Finding Positive Meaning and Its Association With Natural Killer Cell Cytotoxicity
Among Participants in a Bereavement-Related Disclosure Intervention
Julienne E. Bower, Ph.D.
Cousins Center fbr Psvchoneuroimnrunology
Department of Psychiatry and BiobehavioraI Sciences
University of California, Los Angeles
Margaret E. Kemeny, Ph.D.
Health Psychology Program
Universitv ol' Califbrnia. San Francisctr
Shelley E. Taylor, Ph.D
Department of Psychokrgy
University of California. Los Angeles
.fohn L. Fahey. M.D.
of Microbiology and Immunology, Inrrnunology,
University of Calilbrnia, Los Angelcs
and Molecular Genetics
This study tested the hypothesis that cognitive fr(,((ssing
about a past hereayement would prcduce inr:reases in goals
and priorities indicative of'finding positive meaning .from the
loss. It wu.s Jurther hypothesiTed that increases in meuning-re-
lated goals would be associated with changes in immune .func-
tion, speciJically increased natural killer cell cvtotoxicity
(NKCC). Cognitive pntcessing was manipulated using written
emotional discLosure. ForN-three vrtomen who had lost o cktse
relcttive to breast cancer wrote about the death (cognitive pro-
group) or about rutnemotional topics week-
l.y.for 4 weeks. Contrary to predictions. written discLosure did
not induce changes in meaning-related goals or NK cel.l pa-
rameters. However, women in both experimental gntups who
reported po.sitive changes in meaning-related goals over the
study period also showed increases in NKCC. Results suggest
that prioritizing goals emphasizing relationships, personal
growth, and striving for meaning in life may have positive bio-
Iog,ical correlates hut that solitary written disclosure may not
be sfficient to induce changes in these goals in response to a
(Ann Behav Med
Preparation of this article was supported in part by the Norman Cousins
Program in Psychoneuroimmunology at the University of California,
Los Angeles; a National Institute of Mental Health (NIMH) Research
Scientist Development Award (MH00820) Dr. Kemeny, and grants
from the NIMH (MH056880) and National Science Foundation
(S8R9905 I 57) Dr. Taylor.
We would like to thank Drs. David Wellisch, Michelle Craske, and
Steve Cole tbr their contributions to this project, and Dr. Annette
Stanton for her assistance in reviewing an earlier drafi of this article.
Reprint Address: J. E. Bower, Ph.D., UCLA Division of Cancer Pre-
vention and Control Research, A2-125 CHS, Box 956900, Los An-
geles, CA 90095-6900. E-mail: firstname.lastname@example.org
O 2003 by The Society of Behavioral Medicine.
changes in the human immune system is now well documented.
A variety ofstresstul lil'e experiences, ranging tiom acute exper-
imental stressors (e.g., mental arithmetic) to more chronic natu-
(e.g., divorce, bereavement)
ated with changes in both enumerative and functional immune
parameters ( l). However, little is currently known about how
psychological responses to these stressors may influence immu-
nological outcomes. In particular, positive psychological re-
sponses to stress, and positive psychological outcomes that may
occur following stressful life events, have rarely been examined
in psychoneuroimmunological investigations.
The possibility that traumatic events may provoke positive
psychological changes has been a theme of literature, poetry,
and philosophy for centuries and has recently gained recogni-
tion in the research literature on recovery from trauma (2-5).
In particular, many individuals report finding positive meaning
in their lives following a major stressor. The term finding
meaninc is used here to describe an enhanced sense of the
value and importance of one's life, coupled with (and poten-
tially inspired by) a more acute awareness of life's fragility
and preciousness. One of the key elements of finding meaning
is a clarification of the values and goals that imbue one's lif'e
with a sense of meaning and purpose and a new or renewed
commitment to those goal pursuits. Of note, this conceptual-
ization of finding meaning, which focuses on the significance
of the stressful event for the individual's life, goals, and future,
can be contrasted with efforts to understand or make sense of
the traumatic experience (6).
A growing number of studies have found that some individ-
uals are able to find positive meaning from even the most stress-
ful of life events, including bereavement (6,7); diagnosis with
life-threatening conditions, such as HIV/AIDS (8.9) and cancer
(10,1l); coping with impaired fenility (12); or suffering a heart
attack (13) or stroke (14). These individuals typically report
positive shifts in their view of themselves. their relationships
between exposure to stressful events and
have been associ-
Volume 25, Number 2,2003
with others, and their priorities and goals in the aftermath of the
stressor. In particular, priorities and goals related to personal
growth and the importance of interpersonal relationships have
been noted to increase. Overall. studies suggest that individuals
who are able to derive positive meaning fbllowing a stressful lif'e
event also show better psychological ad.lustment
To date. there has been only minimal examination of the ef-
f'ects of flnding meaning on physical health outcomes or specific
systems, including the immune system. Affleck et
al. ( I 3) found that heart attack patients who perceived benefits
fiom their attack, including changes in philosophy of lif'e and
values, were less likely to have another attack and had lower lev-
els of morbidity at an S-year fbllow-up. Crue ss et al. ( l5) lbund
that women diagnoscd with early-stage brcast cancer who re-
ported finding more benef rts fiom thc cancer experience tbllow-
ing a structured psychological intervention also showed de-
creases in serum cortisol. We examined the associution
tinding meaning. changes in the immune syslenl, and mortality
in a group of HIV-seropositive
gay men who had recently lost a
cfose friend or partner to AIDS (7). Finding meuning was de-
fined in this study as a major shifi in values, priorities, or per-
spectives in response to the bereavement. Thc discovery of
meaning was associated with positive changes in a key immuno-
logical marker of HIV progression and a lower rate of AIDS-re-
lated mortality over a 4- to 9-year follow-up period, controlling
for potential biobehavioral and psychosocial confounds. These
data provide preliminary evidence that linding meaning may be
an important outcome of stressful lil-e experiences for physical
as well as psychological well-being.
Atthis point, it is unclear ftow people are able to construe pos-
itive meaning fiom trauma, and specifically whether psychologi-
to the event play a role. We have proposed that cog-
nitive processing in response to a stressful event may promote the
discovery of positive meaning from the experience (l). Cognitive
processing can be defined as actively thinking about a stressful
situation, the thoughts and feelings it evokes, and its implications
fbr one's lif'e and future (16). Cognitive processing is hypothe-
sized to promote the revision of pre-existing belief s and assump-
tions about the self and the world to be consistent with the reality
of the traumatic experience ( l7). We hypothesize that cognitive
processing may also promote a reconsideration
one's life and one's pnorities and goals, thereby facilitating the
discovery of positive meaning liom the event. In support of this
hypothesis, we have shown that men who engaged in cognitive
processing in response to the loss of a friend orloverto AIDS were
more likely to find meaning from the death (7). Similar results
were reported by Calhoun et al. ( l8), who found that stressor-re-
lated cognitive processing was significantly correlated with
posttraumatic growth in a study of college undergraduates.
ever, because cognitive processing and meaning-related
were assessed cross-sectionally in both of these reports. the
causal link between these constructs could not be determined.
This study was designed to experimentally test our theoreti-
cal model linking cognitive processing, finding positive meaning.
and changes in specific immune parameters. Despite increased
interest in finding meaning, no previous studies have specifically
tried to induce meaning shifts using an experimental intervention.
(6.9. 10, 12).
of the direction of
Meaning and NK Cytotoxicity 147
tional disclosure paradigm developed by Pennebaker
this paradigm because
active contemplation of a stressful experience. including one's
"deepest thoughts and feelings" about the experience.
dard disclosure paradigm. individuals are assigned
about the most stressful clr traumatic event they have ever experi-
enced or about a variety of'trivial topics lirr briel'periods
over scveral days. Disclosurc has been associated
changes in immune parameters (20-22). and l'ewer health center
visits (23) among college undergraduates,
ments in disease status among patients with immunologically
lated medical conditions (24,25). Evidence suggests
related to cognitivc processing) are associated with more positive
health outcomes (20.26). supporting the idea that cognitive pro-
cessing may be an important part of the disclosure process.
Study participants were women who had lost a close rela-
tive to breast cancer and perceived themselves to be at increased
risk fbr developing this disease. We were intercsted
because they. like the bereaved HlV-positive men in our earlier
study, faced both the loss of a significant other to a particular
disease and the threat of developing that disease themselves. We
hypothesized that this dual stressor may heighten the threat of
mortality associated with bereavement, leading to more
in-depth reflection on one's own life and goals and potentially
provoking more profound shifts in meaning. Prcvious research
has shown that women who have lost a parent to cancer report
elevated levels of cancer-related
death (27), suggesting that this group may be an important target
for psychological interventions (28).
Immune analyses for this study focused on the natural killer
(NK) cell subset of the immune system, because of both its im-
portance for general immune regulation and its potential role in
controlling the development and metastatic spread of tumor
cells. NK cells are part of the nonspecific immune response
have the ability to kill both virally infected cells and certain
types of tumor cells. Some investigators have theorized that the
NK cell system may play an important role in immuno-
surveillance against tumor cells (29), based in part on studies
showing lower levels of NK activity in healthy individuals with
a high familial incidence of cancer (30), although this is a con-
troversial area. The NK system was also of interest because it
has been shown to respond to psychological states similar to
those considered in this study, including emotional manipula-
tions (3 I ) and psychological interventions
The central hypotheses of this study were as fbllows: (a)
Bereavement-related disclosure will lead to increases in goals
and pnonties related to hnding meaning, and (b) increases in
importance of meaning-related goals and priorities will lead to
increases in NK cell cytotoxicity (NKCC). We focused here on
changes in the importance of meaning-related priorities and
goals, such as cultivating relationships and seeking personal
growth, because increases in these goals are one of the central
components of finding meaning. In addition, there is evidence
that goal engagement is important for overall psychological
well-being (34) and may have physiological relevance. includ-
using a version of the emo-
to elicitit is specifically designed
ln the stan-
to write either
and with inrprove-
in this group
distress even l4 years after the
148 Bower et al.
ing links to the immune system (35). Although the primary anal-
yses focused on changes that occurred after disclosure, we also
examined the association between meaning-related goals and
NK cell parameters at baseline. We hypothesized that women
who rated meaning-related goals as more important at study en-
trv would also show hisher levels of NKCC.
Forty-three women who had lost a close relative to breast
cancer and perceived themselves to be at higher than average
risk for developing breast cancer participated in this study be-
tween February 1 997 and May 1998. Participants were recruited
from cancer-related organizations (i.e., Revlon-University of
California, Los Angeles (UCLA) Breast Center, UCLA Tumor
Registry, Wellness Community), bereavement-related
tions (i.e., bereavement support groups, hospice-related be-
reavement services, Motherless Daughters), advertisements in
local newspapers, public service radio announcements, and in-
formational f'lyers. Recruitment materials indicated that the pur-
pose of the study was to assess the psychological and immuno-
logical effects of writing about "personal life events." To avoid
priming participants for writing about a traumatic topic and spe-
cifically about the death of their relative (36), women who
phoned about the study were informed only that the writing ses-
sions might involve writing about life events "of an emotional
nature." In addition, they were infbrmed that study participation
involved completing questionnaires and an interview focusing
specifically on their relative's death.
Of the 167 women who responded to our recruitment mate-
rials, 156 agreed to complete a telephone screening interview to
assess their eligibility for the study. Women were eligible for
study participation if they had lost a mother, sister, grandmother,
and/or aunt to breast cancer; if their relative had died after they
were l0 years old; and if they perceived themselves to be at
higher than average risk for developing breast cancer. Exclusion
criteria included (a) age under 20 or over 60 years old; (b)
blood/needle phobic; (c) history of immunologically related dis-
ease or condition that might affect the immune system; (d) regu-
lar use of immunosuppressive medication; (e) history of psychi-
atric hospitalization, severe psychological distress in last 6
months, or current use of psychotropic medication; (0 con-
sumption of more than 15 alcoholic beverages per week; and (g)
use of illicit drugs. Women who had lost their aunts or grand-
mothers to breast cancer were asked additional screening ques-
tions to determine the impact of the loss. Only those women
who indicated that they were close to their aunUgrandmother,
that their aunt/grandmother had played an important role in their
lives, and that they remembered the death of their aunt/grand-
mother were eligible for study participation.
Eighty-four (54Vo) of the 156 women screened were ex-
cluded for reasons related to age, medical status, medication
use, and failure to meet bereavement- and risk-related criteria
(e.g., relative diagnosed with breast cancer but still alive, did not
perceive self to be at increased risk for breast cancer). Of the 72
women who were eligible, l0 elected not to participate because
Annals of Behavioral Medicine
of scheduling problems or distance from UCLA (or both), 18
scheduled initial appointments but were unable or elected not to
participate and were never seen, and I did not complete the fol-
Participants averaged 42.14 years of age (SD = 8.32, range
= 24-60). The majority were White (n = 35), with a smaller
number of Asian American/Pacific Islander (n = 4), African
American (n = 3 ), and Latina ( = I ) participants. The sample was
quite well educated; all participants had completed some col-
lege, l7 were college graduates, and 18 had postgraduate train-
ing. Most of the participants were employed either full time (n =
23) or part time (n = I I ) at study entry, and just over one half of
the women were either married (n = l5) or involved in an inti-
mate relationship (n = 8). Twenty-two had children.
The majority of study participants had lost their mother to
breast cancer (n = 34), with a smaller number having lost their
sister (n = I ), their aunt (r? = 4), or their mother and their sister (n
= 4). The deaths occurred an average of l2 years prior to study
entry (range = 0'-34 years). The average age of the participants
at the time of their relative's death was 30 years (range = l4-58
years). Many participants had other relatives who had been di-
agnosed with breast cancer (n = 24), and a substantial minority
had lost more than one relative to this disease (n = l3). There
was no difference between the women who participated in the
study and those who were eligible but did not participate in
terms ofcurrent age or age at relative's death. A higher percent-
age of women who did not participate had lost a sister to breast
cancer, whereas a higher percentage of women who did partici-
pate had lost both a mother and a sister.
Participants were screened over the phone to determine
their eligibility fbr the study then scheduled for a baseline (Time
1) assessment. All assessments
were scheduled within the same
2-hr time period in the morning (8:00 a.m.-I0:00 a.m.) to con-
trol for diurnal variations in immune parameters. The timing of
the assessments was also scheduled to coincide with the
follicular phase of each participant's menstrual cycle (i.e., Days
l-7) to control for possible effects of menstrual phase on im-
mune outcomes. Participants were asked to refrain from con-
suming food, drinking alcohol or caffeine, using tobacco, taking
nonprescription medication, and engaging in strenuous exercise
during the l2-hr period prior to their appointment. Participants
were mailed a consent form and questionnaire packet to com-
plete before their first appointment.
At the initial assessment, a baseline blood sample was drawn
after a l5- to 20-min rest period. A semistructured interview was
then conducted by Julienne E. Bower, focusing on the circum-
stances ofthe relative's illness and death. In addition to objective
information (e.g., date of death), the interview assessed the cir-
cumstances of the death and the participant's psychological re-
sponse to the loss. After the interview, participants were ran-
domly assigned to either the emotional disclosure (n = 20) or
control condition (n = 23) by a trained research assistant. Bower
was blind to the random assignment. Participants were blocked
on menopausal status to ensure equal proportions of pre- and
Volume 25, Number 2,2003
group were assigned to write about a particulartopic at home once
a week for 4 weeks; this weekly schedule was used to maximize
the effectiveness of the intervention (37). The instructions for
each condition followed those used in previous disclosure stud-
ies, with one notable exception: Rather than having participants
in the emotional disclosure condition write about any stressful or
they were specifically instructed to write
about the death o1'their relative to cancer, including their emo-
tional reactions to her death and the implications of her death fbr
their own lives. Participants in the control condition were asked to
write about a variety of non-emotional
day and a description of their homes.
All participants were asked to schedule one 20-min block of
time per week during which they could sit down and write undis-
turbed about their assigned topic. Four mailing envelopes were
provided, along with a booklet in which to write the essays.
day betbre each scheduled writing session, a research assistant
contacted participants to remind them of the writrng assignment.
Participants were asked to mail back the assignments
and were contacted by phone if the assignments
Four weeks afier the initial assessment
of the final writing session), participants returned to UCLA lbr a
follow-up (Time 2) appointment. Participants again completed
questionnaires the evening before their scheduled appointments
for collection that morning. After a 15- to 20-min rest period,
blood samples were drawn, and a second interview was con-
ducted focusing on changes the participants may have experi-
enced following their relative's death. After the interview. par-
ticipants were thoroughly debriefed and thanked. They were
paid $50 for their participation.
women in each condition. Panicipants in each
topics, such as plans fbr the
were not received.
(and within I week
ventory (38) was used to assess the importance of values and
goals related to finding meaning. This inventory includes l6 lif-e
goals that individuals frequently endorse as important in their
lives. Participants were asked to indicate the importance of each
goal on a 7-point scale that ranged trom | (not at ull important)
to 7 (extremelt- important). Previous factor analysis of this scale
in a sample of HIV positive gay men revealed two factors, la-
beled Intrinsic Life Goals and Extrinsic Life Goals by Kemeny
et al. (39). This factor structure was recently corroborated and
replicated in two additional populations using structural equa-
tion modeling (40). Intrinsic and Extrinsic Life Goals emerged
as distinct latent factors, and the individual paths from the indi-
cators to the latent constructs were each significant. The
strength of associations between the latent factors and the item
indicators was the same across the three samples (HlV-positive
gay men, HlV-negative gay men, and college undergraduates).
This study focused on the Intrinsic Goals subscale of the
Life Goals Inventory, which includes nine goals that are closely
related to finding meaning, including personal development
("develop myself as a person," "explore new possibilities or be
adventurous"), interpersonal relationships ("cultivate close
friendships," "give and receive love"), and "strive for meaning
in my life." The Intrinsic Goals subscale has demonstrated
discriminant validity in samples of healthy college students and
variables. The Life Goals In-
Meaning and NK Cytotoxicity 149
in both HlV-positive and HlV-negative gay men (40). Spe-
cifically, the scale does not correlate with measures of depressed
mood, self-esteem, social desirability, and negative affectivity in
these samples. Internal reliability of the Intrinsic Goals subscale
in this sample was .8-5. The Intrinsic Goals subscale was admin-
istercd at Tinre I and Time 2.
was assessed with the Center for Epidemiologi-
cal Studies-Depression scale (CES-D), a 20-item scale de-
signed by Radloff (41 ) to assess depressive symptomatology in
the general population. Respondents indicated how often they
had experienced a variety of affective and vegetative symptoms
in the past week on a 4-point scale that ranged fiom 0 (rarely- or
none of the time) to 3 (most or all of the time). The CES-D has
good reliability and validity in both clinical and community
samples (41). Internal reliability of the CES-D in this sample
was .94. Higher scores on this measure indicate a higher level of
depressive symptomatology, with scores greater than l6 sug-
gesting the presence of clinically significant symptoms of de-
pression. The CES-D was administered at Time I and Time 2.
A subscale of the Texas Revised Inventory of Grief (42)
(TIG) was used to assess curent feelings of grief about the tar-
get relative's death. This l3-item scale taps common symptoms
of grief, including "l very much miss her," "l can't help thinking
about her," and "I still get upset when I think about her." Respon-
dents rated the truthfulness of each item on a 5-point scale that
ranged from I (completely .false) to 5 (completeLy true). This
measure has acceptable reliability and validity, with a coeffi-
cient alpha of.86 and evidence showing that the intensity of
f'eelings varies as expected over time (42). Cronbach's alpha fbr
the TIC in the current sample was .85. The TIG was adminis-
tered at Time I and Time 2.
An essay evaluation scale modeled afier that of Pennebaker
and Beall (23) was administered as a manipulation check at
Time 2. This scale asks respondents to report about the extent to
which their essays were personal and revealing of their emo-
tions; how much they had told others about the essay topics, had
wanted to talk about them, and had actively held back from tell-
ing others about them; and how diffrcult and frustrating the writ-
ing had been. As an additional measure of cognitive processing.
participants were asked at the Time 2 interview to rate the extent
to which they had reflected on their relative's death since the
study began on a l0-point scale that ranged from | (not at all) to
l0 (a ereat deal).
Measures ofpotential confttunding variables. Health be-
haviors that may influence immune status were assessed
self-report questionnaire, including use of caffeine, alcohol, rec-
reational drugs, vitamin supplements, and medications (pre-
scription and nonprescription) in the last 7 days. Nutrition, exer-
cise, and sleep (average number of hours of sleep nightly and
number of nights on which less sleep was obtained than needed)
in the last 7 days were also assessed.
Immunologic measures NK cells were defined as CD3-/
CD56+/CDl6+ and enumerated using flow cytometry with three-
color immunofluoresence. NKCC was measured in a standard
3-hr 5lCr release assay with K562 cells as target cells (see 3l
Bower et al.
for a detailed description of NK cytotoxicity procedures used in
our immunology laboratory). NKCC values are expressed as
percentage specific lysis at a 50: I target:effector cell ratio.
An examination of baseline psychological characteristics
revealed significant differences between the control and experi-
mental groups on the CES-D and TIG, with the control group
reporting higher levels of depressive symptomatology and grief
than the experimental group at baseline (for CES-D, M = ll .O,
SD = 13.5 for controls, and M = 9.6, SD = 8.0 fbr experimental
participants; tbr TIG. M = 3.8, SD = 0.6 for controls, and M =
3.3, SD = 0.9 fbr experimental participants). These dill-erences
appeared to be attributable to the random assignment of all 4
women who had lost both their mother and their sister to the
control condition, these women had higher levels of both de-
pression and grief at baseline compiued with the others. and
when they were excluded from the analysis diff'erences
measures were no longer significant. No group difl'erences were
observed on other bereavement-related
tance of meaning-related goals, NK cell parameters, or demo-
graphic characteristics (i.e., ethnicity, employment status, part-
ner status. educational level, number of children).
Mean TIG scores for both groups fell between the scale
midpoint of 3 (partly true and partlv fttlse) and 4 (mostly true),
suggesting a moderate level of current grief. The scale items that
were most strongly endorsed included "No one will ever take
her pf ace in my life" (M = 4.7 , Mdn = 5), "l very much miss her"
(M = 4.5, Mdn = 5.0), and "I feel it's unfair that she died" (M =
3.8. Mdn = 5\.
Essay Characteristics and Reflection Ratings
The mean number of essays received fiom women in the ex-
perimental group was 3.45 (range = 0-4) and from women in the
control group was 3.65 (range = l-4).There were 5 women (3 in
the experimental group and 2 in the control group) fiom whom
none or only one of the required essays was received, although
all reported that they had completed each of the four writing as-
signments. When data for these women were deleted from the
main analyses, the effects were essentially identical; thus, we
decided to retain these participants for analyses examining ef-
fects of disclosure.
Women in the experimental group rated their essays as
more personal and more emotionally revealing than women in
the control group and reported that they had wanted to talk to
someone about the essay topics more and had actively held back
from telling others about the topics more than women in the con-
trol group (ps < .01). Participants in each group rated the essays
as equally frustrating. Essay ratings for the experimental group
were comparable to those seen in other disclosure studies (21).
Overall, these results support the validity of the disclosure para-
digm in that the disclosure condition appeared to elicit more
emotional writing about personally meaningful topics than did
the trivial topics condition.
Annals of Behavioral Medicine
During the Time 2 interview, all but 2 participants indicated
that they had reflected on their relative's death since the study
began (both of these were in the control condition). There was
no difference between disclosure and control group participants
in how much they had reflected on the death during the 4-week
study period (disclosure group mean ref-lection rating = 6.9, SD
= 2.2: control group mean reflection rating = 6.2. SD = 2.8).
Potential Biobehavioral Confounds
Befbre conducting our primary analyscs. we assessed
number of potential confounding variables that might account
for the variance observed in the immunological outcome mea-
sures (43). Results showed that use of caffeine, alcohol, and vi-
tamins were correlated with NKCC at Time I. Time 2, or bothl
hours of sleep and use of vitamins were correlated with NK cell
percentage at Time I , Time 2, or both (all ps < . l0). These van-
ables were thus used as controls in analyses involving these im-
munologic outcomes. Neither age nor menopausal status was
associated with NK cell parameters.
Many of the study participants (n = 29) reported using pre-
scription or nonprescription medications (or both) in the week
prior to study entry. Most of the medications used could be
grouped into two general categories: (a) hormonal preparations
(e.g., oral contraceptives) and (b) non-narcotic analge-
(e.g., ibuprofen). We conducted inde-
pendent-samples / tests to determine whether use of either type
of medication was associated with the primary immune parame-
ters. Neither use of hormonal preparations nor use of analge-
was associated with NKCC or NK cell
percentage at Time I or Time 2 (all ps > . I 0). Of note, although
participants were screened for use of immunosuppressive
cations, a small number did report using nasal inhalers contain-
ing steroids (n=2) and narcotic analgesic combinations (n=2).
Use of each of these medications was treated as a covariate in
analyses involving immune parameters.
We first examined the association between importance of
meaning-related values and NK cell parameters at Time l. As
predicted, importance of meaning-related goals was positively
correlated with NKCC at Time | (r = .33, p < .05); women who
placed more importance on meaning-related goals had higher
levels of NKCC. This correlation remained signihcant in analy-
ses controlling lbr potential behavioral confounds (i.e., use of
caffeine, alcohol, and medication use) as well as analyses con-
trolling for depression. Of note, depressive symptomatology
was negatively correlated with importance of meaning-related
goals atTime I (r = -.44, p<.01) but was not significantly cor-
related with NK cell parameters.
Effects of Essay Writing
We hypothesized that bereavement-related
sure would induce increases in importance of meaning-related
goals. However, a 2 x 2 Group (disclosure vs. control) x Time
(Time I vs. Time 2) mixed-model repeated measures analysis of
variance (ANOVA) revealed no significant main effects of
Volume 25, Number 2,2003
group or time and. more important, no Group x Time interaction
on importance of meaning-related goals (all Fs < l; see Table I
for mean scores on meaning-related goals at Time I and Time
2). Thus. contrary to predictions, written disclosure did not
cause changes in the importance of goals related to finding
meaning over the course of this l-month study.
We also examined the effect of written disclosure on
changes in NK cell parameters.
mune effects that were independent of goal shifts. A 2 x 2 Group
(disclosure vs. control) x Time (Time I vs. Time 2) mixed-
model repeated measures ANOVAs on both NK cell percentage
and NKCC revealed no significant main effects of group and no
significant Group x Time interaction for either measure (all Fs <
l: see Table I for mean NKCC and NK cell percentage
Time I and Time 2). There was a marginally significant eff'ect of
time on NK cell percentage.
f( | , 38) = 3.12, p < .09; in both ex-
perimental groups, NK cell percentage decreased slightly from
Time I to Time 2. Thus. in contrast to results obtained in previ-
ous studies with other immune subsystems
sponse to Phytohemagglutinin,
antibody titers to Epstein-Barr
virus, antibody response to hepatitis vaccination), written dis-
closure was not associated with changes in the percentage or
cytotoxicity of circulating NK cells.
Additional analyses were conducted excluding the 4 partic-
ipants with elevated CES-D scores at baseline; these yielded the
same pattern of results for both meaning-related goals and NK
cell parameters. We also examined a number of bereavement-re-
lated factors that may have influenced study results, including
number of years since relative's death (death less than l0 years
ago vs. over I 0 years ago based on median split), type of relative
lost (first vs. second degree), and other family history of breast
cancer diagnosis or death. There was no evidence that these
variables moderated the effect of disclosure on meaning-related
goals, NK cell percentage, or NKCC. Finally, we conducted
analyses to determine whether women who were more dis-
tressed or reported more grief at study entry were more likely to
show changes in goals or NKCC. We found no evidence that
scores on the CES-D or the TIG moderated the experimental
as writing may have had im-
(i.e., proliferative re-
Descriptive Statistics for Meaning-Related Goals
and Natural Killer (NK) Cell Parameters at Time 1
and Time 2 (Postintervention)
l\leaning and NK Cytotoxicity
in Goal Changes and NKCC
Our second major hypothesis was that changes in mean-
ing-related goals from pre- to postintervention
ated with changes in NKCC. We were unable to test this hypoth-
esis by comparing experimental groups because the disclosure
intervention did not lead to changes in meaning-related
Instead, we examined individual differences in goal shifts
among participants in both experimental
goal importance were observed among both disclosure and con-
trol group participants (mean change = .33, range = -6-10). Par-
ticipants were classilled into groups on the basis of changes in
their scores on the Lif'e Goals lnventory from Time I to Time 2.
Participants whose scores either changed in a positive direction
(i.e., rated meaning-related
goals as more important; n = l4) or
stayed at the samc high level (n = 3) were classified as "im-
proved/high meaning," whereas participants whose scores
changed in a negative direction (i.e., rated meaning-related
goals as less important; rr = l9) or stayed at the same low level (n
= l) were classified as "declined/low meaning." Of note. therc
was no association between experimental group assignment and
improvement group classifi cation.
A 2 x 2 Group (improved/high vs. declined./low) x Time
(Time 1 vs. Time 2) mixed-model repeated measures ANOVA
computed on NK cell parameters revealed a significant Group x
Time interaction for NKCC. As shown in Figure 1. NKCC in-
creased from Time I to Time 2 among participants classified as
improved/high on meaning-related goals, whereas NKCC de-
would be associ-
conditions. Changes in
I lmproved/High on Meaning-Related Goals
! Declined/Low on Meaning-Related Goals
FIGURE 1 Change in natural killer cell cytotoxicity (NKCC) associ-
ated with change in importance of meaning-related goals from pre- to
postintervention. Individuals who rated meaning-related goals as more
important or whose goal ratings remained high over the course of the
study ("improvedAigh" group; 11 = 17) showed an increase in NKCC,
whereas those who rated these goals as less important or whose ratings
stayed low ("declined/low" group; n = 20) showed a decrease in
( ) A
Importance of meaning-related goals
NK cell cytotoxicity
NK cell percentage
18.2 4'7 .2
I52 Bower et al.
creased among those classified as declined/low. F( 1, 35) = 4.29.
p <.05. This effect remained significant in analyses controlling
fbr baseline depression scores, changes in depression scores
from pre- to postintervention, and potential behavioral con-
founds. In addition. the Group x Time interaction for NKCC re-
mained significant after excluding the 4 women whose scores on
the life goals measure did not change liom Time I to Time 2.
Thus. shifts in meaning-related goals were correlated with
changes in NKCC from pre- to postexperiment. Change in per-
centage ofNK cells was not observed, suggesting
tional capacity of these cells rather than their number or distri-
bution was associated with soal shifts.
that the func-
The theoretical model guiding this investigation proposed
that actively thinking about a past stressor,
ing, would lead to positive shifts in goals and priorities related to
finding positive meaning from the experience. Furthermore, this
model predicted that increases in meaning-related goals would
be associated with changes in immunological parameters, and
specifically in increases in NKCC. Results offered only partial
support for these hypotheses. We found no evidence that cogni-
tive processing, in the form of writing about the loss of a relative
to breast cancer, was associated with changes in meaning-re-
lated goals or NK cell parameters; however, we did find a posi-
tive association between importance of meaning-related goals
and NKCC, both at baseline and over the course of study partici-
pation. These results are discussed later.
or cognitive process-
Effects of Written Disclosure
Contrary to predictions, writing about the death of one's
relative did not elicit positive changes in importance of mean-
ing-related goals, and disclosure did not lead to changes in NK
cell percentage or NKCC. Prior research conducted with college
undergraduates has shown positive effects of disclosure on psy-
chological and immune variables (37), although the specific
outcomes of interest here have not been assessed previously.
Why were effects not seen in this study?
It is possible that differences in our experimental paradigm,
such as having participants write at home rather than in a super-
vised laboratory setting, may have affected results. Conducting
the follow-up assessment within I week of the final writing as-
signment may not have allowed sufficient time for disclosure ef'-
fects to emerge, although immune effects were observed I week
after disclosure in an earlier study (20). ln addition, the hetero-
geneity of our sample may have influenced study results. There
was considerable variability in the participants' age at their rela-
tive's death, the number of years since the loss, and the extent to
which they had already thought and talked about the experience;
although we found no evidence that these or other bereave-
ment-related factors influenced responses to disclosure, our
sample may not have been sufficiently large to capture these
The intervention might have been more effective if partici-
pants had been allowed to select their own disclosure topic, as
has been done in most previous disclosure studies. Disclosure
Annals of Behavioral Medicine
interventions have been successful when participants are in-
structed to write about a specific topic; fbr example. Stanton et
al. (41) fbund that breast cancer patients who were asked to
write about their thoughts and feelings about their cancer expe-
rience showed improvements in psychological and physical
well-being conrpared with women who wrote only about the
factual aspects ofthe experience. However, several studies have
nol shown eff'ects of writing about bereavement on mental or
physical health outcomes, even among recently bereaved indi-
viduals (45,46). Positive effects of bereavement-related
sure may be particularly difficult to detect when the study oc-
curs several years afier the bereavement experience, as was the
case fbr many participants
in our study. Although women who
have losl rclatives to breast cancer do report elevated levels of
canccr-related distress up to 1.1 ycars afier the relative's death
(21), it is unclear whether these women also experience pro-
longed or complicated grief reactions that might be amenable to
Of note, we lbund no evidence that
women who reported higher levels o1' grief or depressive
symptomatology at study entry were more likely to benefit fiom
the disclosure intervention.
Finally, it is possible that the standard disclosure paradigm
may not be efTective in eliciting changes in meaning-related pri-
orities and goals. To date, research on finding meaning has been
primarily descriptive in nature; relatively little is known about
how individuals are able to find meaning for themselves, and
even less is known about the role of external interventions in
promoting meaning changes. We used disclosure as a means of
eliciting cognitive processing, which has been linked to finding
meaning in observational studies (7,18). However, a few ses-
sions of solitary, written disclosure may not be sufficient to in-
duce the level of cognitive processing necessary for provoking
changes in intrinsic values and goals in a systematic way: in-
deed, our results suggest that this type of writing did not lead to
an overall increase in reflection on the relative's death, outside
of the 20-min writing session. Instead, a more intensive inter-
vention may be required, involving more time and possibly an
interpersonal context providing guidance and support. One ex-
ample of this type of intervention is supportive-expressive
apy, which fbcuses specifically on existential themes related to
finding meaning and is conducted in groups over an extended
time frame (47). There is also evidence that a more didactic,
structured group intervention emphasizing reappraisals of the
event and positive coping strategies may be effective in helping
individuals find benefits from certain types of stressors (48);
whether this approach specifically affects life goals has not been
determined. Given the potential importance of meaning-related
goals for psychological and physical health, future research
should more carefully explore the ways in which individuals are
able to derive meaning from traumatic life events, particularly
whether meaning changes can be prompted by disclosure or
other psychological interventions.
We have discussed why the writing intervention used in this
study may not have prompted changes in meaning-related goals.
What. then, might have caused the goal changes that did occur?
Recall that shifts in goal importance were observed among indi-
viduals in both the disclosure and control conditions over the
lblume 25, Number 2, 2003
course o1'the study. These changes may have resulted from natu-
ral fluctuations in goal importance or fiom external events or
triggers in participants' lives. It is also possiblc that some
women may have reordered their goals as a result of study par-
ticipation. Simply being part of a study focusing on breast can-
cer and, in particular, having the opportunitv to discuss their rel-
ative's death during the initial interview may havc prompted a
reconsideration ol' goals and priorities fbr some participrnts.
This finding is consistent with the idea that in the absencc of a
powerful intervention the discovery of mcaning seems to be an
idiosyncratic process. dependent on aspects ol'an individual's
personality, lil'e circumstances.
event. The identiflcation of these individual difference factors is
another critical step in understanding positive adaptation t<r
stressful events and determining who may benelrt most fronr
and response to the triggering
Finding Meaning and NKCC
The second rnajor goal of this study was to test the hypothe-
sis that increases in the impo(ance of meaning-related
would be associated with increases
uate the association between goal changes and NKCC in the ab-
sence of a disckrsure effect, we collapsed data across experi-
mental groups. The results showed that at baseline, women who
placed greater importance on goals such as cultivating relation-
ships, personal development, and striving fbr meaning in lif'e
had higher levels of NKCC. Furthermore, women who rated
these goals as more important after study participation also
showed increases in NKCC from pre- to postintervention. The
relation between meaning-related goals and NKCC remained
significant in analyses controlling fbr a number of potential
psychosocial and biobehavioral confounds, including depressed
mood, reducing the plausibility that other fhctors may have me-
diated this eff'ect.
Because the changes in meaning-related goals were not
caused by the experimental manipulation, we cannot conclude
that shifts in goals provoked changes in NK cell activity. How-
ever, both the cross-sectional and longitudinal data do suppoft
an association between importance of meaning-related goals
and NK cell function. To our knowledge, this is the first study to
show an association between prioritization of goals related to
finding meaning and an immunological outcome. These find-
ings are consistent with results from an earlier study, in which
we fbund that shifts in goals and priorities indicative of tinding
meaning were associated with both positive immunological
changes and a lower rate of mortality among bereaved HIV-pos-
itive men (7). Statements reflective of finding meaning in that
study (coded fiom transcripts of bereavement interviews) in-
cluded greater appreciation for loved ones, emphasis on rela-
tionships, new personal growth goals, and a commitment to en-
joying and appreciating life more fully; these statements are
very similar to items on the Life Goals Inventory used in this in-
vestigation. Together, results from these studies provide prelim-
inary evidence that individuals who place greater importance on
intrinsic goals, such as personal growth, relationships with oth-
in NK cell function. To eval-
Meaning and NK Cytotoxicity 153
ers. and finding meaning in lit'e. may show positive changes in
certain components of the immune system.
Physiological mediators of the association
ing-related goals and NK cell parameters
mined. The sympathetic nervous system and hypothalamic-pi-
tuitary-adrenal axis are responsive to psychological triggers
such as stress and depression and have demonstrated effects on
NK cells in vivo and in vitro (49.50). However. little is known
about the relation between "positive" psychological states and
these physiological systems. It is possible that prioritizing
goals may promote more adaptive autonomic
clr neuroendocrine responses to stress, such as more rapid
to repeated stressors
beneficial cff-ects on neural and hormonal factors. such as de-
creases in baseline cortisol ( | 5). The physiological mechanisms
through which finding meaning gets "under the skin" and inf'lu-
ences thc immune system is a current tbcus of our research.
have not been deter-
(51). or may have direct
Goal Engagement and Immunity
This study adds to a growing body of literature on positive
psychological states (52) and the potential fbr positive change
and growth lbllowing exposure to trauma (2--5). We fbcused
here on one dimension ofpositive change: increased or renewed
importance of goals linked to intrinsic values such as relation-
ships. personal growth, and finding meaning in life (53). In-
creased goal importance may indicate a greater striving toward
or engagement with particular life goals. This type of goal en-
gagement may represent
an important motivational domain that
has linkages to neurophysiological
mune system. Older theories have posited that a generalized
(e.g., giving up) may have health implica-
tions (54), and current animal research on defeat suggests
neurophysiological and immunological changes coincident with
a disengaged or defeated state (55,56). This study supports the
possibility that a generalized form of goal engagement----one
relevant to more intrinsic goals. such as finding meaning-may
have pclsitive immunological correlates.
that this type of goal engagement may be difficult to promote
and that a relatively briet-, solitary intervention such as written
disclosure may be insufficient to affect changes in this domain.
systems relevant to the im-
Our results also suggest
(l) Herbert TB, Cohen S: Stress and immunity in humans: A
meta-analytic review. Psychosomatic Medicine. l993,
R, Frantz CM: The impact of trauma on mean-
ing: From meaningless
world to meaningful
Brewin C (eds), Iher TransJormation
London: Wiley, 1997.
(3) Taylor SE: Adjustment
(4) Tedeschi RG, Calhoun LG: Trauma and Transformation:
Growirtg in the A.fiermath
of Sufering. Thousand Oaks, CA:
(5 ) Park CL, Folkman S: Meaning in the context of stress and cop-
ing. Reviel' o.f General
life. In Power M,
of Meaning in Psvchologi-
A theory of cogni-
Bower et al.
(6) Davis CG, Nolen-Hoeksema
and benefiting from the experience: Two construals of meaning.
Journal oJ Personalirr* and Sociol Psychologt'. 1998,
(7) Bower JE, Kemeny ME, Taylor SE, Fahey JL: Cognitive pro-
cessing, discovery of meaning, CD4 decline, and AIDS-re-
lated mortality among bereaved HIV seropositive gay men.
Journol o.f Consulting and CLinical Ps'-cholog'-. 1998.
(8) Schwartzberg SS: Struggling for meaning: How HlV-positive
gay men make sense of AIDS. Proftssional PsvL:hoktgv:
s earch and P ruc tic e. 1993, 24 :483490.
(9) UpdegraflJA, Taylor SE, Kemeny ME. Wyatt GE: Positive and
negative effects of HlV-infection in women with low socioeco-
nomic resources. I'ersonali6- and Sociul Ps,-cfutlogy
( l0) Taylor SE, Lichtman RR, Wood JV: Attributions, belief.s about
ctrntrol, and adjustment to breast cancer. Jourrutl of Personalin'
and S oc ia I P st cho lo gy'. | 984. 46:489-502.
( I l)
Andrykowski MA, Brady MJ, Hunt JW: Positive psychological
adjustment in potential bone marrow transplant recipients: Can-
cer as a psychosocial transition. P.s.vcho-Oncoktgl'.
(12) Tennen H, Affleck G, Mendola R: Causal explanations for in-
fertility: Their relationship to control appraisals and psycholog-
ical adjustment. In Stanton A, Dunkel-Schetter C (eds), Infertil-
ity: Perspectives From Stress and Coping. Researci. New York:
Plenum. 1991. 109-132.
( 13) Affleck G, Tennen H, Croog S, Levine S: Causal attributions,
perceived benefits, and morbidity after a heart attack: An 8-year
study. Journal of Consulting and Clinical Psychology. 1987,
( 14) Thompson SC: The search for meaning following a stroke. Ba-
sic and Applied Social Psychologv.1991, l2:81-96.
(15) Cruess DG, Antoni MH, McGregor BA, et al.: Cognitive-be-
havioral stress management reduces serum cortisol by enhanc-
ing benefit finding among women being treated for early stage
breast cancer. P s y c ho s o mat i c M e di c i ne. 2000, 62 :304-3O8.
Greenberg MA: Cognitive processing of trauma: The role of in-
trusive thoughts and reappraisals. Journal ol Applied Social
P sy c ho lo gy. 1995, 2 5 : 1262-1296.
Horowitz MJ: Stess Response Syndromes (.2nd ed.). New York:
(18) Calhoun LG, Cann A, Tedeschi RG, McMillan J: A
correlational test of the relation between posttraumatic growth,
religion, and cognitive processing. Journal rf Truumatic Stress.
Pennebaker JW: Writing about emotional experience as a thera-
peutic process . P sychological Scie nce. 199'7, 8:162-166.
(20) Esterling BA, Antoni MH, Fletcher MA, Margulies S,
Schneiderman N: Emotional disclosure through writing or
speaking modulates latent Epstein-Barr virus antibody titers.
Joumal of Consulting and Clinical Psltchology. 1994.
(21) Pennebaker JW, KiecolrGlaser JK, Glaser R: Disclosure of
traumas and immune function: Health implications for psycho-
therapy. Journal ofConsulting and Clinical Psychology. 1988,
(22) Petrie KJ, Booth RI, Pennebaker JW, Davison KP, Thomas
MG: Disclosure of trauma and immune response to a hepatitis
B vaccination program. Journal of Consulting and Clinical
P src ho lo gy. 199 5, 6 3 :7 87 -7 92.
S, Larson J: Making sense of loss
Annals of Behavioral Medicine
(23) Pennebaker JW, Beall SK: Confronting a traumatic event: To-
ward an understanding of inhibition and disease. Journal of Ab'
rut rma I P s1'c
holo gy. 1986, 9 5 :27 4-28 L
(21) Kelley JE, Lumley MQ, Leisen JCC: Health effects of emo-
tional disclosure in rheumatoid arthritis patients. Health Ps-v-
cholo14t. 1991 , 16:331-340.
(25) Smyth JM, Stone AA. Hurewitz A, Kaell A: Effects of writing
about stressful experiences on symptom reduction in patients
with asthma or rheumatoid arthritis. Journal ot the Amerir:an
Medical Assot'iation. 1999.281: I 304-l 309.
(26) Pennebaker JW, Mayne TJ, Francis ME: Linguistic predictors
of adaptive bereavement. Journal of Personalim^ and Social
P s v t' tu t l o gr. 199-l, 7 2 :863-tl'l l .
(27) Zakowski SG, Valdimarsdottir HB, Bovbjerg DH. et al.: Predic-
tors ol'intrusive thoughts and avoidance in women with family
histories ol' breast cancer. Annals of Behaviorttl Medir:ine.
(28) Bovbjerg DH. Valdimarsdottir HB: lntcrventions firr healthy in-
dividuals at lamilial risk tbr cancer: Biobehavioral mechanisms
fbr health benefits. In Baum A, Anderson B (eds). Psvt:hosocial
I nt e rv e nt il t ns .fo r C anc e r. Washington, DC : American Psycho-
logical Association, 2001.
(29) Whiteside TL, Herberman RB: The role of natural killer cells in
human diseasc . C linical I mmunolo g r- and I mmunopatho lrt gl'.
(30) Strayer DR, Carter WA, Brodsky I: Familial occurrence of breast
cancer is associated with reduced natural killer cytotoxicity.
Breast Cancer Research and Treatment. 1986, 7: l8'l -192.
(31) Futterman AD, Kemeny ME, Shapiro D, Fahey JL: Immuno-
logical and physiological changes associated with induced pos-
itive and negative mood. Psycfutsomatic Medicine. 1994,
56:499-5 1 l.
(32) Fawzy Fl, Kemeny ME, Fawzy NW, et al.: A structured psychi-
atric intervention for cancer patients: Changes over time in im-
munological measures. Archives of General Psychiatry. 199Q,
(33) Kiecolt-Glaser JK, Glaser R, Williger D, et al.: Psychosocial
enhancement of immunocompetence in a geriatric population.
Health Psychoiogy. 1985,4:2541 .
(34) Carver CS, Scheier MF: On the Self-Regulation of Behavior.
New York: Cambridge University Press, 1998.
(35) Kemeny ME, Gruenewald T: Affect, cognition, the immune
system and health. In Mayer EA, Saper C (eds), The Biological
Basis Jor Mind Body Interactions: Progress in Brain Research
Series. Amsterdam: Elsevier Science B.V., 2000, 291-308.
(36) Cole SW: Negative physical health effects of psychological in-
hibition in disclosure study control groups. Manuscript submit-
ted for publication.
(371 Smyth JM: Written emotional expression: Effect sizes, our
come types, and moderating variables. Journal of Consulting
and C linical P s1'cholo gy. 1998, 66 : 17 4-184.
(38) Kemeny ME: The Life Goals Inventory. Unpublished manu-
(39) Kemeny ME, Reed G, Taylor SE, Visscher B, Fahey JL: Nega-
tive HIV-specilic expectancies predict immunological evidence
of HIV progression.
(40) Kemeny ME, Dickerson S, Gruenewald T, Bower J, Lu Q: Vali-
dation of a measure of life goals and priorities. Unpublished
(41) Radloff LS: The CES-D scale: A self-report depression scale
for research in the general population. Applied PsychologicaL
Measurement. 1977 , 1:386401.
Volume 25, Number 2,2003
ventory of Grief. In Zisook S (ed). Brops'vr:hoxtcial
Bereavement. Washington, DC: American Psychiatric Press.
(431 Kiecolt-Glaser JK, Glaser R: Methodological issues in behav-
ioral immunology research with humans. Brain, Behovior ond
lmnmnin. 1988.2:67 78.
(14) Stanton A[,, Danoff-Burg S, Sworowski I-A. et al.: Random-
ized. controlled trial of written emotional expression
fit finding in brcast cancer patients. Journnl tf (-linical Oncol-
ogv. 2002,20:4 | 60-4 1 68.
(45) Range LM. Kovac SH, Marion MS: Does writing about be-
reavement lessen griel' lbllowing sudden. unintentional death'l
Deeth Studies. 2000. 24: I l-5 134.
(46) Stroebe M, Stroebe W, Schut H.7ech E, van dcn Bout J: Does
disclosure of emotions facilitate recovery from bereavement'l
Evidence from two prospective studies. Journal o.l'Consulting
and C I inic a I P st't fu tlo 14t,. 2002. 70 ; l 69- I 78.
(47) Spiegel I), Classen C: Group Therapy .fttr Cnrccr Patient.r: A
Handbook of Psychosocial Care. New York:
Basic Books. 20ffi.
(48) Antoni MH, Lehman JM, Kilbourn KM, et i l.: Cognitive be-
havioral stress nranagement intervcntion decreases the preva-
lence o1' depression and enhances benetit flnding amclng
women under treatment for early stage breast cancer. Health
P s.vc fu t I o g,,t. 200 | . 2 0. 2V32.
TR. Zisook S. DeVaul R: The Texas Revised In-
Meaning and NK Cytotoxicity
(49) Friedman EM, Irwin MR: Modulation of immune cell function
by the autonomic nervous system. Pharmacologt' and Thera-
peutic s. 1997. 71 :27 -38.
(-50) McEwen BS, Biron CA, Brunson KW, et al.: The role of
adrenocorticoids as modulators of immune function in health
Re.sean'h rteylew.r. 1997. 2-1.19 133.
(51) Epel ES. McEwen BS. Ickovics JR: Embodying psychological
thriving: Physical thriving in responsc to strcss. Journal ol So-
c ial Ls,sut',s.
I 99u. .54:30 l-322.
(-52) Seligman MEP, Csikszentmihalyi M: Positive psychology
ISpecial issue j. Amencan Psvchoktgi,t/.
(-53) Ryan RM. Sheldon KM, Kasser T, Deci EL: All goals are not
created equal: An organismic perspective
and their regulation. In Gollwitzer PM, Bargh JA (cds), 7ftc
Ps,vchohtgt ttf At'tiort: Linking Cognition and Motit'ution trs Be-
halior. New Vrrk: Guilford, 1996.'7-26.
(54) Schmale A H: Giving up as a tinal common pathway to changes in
health. A dv a n t' e s in P sv c ho s o mat i c M e d i c i ne. | 9'7 2, 8: 2G-40.
(-55) Fleshner M. Laudenslager
ML. Sinrons L, Maier SL: Reduced
serum antibodies associated with social defbat in rats. Phvsiol-
ogv and Behavior. 1989.4J: l 1 83-1 1 U7.
(56) Stetanski V Ben-Eliyahu S: Social confiontation and tumor
metastasis in rats: Del'eat and beta-adrenergic mechanisms.
P h1: s io kt 14v
, and immune intcractions.
on the nature ofgoals
and B e hav io r 1996, 60 :27-l -282.