars depict standard errors of the mean. From "The Salience of
Emotion Across the Adult Life Course," by L. L. Carstensen and S. Turk-Charles,
1994,
Psychology and Aging, 9, p. 262. Copyright 1994 by the American
Psychological Association.
velopmental account of change and implicates the approach
of an important ending in the instigation of cognitive shifts.
Age Differences in the Regulation of
Emotion
Evidence that emotion is more prominent in social cogni-
tive processing in groups of people nearing the end of life
is consistent with socioemotional selectivity theory's con-
tention that emotion and emotional goals are increasingly
important as people approach endings. To the extent that
greater value is placed on emotionally meaningful goals,
there also should be a concomitant enhancement of emo-
tional experience and, very likely, better regulation of
emotional experience (see Carstensen & Charles, in press).
That is, more investment in emotional goals should be
related to more resource allocation to these goals. We argue
that as people approach endings, they pay more attention to
the emotional quality of social exchanges and engage in
strategic attempts to optimize emotional aspects of impor-
tant social relationships. In everyday terms, awareness of
limited time provides the sense of perspective that softens
the experience of negative emotions (why get angry now?)
and enhances the appreciation of positive aspects of life.
The sense that "this may be the last time" changes emo-
tional reactions to positive and negative social exchanges.
We do not believe that this occurs only as people approach
the end of life. On the contrary, the theory suggests that this
experience is common. We expect that saying your last
goodbyes to a friend (even if the friendship has been
rocky),
or approaching graduation or retirement, should
also be characterized by efforts to make the experience
emotionally positive. Yet, old age is the life stage where
potential "last times" are ubiquitous, and thus the theory
makes clear predictions about emotional functioning dur-
ing the last stage of life.
Although emotion in old age has been decidedly un-
derstudied, diverse evidence from a number of laboratories
is beginning to converge to suggest that people function
very well in emotional domains of life in the later years.
Studies examining social reasoning and decision making
are suggestive of improved understanding of basic emotion
states well into adulthood as well as better integration of
emotion into cognitive processing (Blanchard-Fields, 1986;
Labouvie-Vief, DeVoe, & Bulka, 1989; Labouvie-Vief,
Hakim-Larson, DeVoe, & Schoeberlein, 1989). Mood in-
ductions under controlled laboratory conditions result in
subjective experiences that are comparably intense for
younger and older adults (Levenson, Carstensen, Friesen,
& Ekman, 1991), but older people nevertheless report
superior self-regulation of emotion, including decreased
lability and surgency and better control over negative emo-
tions (Lawton, Kleban, Rajagopal, & Dean, 1992).
Although there have been no longitudinal studies of
emotion regulation, leaving open the possibility that cohort
differences are responsible for age differences, Gross et al.
(1997) recently reported findings from a project based on
multiple samples including Norwegians, Catholic nuns,
African Americans, Chinese Americans, and European
Americans. Across these diverse samples, older people
reported better control of emotion, and consistent with
Lawton et al.'s (1992) earlier report, older adults in all
samples reported fewer negative emotions. Thus, if cohort
accounts for generational differences in the perceived reg-
ulation of emotion, they are surprisingly widespread. Fi-
nally, the self-reported reduction in surgency (Gross et al.,
1997;
Lawton et al., 1992) finds convergent support from
studies in which autonomic reactivity is measured directly
during emotional episodes; older people display relatively
lower levels of physiological activity during mood in-
ductions and while engaging in discussions of emotion-
ally charged topics (Levenson et al., 1991; Levenson,
Carstensen, & Gottman, 1994). All told, empirical research
on emotion and aging suggests that emotional functioning
is at the least well-preserved in old age, and may even
improve.
In an effort to understand the dynamics of emotional
exchanges among intimates at different ages, Carstensen
and two colleagues, John Gottman and Robert Levenson
(1995;
see also, Levenson, Carstensen, & Gottman, 1993,
1994),
conducted a study of middle-aged and older couples
who had been married for many years. After an initial
screening to ensure that both happy and unhappy couples
were included in both age groups, couples completed a
number of questionnaires about married life. They indi-
March 1999 American Psychologist
171
cated
the
degree
to
which various issues presented prob-
lems
in
their relationships (e.g., finances, children, in-laws)
and
the
degree
of
pleasure they derived from other activi-
ties along with
a
variety
of
instruments that assessed
emo-
tional
and
physical health. Couples were then observed
while they discussed
a
conflict
in
their relationship. After
an experimenter helped couples agree
on a
conflict appro-
priate
for
discussion (e.g., both parties
had to
agree
it was
a conflict), they were left alone
in a
room equipped with
remote cameras
and
asked
to
discuss
the
conflict
for 15
minutes. Throughout
the
interaction, cameras recorded
the
interaction,
and
physiological activity
was
monitored
as
couples discussed conflictual aspects
of
their relationships.
Subjective evaluations
and
direct observations
of dis-
cussions both pointed
to
superior emotion regulation
in
older couples.
By
self-report,
the
conflicts
in a
number
of
domains (e.g., finances, children,
and so on)
were less
severe
in
older
as
compared with middle-aged couples;
it is
important
to
note that this
was
true even
for
unhappily
married couples. Moreover, older couples reported that
they derived greater pleasure than middle-aged couples
in
four arenas: talking about children
and
grandchildren,
do-
ing things together, taking vacations,
and
"dreaming."
Thus,
compared with middle-aged couples, older couples
reported experiencing less conflict
and
taking greater plea-
sure
in
their marriages (Levenson
et al.,
1993).
However, this project went beyond self-reported eval-
uations
of the
relationships
we
studied.
The
videotaped
discussions
of
conflict provided
a
rich source
of
observa-
tional data. Coding
of
specific affects revealed direct
evi-
dence
for
emotion regulation. Even after controlling
for
marital satisfaction, older couples, compared with their
middle-aged counterparts, expressed lower levels
of
anger,
disgust, belligerence,
and
whining. Moreover, they were
more likely than middle-aged couples
to
express affection
to
one
another, even
as
they discussed
a
problematic aspect
of their relationship. Older couples were
not
less involved
in
the
task; they displayed levels
of
tension
and
domineer-
ingness similar
to
those
of
middle-aged couples. Rather,
they interwove expressions
of
affection along with expres-
sions
of
discontent (Carstensen
et al., 1995;
Carstensen,
Graff,
Levenson,
&
Gottman, 1996).
Previous research,
our own
included, suggests that
where there
are
differences, older people experience fewer
negative emotions
and
have greater control over their emo-
tions
in
everyday life. Our.observational research
on mar-
ried couples
is
certainly consistent with these reports. Still,
the bulk
of the
evidence
in the
literature relies
on
self-
reports, which
are
susceptible
to
distortion
as a
function
of
demand characteristics
and
implicit theories about what
behavior "should
be
like"
at
different life stages. Very
recently,
we
completed
a
project designed
to
sample
emo-
tions