You are on page 1of 14

Evaluation of an Evolutionary Model of

Self-preservation and Self-Destruction


R. Michael Brown, PhD, Eric Dahlen, PhD, Cliff Mills, BA,
Jennifer Rick, BS, and Arturo Biblarz, PhD
According to decatanzaros mathematical model of self-preservation and self-destruction,
staying alive actually may reduce inclusive fitness for an individual who is low in reproduc-
tive potential and, at the same time, poses such a burden to close kin that it costs them
opportunities for reproduction. Predictions generated from this model were tested using
175 university students as subjects and variables constructed from a 164-item question-
naire. The criterion variables were separate measures of depression, hopelessness, and
suicide ideation and behavior. The predictor variables derived from the model were separate
measures of reproductive potential of the individual, the individuals perceived benefit or
cost to kin, and reproductive potential of the individuals kin. As predicted, there were
negative and significant bivariate correlations between each of the model-generated pre-
dictor variables and one or more of the criterion variables. Multiple regression analyses
showed that benefit to kin was the best predictor of both depression and hopelessness.
Discriminant analysis showed that reproductive potential of kin significantly differentiated
suicide attempters from nonattempters. Overall, our results support and extend decatan-
zaros model and empirical findings.

Drawing mainly upon Fishers (1950) con- y,= P,+ CbLPLJ-L


cept of reproductive value, later refined
by Hamilton (1966), and upon Hamiltons where Y,= the optimal degree of self-pres-
(1964) notion of inclusive fitness, deCa- ervation expressed by individual i
tanzaro (1986, 1987, 1991, 1992) has for- (the residual capacity to promote inclu-
mulated a simple mathematical model sive fitness);
designed to explain the evolution of self- pi = the remaining reproductive potential
preservative and self-destructive emo- of i;
tional and motivational mechanisms in P k = the remaining reproductive potential
humans and other social species (e.g., so- of each kinship member k ;
cial Hymenopterans). Stated simply, self- bk = a coefficient ofbenefit (positive values
preservation in an individual member of of b k ) or cost (negative values of bk) to
a social species will be influenced by the the reproduction of each k provided by
individuals capacity for sexual reproduc- the continued existence of i (-1 < b < 1);
tion, and by the individuals ability to facil- rk = the coefficient of genetic relatedness
itate (or interfere with) reproduction in of each k to i (sibling, parent, child=
kin. The model, in its original form (deCa- .5; grandparent, grandchild, nephew or
tanzaro, 1986), is represented mathemati- niece, aunt or uncle = .25; first cousin =
cally as follows: .125; etc.).

R. Michael Brown is Psychology Professor, Eric Dahlen is a former student, Cliff Mills is a former student,
Jennifer Rick is a former student, and Arturo Biblarz is SociologyProfessor, Department of Psychology, Pacific
Lutheran University, Tacoma, Washington.
Address correspondence to R. Michael Brown, Department of Psychology, Pacific Lutheran University,
Tacoma, WA 98447.
Portions of this research were presented at the seventh annual meeting of the Human Behavior and
Evolution Society, Santa Barabra, California, June, 1995.
Suicide and Life-Threatening Behavior, Vol. 29(1), Spring 1999
58 0 1999 The American Association of Suicidology
BROWN ET AL. 59

Self-preservation (YJ is assumed to be lighted perceived burdensomeness to kin


variable, ranging from "strong promotion and unstable heterosexual relationships
and defense of survival . . . to outright sui- as potentially important predictors of sui-
cidal tendencies" (decatanzaro, 1991, p. cide ideation. His 1984 survey of univer-
16).Where an individual lies on the distri- sity students and the general public was
bution of self-preserving behavior is influ- not a formal test of his model; rather, it
enced by that individual's expected repro- appears to have formed an empirical basis
duction during his or her remaining for generating the model. The more recent
lifetime ( p i ) , and by the remaining repro- (1995) study reports survey data collected
ductive value of the individual's kin (pJ, from groups representing the general pub-
weighted by degree of genetic relatedness lic, the elderly, psychiatric patients, crimi-
( r k )and by the benefit or cost to the repro- nal psychiatric patients, and homosexu-
ductive value of the individual's kin pro- als. Unfortunately, in 9 of the 11 groups
vided by the continued existence of the surveyed in this study, small sample sizes
individual (bk).It should be noted that p (ns range from 27 to 70) relative to the
represents reproductive potential calcu- number of predictor variables (30) raise
lated from a given point in time forward; serious questions about the reliability of
it does not include offspring already pro- the regression findings. Also, decatanz-
duced. According to decatanzaro p is de- aro has restricted his dependent variable
pendent on age, but will show variation in these studies to a composite measure of
even within a given age cohort due to fac- suicide ideation and behavior. While this
tors such as dominance, health, and ade- measure is clearly appropriate, there are
quacy of relationships with the opposite other components of self-destructive moti-
sex. vation and behavior that should also be
In his discussions decatanzaro has fo- affected by Y, variables, namely, depres-
cused mainly on the model's potential for sion and hopelessness. Finally, although
explaining the adaptive significance of decatanzaro raises the possibility of inter-
self-destructive behaviors (decatanzaro, actions among Y, variables, his theoretical
1986, 1987, 1992). Although positive Y, rationale fails to make them explicit, and
values that approach 0 will be associated his empirical work reports no attempt to
with progressively weaker attempts at assess such interactions. The present
self-preservation, negative '', values raise study addresses these limitations and pro-
the possibility of outright self-destruction. vides an independent and formal test of
Negative Y, values will occur only when the Yi model.
the impact of the individual's continued
existence on the reproduction of closely re-
lated kin is so devastating that it overpow-
ers his or her own residual reproductive QUESTIONS AND HYPOTHESES
potential; that is, when bk is negative and
the absolute value of SkPkrk exceeds the In his 1984 study, decatanzaro does not
value of p i . It is assumed that, at some report assessing the effects of perceived
level, the individual must be able to per- benefithost to kin (b k )independently of
ceive predictive contingencies between be- reproductive potential of kin ( p , ) . In fact,
havior and reproductive outcomes, and his measure of the former appears to be
that this perception can influence gene ex- weighted by the latter, and by degree of
pression. genetic relatedness of each kin member
DeCatanzaro (1981, 1991) has mar- summed over all kin (Xbkpkrk). Whether
shaled arguments and evidence from a va- this same procedure was followed in the
riety of sources that make his model plau- 1995 study is not clear, but there is no a
sible. His own published survey data priori reason why the separate effects of
(decatanzaro, 1984, 1995) have high- each variable (bkand Zpkrk)could not, or
60 SUICIDE AND LIFE-THREATENING BEHAVIOR

should not, be assessed. It is conceivable produce negative YLvalues and possible


that a sense of onesperceived contribution self-destructive behavior. If individual re-
or burden to family is sufficient to drive productive potential is extremely high and
self-preserving or self-destructive cogni- bk is negative, then the same small in-
tive and behavioral algorithms, indepen- creases in the absolute value of bkwill not
dently of the number, reproductive status, have the same effect; much larger in-
and genetic relatedness of kin who stand creases in bk will be required in order to
to gain or lose by virtue of the individuals produce negative r, values and possible
continued existence (assuming the num- self-destructive behavior.
ber of kin is greater than 0). Similarly, 2. Reproductive potential of kin
reproductive potential of kin may affect weighted by degree of genetic relatedness
self-preserving and self-destructive be- (Pkrk) should interact with benefitkost to
haviors independently of ones sense of kin (bk),as indicated by the multiplicative
contribution or burden to kin. Moreover, relationship between the two terms. Spe-
there are serious questions that can be cifically, the effect of benefithost to kin on
raised about the way in which CbkPkrk af- Y, will not be the same at all levels of Pkrk.
fects Y,in the model. When an individuals Assume bk is negative (a perceived cost t o
continued existence represents a contribu- kin associated with the continued exis-
tion to kin, bk will be a positive number. tence of the individual) and pL2 0. An in-
In this case Y ,will increase as the number crease in the absolute value of bk from,
of kin, their reproductive potential, and say, .2 to .4will result in a proportionately
their degree of genetic relatedness to the steeper drop in Y L values when Pkrk is high,
individual increase. However, when an in- compared to when it is low.
dividuals continued existence represents Our first step in evaluating the model
a burden to kin, bkwill be a negative num- and testing our hypotheses was to com-
ber. In this case, Y t will decrease as the pute and examine all possible correlations
number of kin, their reproductive poten- among our suicide-related measures, Y,
tial, and their degree of genetic related- variables, and other relevant variables
ness to the individual increase. DeCatanz- that have proven t o be reliable correlates
aro does not present a rationale for the of suicide-related measures in our own
models predicted paradoxical effect of Pkrk work (Biblarz, Brown, Biblarz, Pilgrim, &
on Y,. This is an unfortunate omission, Baldree, 1991) and in the work of others
and a further reason for evaluating the (e.g., Strang & Orlofsky, 1990; Topol &
contributions of bk and CPkrk separately. Reznikoff, 1982). The Y t variables in-
The results of such an evaluation might cluded a measure of individual reproduc-
argue for qualifying decatanzaros mathe- tive potential, perceived benefit to kin,
matical model in important ways. and reproductive potential of kin. The
Although decatanzaro (1992) acknowl- other relevant variables were perceived
edges the possibility of complex interac- quality of relationships with parents, per-
tions among Y, variables, to this point he ceived quality of relationships with
appears not to have discussed or evaluated friends, and measures of locus of control.
the interactions that are inherent in his Based on decatanzaros model, as well as
model: previous research, we expected our sui-
1. Individual reproductive potential (p,) cide-related measures to correlate nega-
should interact with perceived benefit/ tively with individual reproductive poten-
cost to kin (bk),even though the additive tial, perceived benefit to kin, quality of
relationship specified in the model seems relationships with parents and friends,
to suggest otherwise. For example, if indi- and internal locus of control, and posi-
vidual reproductive potential is extremely tively with external locus of control.
low and bkis negative, then even negligible Our next step was to regress our suicide-
increases in the absolute value of bk will related measures on those variables that
BROWN ET AL. 61

had proven to be significant correlates of of the opposite sex. The benefit-to-kin


these measures. Finally, we conducted a items asked subjects to estimate the fol-
second regression analysis which focused lowing: burdensomeness to family, impact
exclusively on Y t variables as predictors of their death on family, their contribu-
of our suicide-related measures. Our pri- tions and assistance to family, and their
mary interest in this final analysis was to familys dependence on them for emotional
evaluate the interactions that appear to or social support. Reproductive potential
be specified by the model. of kin was a composite measure reflecting
each biological relatives expected health,
fertility, and degree of genetic relatedness
METHOD to the subject, summed over all kin. The
reproductive-potential-of-kin items asked
Subjects subjects to indicate how many of their bio-
logical relatives were living for each of the
One-hundred and eighty one undergradu- following levels of kinship: parents, sib-
ate students enrolled in psychology lings, children; grandparents, aunts and
courses at Pacific Lutheran University uncles, nieces and nephews; great grand-
volunteered to participate in the study in parents, great aunts and great uncles,
return for course credit. The students were first cousins; second cousins. A parallel set
asked to complete a questionnaire and, of of items asked subjects to rate the health
those returned, 175 were usable (130 status of each kin member. Fertility rat-
women and 45 men). Subjects in the sam- ings were taken from U.S. Department of
ple were predominantly White (go%),reli- Commerce (1991) tables, and based on es-
gious (84% expressed a religious prefer- timated age cohorts of the subjects kin.
ence), from an urban environment (68%), Genetic relatedness coefficients were as-
middle class (59% with annual family in- signed based on level ofkinship. Other rele-
comes 2 $40,000), first- or second-year vant predictor variables included quality of
students (64%), in late adolescence or relationships with parents (10 items),
early adulthood (89% ranged in age from quality of relationships with friends (7
17 to 23 years), single (81%),and hetero- items), and Levensons (1981) locus of
sexual (98%). control inventory (24 items that yield
three scales: Internal, Powerful Others,
Questionnaire Chance). The remaining questionnaire
items focused on demographic informa-
All subjects completed a 164-item ques- tion, sexual behavior, and control issues.
tionnaire that included items designed to
assess affective states, cognitions, and be- Procedure
haviors of potential causal relevance to
suicide: depression (3 items), Becks Hope- Potential subjects were told that (1)the
lessness Scale (20 items; Beck, Weissman, purpose of the study was to assess rela-
Lester, & Trexler, 19741, and a composite tionships among peoples attitudes, emo-
measure of suicide ideation and behavior tions, and motivations; (2) the question-
(4 items). The Y ipredictor variables in- naire would require approximately 1hour
cluded individual reproductive potential or more to complete; (3) they would have
(4 items), perceived benefit to kin (6 4 days (which coincided with a midterm
items), and reproductive potential of kin break) to complete the questionnaire; (4)
(14 items). The individual reproductive po- they should feel free to omit questions that
tential questions asked subjects to rate the seem too difficult or to withdraw from par-
following: importance of sex, hurtfulness ticipation at any time; and (5) they were
of romantic relationships, satisfaction to take the questionnaire seriously and re-
with partners in romantic relationships, frain from talking to other participants
and perceived attractiveness to members about the questions. Those who volun-
62 SUICIDE AND LIFE-THREATENING BEHAVIOR

teered were assigned random identifica- dictors. Higher-order interaction terms


tion numbers and assured confidentiality. were dropped from the equation systemat-
ically following the step-down procedures
for testing interactions recommended by
Design and Analysis Aiken and West (1991).

Correlations. The design was completely


correlational, and major analyses in- RESULTS
cluded bivariate correlations (Pearson r )
computed between each of the continuous Correlations with Suicide-
criterion variables (depression, hopeless- Related Measures
ness, suicide ideation and behavior) and
each of the eight predictor variables (indi- Each of the three 'Pivariables correlated
vidual reproductive potential, benefit to significantly' with one or more of the sui-
kin, reproductive potential of kin, quality cide-related measures, and the significant
of relationships with parents, quality of correlations were all in the direction pre-
relationships with friends, internal locus dicted by decatanzaro's model (see Table
of control, belief in powerful others locus 1).Subjects with lower individual repro-
of control, belief in chance locus of control). ductive potential scores were more likely
than other subjects to report higher levels
First Regression Analysis. The purpose of of depression and suicide ideation and
the first regression analysis was to evalu- behavior. Subjects with fewer fertile,
ate the contribution of each of the Yivari- healthy, and closely related kin were more
ables alongside other predictor variables likely than others to report higher levels
that have proven to be reliable correlates of suicide ideation and behavior. Subjects
of hopelessness in previous research. Sep- with lower benefit to kin scores were more
arate stepwise multiple regressions of de- likely than others to report higher levels
pression, hopelessness, and suicide ide- of depression, hopelessness, and suicide
ation and behavior were computed. For ideation and behavior.
each of the three regressions, only those For the other relevant variables, sub-
variables that had proven to be significant jects scoring lower on quality of relation-
(two-tailed)correlates ofthe criterion vari- ships with parents and with friends were
able being regressed were entered as pre- more likely than others to score high on
dictors in the equation. Finally, a discrimi- all three suicide-related measures; sub-
nant analysis was computed to determine jects scoring lower on internal and higher
which of the eight predictor variables reli-
ably differentiated self-reported suicide
attempters from nonattempters. 'Reported correlations and regression findings were
calculated on the entire sample. Because marital sta-
tus could alter subjects' responses, especially to the
Second Regression Analysis. The second individual reproductive potential items, we also ana-
regression analysis focused exclusively on lyzed data with married subjects excluded from the
the potential contributions of Y Lvariables sample. Regression findings were identical (direction
and significance of predictors) to those of the entire
to our criterion measures. Of particular sample. Four of the correlational findings differed
interest was the assessment of hypothe- from those of the entire sample: the correlation be-
sized interactions among the Y:variables. tween individual reproductive potential and suicide
Separate regressions were performed on ideation and behavior only approached significance,
r(133)= -.17,p c .057;the correlation between rela-
each of the suicide-related measures, with tionships with friends and suicide ideation and be-
the effects of the remaining two criterion havior was not significant, r( 158)= -. 12;the intercor-
variables removed before proceeding relation between individual reproductive potential
and relationships with parents was not significant,
further. Subsequently, each of the Y ,vari- r(132)= .13;the intercorrelation between individual
ables, and all possible interactions involv- rerpoductive potential and relationships with friends
ing these variables, were entered as pre- was not significant, r(132)= .14.
BROWN ET AL. 63

TABLE 1
Correlates of Suicide-Related Measures
Suicide variables
Ideation and
Correlates Depression Hopelessness behavior
Individual reproductive potential -.29*** -.04 -. 19*
Benefit to kin -.44*** - 5o*w -.28***
Reproductive potential of kin -.08 .01 -.20**
Parents -.47*** -.27*** -.38***
Friends -.39*** -.38*"* -.18*
Internal locus of control -.21** -.14 -.08
Powerful Others locus of control .23** -.03 .22**
Chance locus of control .35*** .08 .28***
-
Note. All probability levels are two-tailed.
*p < .05; **p < .01;***p < ,001.

on external locus of control measures were tions of one or more of the Y Lvariables
more likely than others to score high on one when compared with other relevant and
or more of the suicide-related measures. traditionally potent predictors (see Table
2). Compared to the other seven predictor
Intercorrelations variables, benefit to kin was the strongest
predictor of depression, accounting for
As expected, the three suicide-related nearly 25%of the variance. Moreover, ben-
measures were positively correlated with efit to kin was the only significant pre-
one another. For depression and hopeless- dictor of hopelessness, accounting for just
ness, 4171) = .20, p < .008;for depression under 20%of the variance. Individual re-
and suicide ideation and behavior, r(170) productive potential was a significant pre-
= .45, p < .001;and for hopelessness and dictor of depression, and reproductive po-
suicide ideation and behavior, r(172) = .13, tential of kin was a significant predictor
p < .079. As for the Yi variables, subjects of suicide ideation and behavior. The other
with lower individual reproductive poten- significant predictors were quality of rela-
tial scores were more likely than others tionships with parents and the chance lo-
to report poor relationships with parents, cus of control measure. Each of these vari-
r(143) = .26, p < .001, and with friends ables reliably predicted both depression
r( 143) = .22, p < .008. Similarly, subjects and suicide ideation and behavior. Results
with lower benefit-to-kin scores reported of the discriminant analysis indicated that
poorer relationships with parents, 4171) six of the eight predictor variables reliably
= 5 7 , p < .001, and with friends r(171) = differentiated suicide attempters from
5 2 , p < .001. Finally, subjects with low nonattempters (see Table 3). Reproductive
benefit-to-kin scores tended to score low potential of kin was the most potent of the
on internal, and high on external, locus significant predictors based on an exami-
of control: Internal, 4172) = .21, p < .005; nation of correlations between the
Powerful Others, 4172) = -.16, p < .035; discriminant scores and the predictor (in-
Chance, r(172) = -.23, p < .002. Reproduc- dependent) variables. Compared to nonat-
tive potential of kin showed no significant tempters, attempters had significantly
intercorrelations. fewer close kin with high expected fertility
and good health. Interestingly, compared
First Regression Analysis to nonattempters, attempters had signifi-
cantly higher individual reproductive po-
Results of multiple regressions of each cri- tential scores. The two groups did not dif-
terion variable underscore the contribu- fer significantly on the benefit-to-kin
64 SUICIDE AND LIFE-THREATENING BEHAVIOR

TABLE 2
First Regression Analysis: Stepwise Multiple Regressions Comparing the Effects of Y,
Variables with Those of Other Relevant Predictor Variables
Suicide variable Step F Adj. R2 Predictor P t
Depression 1 46.28*** .25 Benefit to kin -.29 -3.61***
2 36.93*** .34 Chance .29 4.23***
3 29.16*** .38 Parents -.21 -2.58*
Individual repro-
ductive
4 23.66*** .39 potential -. 15 -2.18*
Hopelessness 1 42.50*** .20 Benefit to kin -.45 -6.52***
Ideation and
behavior 1 19.02*** .ll Parents -.29 -3.8 1***
2 16.90*** .19 Chance .26 3.40***
Reproductive
3 13.65*** .21 potential of kin -. 18 -2.44*
Note. All probability levels are two-tailed.
*p < .05;**p < .01;***p < ,001.

variable, though the group means were to kin, or reproductive potential of kin x
in the expected direction, with attempters benefit to kin. However, for each of the
showing lower scores ( M = 23.625) than suicide-related measures, at least one of
nonattempters ( M = 24.632). the Y ivariables attained significance as
a predictor (see Table 4).As in the first
regression analysis, benefit to kin was a
Second Regression Analysis significant predictor of both depression
Results of multiple regressions of each sui- and hopelessness. Individual reproductive
cide-related measure using only the Yz potential also predicted depression signif-
variables as predictors showed no support icantly. Reproductive potential of kin was
for the two hypothesized interactions: in- the only significant predictor of suicide
dividual reproductive potential x benefit ideation and behavior.

TABLE 3
Discriminant Analysis: Differentiation of Suicide Attempters from Nonattempters
~~~ ~ ~

Means for Discriminant


significant suicide Means for non- correlations
independent attempters attempters Wilkss with
Step variables ( n = 8) ( n = 133) Lambda predictors
Reproductive potential
1 of kin 3.14 3.97 .95** .71
2 Parents 32.88 37.61 .94* .37
3 Friends 31.12 30.38 .93* -.13
4 Chance 14.50 11.82 .92* -.34
Individual reproduc-
5 tive potential 7.62 7.38 .91* -.08
6 Powerful others 13.00 12.71 .91* -.04
Note. All probability levels are two-tailed.
*p < .05;**p < .01;***p < .001.
BROWN ET AL. 65

TABLE 4
Second Regression Analysis: Multiple Regressions Evaluating the Effects of the
Variables and Hypothesized Interactions among These Variables
Suicide variable F Adj. R2 Predictor s t
Depression 10.71*** .35 Benefit to kin -.38 -4.50***
Individual reproductive
potential -.17 -2.34*
Hopelessness 7.55*** .27 Benefit to kin -.55 -6.56***
Reproductive potential
Ideation and behavior 6.58*** .24 of kin -.19 -2.50"
Note. All probability levels a r e two-tailed.
*p < .05; **p < .01;***p < ,001.

DISCUSSION to suicide ideation, hopelessness has been


shown in prospective studies to be a far
Perceived Benefits and better predictor of suicide (Beck, Steer, Ko-
Costs to Kin vacs, & Garrison, 1985), and is arguably
Our benefit-to-kin variable correlated sig- the best predictor overall of suicide com-
nificantly with all three suicide-related pletions in clinical populations (Steer, Ku-
measures-depression, hopelessness, and mar, & Beck, 1993). The fact that benefit
suicide ideation and behavior. In addition, to kin is related inversely not only to sui-
compared to the seven other predictor cide ideation but also to hopelessness
variables, benefit to kin explained more raises the odds that we are tapping into a
of the variance in depression and hopeless- motivational path that can eventuate in
ness. Other researchers operating from actual suicide, not just thoughts about sui-
very different theoretical perspectives cide or parasuicide.
have also reported evidence for a relation- Given the correlational design of our
ship between perceived burdensomeness study, we can only speculate about the pos-
t o others and self-destructive behavior sible causal role of perceived benefits and
(Motto & Bostrom, 1990; Woznica & Sha- costs to kin. It may well be that, in some
piro, 1990). instances, perceiving oneself to be a bur-
The benefit-to-kin findings are quite den to kin elevates depression and hope-
consistent with decatanzaro's model lessness and activates a self-destructive
(19861, and provide a needed extension of motivational system. However, it is also
his empirical work linking burdensome- possible that the causal arrow points in
ness to kin with suicide ideation (deCa- another direction; specifically, high levels
tanzaro, 1984, 1995). Suicide ideation is of depression may cause a person to think
clearly an appropriate measure of self-de- of himself or herself as a burden. In short,
structive cognitive content. However, the perceived burdensomeness may be an ef-
extent of ideation in nonclinical adolescent fect rather than a cause of depression. This
and young adult populations (e.g., 44% in could explain why benefit to kin was a
Rudd, 1989; 63% in Smith & Crawford, potent predictor of both depression and
1986; 61% in Strang & Orlofsky, 1990) hopelessness in our study. However, there
raises serious questions about its ability are other lines of evidence that oppose this
to predict actual suicide, and its role in the interpretation, and which suggest that
motivational sequences that eventuate in burdensomeness can have a causal influ-
suicide. Although most suicides probably ence that is not contaminated by subjects'
have a history of ideation, the vast major- affective states:
ity of even serious ideators never attempt, 1.Extremely high benefit-to-kin scores
let alone complete, suicide. And, compared and extremely low depression scores were
66 SUICIDE AND LIFE-THREATENING BEHAVIOR

rare in our study. The mean total benefit- tive behavior. Maris (1992) has noted that
to-kin score was high (24.75 out of a possi- nonfatal suicide attempts may confer on
ble 301, while the mean total depression the attempter benefits as well as costs.
score was low (3.905 out of a possible 16). Benefits include attention from family
These data do not support the conclusion members, friends, and professionals, and
that depression was responsible for rela- short-term elevation of affect. Such bene-
tively low benefit-to-kin scores, but they fits may have reduced feelings of burden-
do not rule out the possibility that changes someness in our attempters. It is also pos-
within the observed range of depression sible that some of our attempters were not
ratings may have influenced the benefit- intent on completing suicide but, rather,
to-kin responses in some way. were motivated by the increased attention
2. When regressing a given dependent they anticipated receiving as a result of
variable (e.g., hopelessness), we partialed their attempt (Kreitman, 1977). In an
out the effects of the other dependent vari- effort to clarify this matter we are now
ables (depression, suicide ideation and be- conducting studies that utilize a revised
havior). Therefore, the observed signifi- questionnaire designed to rate the lethal-
cant regression of hopelessness on benefit ity of any reported suicide attempts.
to kin indicates that benefit to kin made
a unique contribution to the variance in Reproductive Potential of Kin
hopelessness. This argues against the con-
clusion that subjects depression ratings Our data also show that reproductive po-
can explain the predictive relationship be- tential of kin may be an important vari-
tween benefit to kin and hopelessness. able in its own right. It was a significant
3. In a recent study, we (Brown, Brown, inverse predictor of suicide ideation and
Johnson, & Lampert, 1997) manipulated behavior, whether entered alongside all of
experimentally the degree to which a tar- our other predictors, or only with Yi vari-
get person in a scenario represented a ables. Moreover, it did the best job of all
burden to kin, while keeping the target our predictors differentiating suicide at-
persons described level of depression con- tempters from nonattempters, revealing
stant across all conditions of the experi- that attempters had significantly fewer
ment. As predicted, we found that univer- fertile, healthy, and genetically related kin
sity students judged the target person to than were reported by nonattempters. The
be significantly more unhappy, hopeless, direction of this relationship is consistent
and suicidal if, in the scenario, the target with decatanzaros model as long as bk is
was depicted as being a burden (compared assumed to be the same (and not a nega-
to a benefit) to kin. Furthermore, subjects tive value) for both attempters and nonat-
depression levels, as measured by the Beck tempters. Our data provide no reason to
Depression Inventory (Beck, 1967),did not doubt these assumptions. Attempters and
predict their ratings of the target persons nonattempters did not differ significantly
unhappiness, hopelessness, or suicidal in- on benefit to kin scores, and the distribu-
tent. These data show clearly that per- tion of benefit to kin scores was skewed
ceived burdensomeness can have a causal negatively.
influence on ratings of depression, hope- Our reproductive-potential-of-kin vari-
lessness, and suicidal intent in others. able is likely not a proxy for size of an
At first glance, it is perhaps surprising individuals immediate family (parents
that benefit to kin, so strongly associated and siblings), even though family size is
with our continuous suicide-related mea- reflected in calculating reproductive po-
sures, failed to differentiate suicide at- tential of kin. Our questionnaire mea-
tempters from nonattempters. However, sured family size in a separate item, and
this failure may have occurred in part the pattern of correlations between it and
because of the difficulties inherent in tak- other variables was not consistent with
ing retrospective measures of self-destruc- the reproductive-potential-of-kin pattern.
BROWN ET AL. 67

Although family size correlated positively In the discriminant analysis, individual


and significantly with reproductive poten- reproductive potential was one of six
tial of kin, it did not correlate significantly predictor variables that reliably differen-
with depression, hopelessness, or suicide tiated suicide attempters from nonat-
ideation and behavior. It should be noted, tempters. Suicide attempters had higher
however, that family size, as measured by individual reproductive potential scores
number of offspring, has been found to be than nonattempters, a finding that seems
an inverse predictor of attempted suicides at odds with the correlation and regression
(Calzeroni, Conte, Pennati, & Vita, 1990; results from all 175 subjects. With so few
Koller & Costanos, 1968; Kreitman, 1977; attempters, it is difficult to know whether
Lipe, Schulz, & Bird, 1993), and of com- this finding reflects sampling error, or a
pleted suicides (Hoyer 8z Lund, 1993; reliable difference between attempters
Humphery, 1977; Iga, Yamamoto, No- and nonattempters. However, the same
guchi, & Koshinaga, 1978). discriminant analysis showed a similar re-
versal of correlational findings for quality
of relationships with friends: Attempters
Individual Reproductive were differentiated from nonattempters
Potential by their better relationships with friends,
Our measure of individual reproductive but correlations for the entire sample
potential was inversely related to depres- clearly indicate an inverse relation be-
sion, consistent with decatanzaros model. tween relationships with friends and each
In both regression analyses, individual re- of our three continuous suicide-related
productive potential was a significant measures. One possible explanation of
source of variance in predicting depres- these findings was presented earlier, in
sion. Naturally, the same cautions that connection with the failure of the benefit-
were raised over interpretation of the ben- to-kin variable to reliably differentiate at-
efit-to-kin findings also apply to the indi- tempters from nonattempters. Ratings of
vidual reproductive potential data. We do sex, romantic relationships, and personal
not know to what extent subjects affective attractiveness may have improved as a
states may have colored their assessments consequence of the suicide attempt. In or-
of relationships with, and attractiveness der to test this hypothesis, prospective
to, members of the opposite sex. However, < .025. Subjecting our second, ostensibly more objec-
as noted before, subjects levels of depres- tive, resource measure to the same analyses yielded
similar results. The partial correlation between in-
sion were low overall, and this suggests come-weighted occupational status and suicide ide-
that depression was not likely responsible ation and behavior was significant for males, r(26)
for relatively low scores on the individual = -.40,p < ,037, but was negligible for females.
reproductive potential items.2 Neither resource measure appears to have been in-
fluenced significantly by our measure of subjects de-
pression or scores on Becks Hopelessness Scale, yet
each resource measure correlated with suicide ide-
More objective indicators of individual reproductive ation and behavior in ways expected by deCatan-
potential, such as age and resources, would be less zaros evolutionary model. The fact that the correla-
susceptible to contamination by subjects affective tions were observed for males only is consistent with
states. We collected age data, but the range was too an empirically defensible evolutionary perspective-
restricted to expect age-related differences in repro- sexual selection theory-which predicts that women
ductive potential, actual or perceived. We also col- will value men mainly for their ability to provide
lected resource data yielding two measures: subject resources, while men will value women mainly for
ratings of family economic status relative to others, their physical attractiveness, a presumed sign of
and reported occupational status of parents weighted health and reproductive capacity (Archer, 1996).Not
by ratio of reported family income to reported family surprisingly, in our study male subjects ratings of
size. High ratings of relative family economic status their physical attractiveness correlated little, if a t
were accompanied by low suicide ideation and behav- all, with the dependent measures. On the other hand,
ior, even with the effects of depression and hopeless- female subjects attractiveness ratings correlated
ness partialed out, r(152) = -.19, p < .019. When the significantly with both depression, r(130) = -.26, p <
partial correlation was recalculated for each gender, ,003, and with suicide ideation and behavior, r(129)
only males showed a signficant effect, r(35) = -.37,p = -.25, p < ,005.
68 SUICIDE AND LIFE-THREATENING BEHAVIOR

studies are needed in which measures of tional substrates of self-preservation lie on


individual reproductive potential are a continuum. Therefore, a priori, it seems
taken prior to, as well as after, nonfatal reasonable to expect that such differences
suicide attempts. would be correlated with fitness variables
regardless of which portion of the self-
preservation continuum is sampled. In
Interactions fact. our data indicate that the exDected
correlations hold for subjects who score
Our hypothesized interactions-indivi-
high on self-preservation (i.e., they show
dual reproductive potential x benefit to
low levels of depression and hopelessness
kin, and reproductive potential of kin x overall). The correlations also appear to
benefit to kin-were not confirmed for any hold for certain populations at risk for self-
one of the suicide-related measures. Per-
destruction (decatanzaro, 1995). Only
haps this is because of restricted variation
further empirical work of a prospective na-
in the benefit-to-kin scores. Even though ture can determine whether there is a con-
these scores ranged from 8 to 30 (possible
nection between fitness and self-preserva-
range: 0-30),the distribution was skewed
tion motivation in suicide completers.
negatively (67% of the scores were above Because reproductive value (expected
the mean). Therefore, interactions involv- future reproduction) peaks during adoles-
ing benefit to kin may have been obscured
cence (Thornhill & Thornhill, 1983) and
by this ceiling effect. It will be necessary
declines sharply with age, and because re-
to explore the full range of the benefit-to-
productive strategies are expected to vary
kin variable, perhaps with clinical sam-
with sex (Trivers, 1972), we might expect
ples, in order to properly evaluate the hy-
differential effects of fitness variables on
pothesized interactions.
self-destructive motivation and behavior
depending on age and sex. In his recent
Qualifications Concerning work with six (nonuniversity) samples
the Sample representing the general public, deCa-
tanzaro (1995) has found that the strong-
Naturally, our generalizations are limited est fitness correlates of suicide ideation
to a rather restricted population-young do, in fact, change depending on age and
psychology undergraduates attending a sex of the sample. For example, self-re-
private (Lutheran) university in a major ported health and financial status failed
metropolitan area in the Pacific North- to correlate significantly with suicide ide-
west region of the United States. Further- ation in men or women comprising the
more, the sample represents a population youngest samples (18-30 years of age).
that is disproportionately female, White, However, these two variables were signifi-
single, middle- to upper-middle class, at cant inverse correlates of ideation for both
the peak of fertility and, overall, not suf- men and women in the older samples
fering from clinical depression or other se- (31-50 years, and 51+ years). Further-
rious psychopathological disorders. Thus, more, one or more measures of sexual ac-
we are limited in what we can say about tivity correlated significantly and in-
suicide completers, and about individuals versely with ideation in each of the male
with demographic characteristics (e.g.,sex samples, regardless of age, but only in the
and age) that differ from those of our youngest female sample. It should be em-
sample. phasized, however, that perceived burden-
Even though our sample precludes gen- someness toward kin correlated signifi-
eralizing our findings directly to individu- cantly and positively with suicide ideation
als who complete suicide, the model we in all six samples, in four of five additional
are testing assumes that individual differ- atr i s k samples, and in a university sam-
ences in evolved emotional and motiva- ple tested previously (decatanzaro, 1984).
BROWN ET AL. 69

This consistent pattern is strengthened by disorders, conduct disorders). The threats


the independent benefit-to-kin findings and reactions by themselves are not suffi-
from the present study and from other re- cient to drive suicidal behavior; they may
cent investigations using different sam- in fact produce a wide variety of outcomes.
ples and methodologies (Brown & Melver, What would make a person truly suicidal,
1996; Brown et al., 1997). according to the evolutionary model out-
lined in the present study, is the (uncon-
scious or conscious) calculation^' that the
Qualifications Concerning the potential genetic costs of staying alive out-
Evolutionary Model weigh the gains. Extrapolating from the
model, this result would occur only when
DeCatanzaros model was designed with expected individual reproduction is ex-
ancestral conditions in mind, when hu- tremely low, and feelings of burdensome-
mans lived in relatively small and geneti- ness to close kin are high. Perhaps it is
cally interrelated groups. According to the the negative outcome of a fitness analysis
model, suicide would have been geneti- such as this that triggers a state of hope-
cally advantageous only when there was a lessness.
conjunction of low individual reproductive Our findings and decatanzaros under-
potential and burdensomeness toward score the importance of understanding in-
kin. However, it is important to note that terpersonal relationships within the fam-
our present environment may differ dra- ily in treating depression, hopelessness,
matically from the environment of evolu- and self-destructive ideation and behav-
tionary adaptedness, as decatanzaro ior. It appears that self-reported burden-
(1995) and others (e.g., Tooby & Cosmides, someness to family members is a salient
1992)have noted. There exist today meth- correlate of suicide ideation and behavior
ods of self-destruction, threats to family for both genders and all age levels sampled
and society, and modes of communication thus far, whether university students, the
and isolation that were not present in the general public, or a variety of at risk popu-
ancestral environment. Consequently, we
lations. It may be useful to evaluate and
should expect that there may now be alter-
attempt to modify perceptions of burden-
native paths to suicide. Some instances of someness in depressed or otherwise at-
suicide in the contemporary environment
risk individuals. Cognitive-behavioral
may be triggered by experimentation, me- techniques might be used to help such in-
dia stories, ineffective strategies for cop-
dividuals identify fallacious conclusions of
ing with evolutionarily novel stress, or burdensomeness, attend to and rehearse
psychopathological processes.
instances of concrete contributions made
to family members, and take advantage of
Clinical Implications opportunities for making contributions to
the community or society. Relatives of a
An evolutionary approach to suicide offers depressed individual may communicate
new and potentially fruitful ways of un- messages of burdensomeness directly, or
derstanding the etiology of suicide. The through emotional neglect. Therefore, it
almost bewildering array of conditions may be especially important to include the
(Felner, Adan, & Silverman, 1992) linked depressed persons siblings, parents, off-
to suicide may have in common the fact spring, or other family members in treat-
that they are actual or perceived threats ment plans. In this way, family members
to inclusive fitness (e.g., parent-child con- might learn to recognize their roles in com-
flict, death of a family member, breakup of municating overt or subtle messages of
a romantic relationship, terminal illness, burdensomeness to the depressed relative,
public humiliation), or reactions to such and to appreciate the potentially lethal
threats (e.g., substance abuse, affective impact of such communications.
70 SUICIDE AND LIFE-THREATENING BEHAVIOR

CONCLUSIONS that have occurred in the social ecology of


our species.
Freuds use of the death instinct to ex-
plain self-destructive behavior met with
skepticism and denial among analysts and REFERENCES
biologists alike almost as soon as the con-
Aiken, L. S., &West, S. G. (1991).Multiple regression:
cept was introduced in the early 1900s Testing and interpreting interactions. London:
(Jones, 1957).In the years hence, concep- Sage.
tualizations linking suicide to biology typi- Archer, J. (1996). Sex differences in social behavior:
cally have stated or implied that genes Are the social role and evolutionary explanations
compatible? American Psychologist, 51, 909-917.
may be responsible for affective disorders Beck, A. T. (1967). Depression: Clinical, experimental
that can give rise to suicide, but not for and theoretical aspects. New York Harper & Row.
suicide directly. The idea that at least Beck, A. T., Weissman, A., Lester, D., & Trexler, L.
some aspects of self-destructive motiva- (1974). The measurement of pessimism: The Hope-
lessness Scale. Journal of Constitution and Clini-
tion may be part of our evolutionary heri- cal Psychology, 42, 861-865.
tage has received little attention, let alone Beck, A. J., Steer, R. A., Kovacs, M., & Garrison, B.
support, in spite of suggestive evidence (1985). Hopelessness and eventual suicide: A 10-
year prospective study of patients hospitalized
from behavioral genetic studies (e.g., with suicide ideation. American Journal of Psychi-
Schulsinger, Kety, Rosenthal, & Wender, atry, 142, 559-563.
1979), the availability of kin selection as Biblarz, A., Brown, R. M., Biblarz, D., Pilgrim, M., &
an explanatory mechanism (Hamilton, Baldree, B. F. (1991). Mediainfluence on attitudes
toward suicide. Suicide and Life-Threatening Be-
1964; Maynard Smith, 1964),and the for- havior, 21, 374-384.
mulation of inclusive fitness models of ani- Brown, R. M., Brown, S. L., Johnson, A., & Lampert,
mal suicide (McAllister & Roitberg, 1987; A. (1997, June). Effects of fitness cues on ratings
of anothers self-destructive tendencies. Paper pre-
OConnor, 1978; Poulin, 1992). The lone sented a t the meeting of the Human Behavior and
voice in the wilderness belongs to deCa- Evolution Society, Tucson, AZ.
tanzaro, whose arguments for the evolu- Brown, R. M., & Melver, K. (1996, June). Evolution-
tion of self-destructive behavior under ary analysis ofsuicide ideation and behavior. Paper
presented at the meeting of the Human Behavior
well-defined circumstances seem plausi- and Evolution Society, Evanston, IL.
ble, and whose mathematical model has Calzeroni, A., Conte, G., Pennati, A,, & Vita, A.
provided the impetus for the present study. (1990). Celibacy and fertility rates in patients with
major affective disorders: The relevance of delu-
Results of our study support and extend sional symptoms and suicidal behaviour. Acta Psy-
decatanzaros theoretical and empirical chiatrica Scandinavica, 82, 309-310.
contributions (decatanzaro, 1984, 1986, decatanzaro, D. (1981). Suicide and self-damaging
1995). While his investigations focused behavior: A sociobiological perspective. New York
Academic Press.
only on suicide ideation and suicide at- decatanzaro, D. (1984). Suicidal ideation and the
tempts as dependent measures, our study residual capacity to promote inclusivefitness: A
has shown that one or more of the three survey. Suicide and Life-Threatening Behavior, 14,
75-87.
Y ivariables predict levels of depression decatanzaro, D. (1986). A mathematical model of
and hopelessness as well. Moreover, al- evolutionary pressures regulating self-preserva-
though decatanzaro appears not to have tion and self-destruction. Suicide and Life-Threat-
considered reproductive potential of kin ening Behavior, 16, 166-181.
decatanzaro, D. (1987). Evolutionary pressures and
an important variable in its own right, our limitations to self-preservation. In C. Crawford,
regression and discriminant analysis re- M. Smith, & D. Krebs (Eds.), Sociobiology and psy-
sults show it to be a significant predictor chology (pp. 311-333). Hillsdale, NJ: Erlbaum.
of suicide ideation and behavior. Finally, decatanzaro, D. (1991). Evolutionary limits to self-
preservation. Ethology and Sociobiology, 12,
although decatanzaro has been careful to 13-28.
note that the Y imodel was intended to decatanzaro, D. (1992). Prediction of self-preserva-
explain the adaptive significance of sui- tion failures on the basis ofquantitative evolution-
ary biology. In R. W. Mans, A. L. Berman, J . T.
cide in ancestral conditions, our data (and Maltsberger, & R. I. Yufit (Eds.), Assessment and
his) suggest that the model has predictive prediction ofsuicide (pp. 607-621). New York Guil-
utility today, in spite of dramatic changes ford Press.
BROWN ET AL. 71

decatanzaro, D. (1995).Reproductive status, family OConnor, R. J . (1978). Brood reduction in birds: Se-
interactions, and suicide ideation: Surveys of the lection for fratricide, infanticide andsuicide? Ani-
general pubic and high risk groups. Ethology and mal Behaviour, 26,79-96.
Sociobiology, 16, 385-394. Poulin, R. (1992). Altered behavior in parasitized
Felner, R. D., Adan, A. M., & Silverman, M. M. (1992). bumblebees: Parasite manipulation or adaptive
Risk assessment andprevention of youth suicide suicide? Animal Behaviour, 44, 174-176.
in schools and educational contexts. In R. W. Maris, Rudd, M. D. (1989). The prevalence of suicidal ide-
A. L. Berman, J. T. Maltsberger, & R. I. Yufit (Eds.), ation among college students. Suicide and Life-
Assessment andprediction ofsuicide (pp. 420-447). Threatening Behavior, 19, 173-183.
New York: Guilford Press. Schulsinger, F., Kety, S. S., Rosenthal, D., & Wender,
Fisher, R. A. (1950). The genetical theory of natural P. H. (1979). A family study ofsuicide. In M.
selection. New York: Dover. Schou & E. Stromgen (Eds.), Origins, prevention
Hamilton, W. D. (1964). The genetical evolution of and treatment of affective disorders (pp. 277-287).
social behavior. Journal of Theoretical Biology, 7, London: Academic Press.
1-16. Smith, K., & Crawford, S. (1986). Suicidal behavior
Hamilton, W. D. (1966). The molding of senescence among normalhigh school students. Suicide and
by natural selection. Journal of Theoretical Biol- Life-Threatening Behavior, 16, 313-325.
ogy, 12, 12-45. Steer, R. A., Kumar, G., & Beck, A. T. (1993). Self-
Hoyer, G., & Lund, E. (1993). Suicide among women reported suicide ideation inadolescent psychiatric
related to number of children in marriage. Archives inpatients. Journal of Consulting and Clinical Psy-
of General Psychiatry, 50, 134-137. chology, 61, 1096-1099.
Humphery, J . A. (1977). Social loss: A comparison Strang, S . P., & Orlofsky, J . L. (1990).Factors under-
of suicide victims, homocide offenders and non- lying suicide ideation amongcollege students: A
violent individuals. Diseases of the Nervous Sys- test of Teicher and Jacobs model. Journal of Ado-
tem, 38, 157-160. lescence, 13, 39-52.
Iga, M., Yamamoto, J., Noguchi, T., & Koshinaga, J . Thornhill, R., & Thornhill, N. W. (1983).Human rape:
(1978). Suicide in Japan. Social Science and Medi- An evolutionary analysis. Ethology and Sociobiol-
cine, 12, 507-516. ogy, 4 , 137-173.
Jones, E. (1957). The life and work of Sigmund Freud Tooby, J., & Cosmides, L. (1992). The psychological
(Vol. 3). New York: Basic Books. foundations of culture. In J. H.Barkow, L. Cos-
Koller, K. M., & Costanos, J . N. (1968). Attempted mides, & J. Tooby (Eds.),The adapted mind: Euolu-
suicide and alcoholism. Medical Journal ofAustra- tionary psychology and the generation of culture
lia, 2, 835-837. (pp. 19-136). New York: Oxford University Press.
Kreitman, N. (1977). Parasuicide. New York: Wiley. Topol, P., & Reznikoff, M. (1982). Perceived peer and
Levenson, H. (1981).Differentiating among internal- family relationships, hopelessness and locus of
ity, powerful others, and chance. In H. M. Lefcourt control as factors in adolescent suicide attempts.
(Ed.), Research with the locus of control construct Suicide and Life-Threatening Behavior, 12, 141-
(Vol. 1, pp. 15-63). New York Academic Press. 150.
Lipe, H., Schulz, A., & Bird, T. D. (1993). Risk factors Trivers, R. L. (1972).Parental investment and sexual
for suicide in Huntingtonsdisease: A retrospective selection. In B. Campbell (Ed.), Sexual selection
case controlled study. American Journal of Medical and the descent of man 1871-1971 (pp. 136-179).
Genetics, 48, 231-233. Chicago: Aldine.
Maris, R. W. (1992). The relationship of nonfatal sui- United States Department ofCommerce. (1991).Sta-
cide attempts to completed suicides. In R. W. Mans, tistical abstract. Washington, DC: Government
A. L. Berman, J. T. Maltsberger, & R. I. Yufit (Eds.), Printing Office.
Assessment andprediction of suicide (pp. 362-380). Woznica, J. G., & Shapiro, J. R. (1990).An analysis of
New York Guilford Press. adolescent suicide attempts: The expendable child.
Maynard Smith, J . (1964). Group selection and kin Journal of Pediatric Psychology, 15, 789-796.
selection. Nature, 201, 1145-1147.
McAllister, M. K., & Roitberg, B. D. (1987). Adaptive
suicidal behaviour in pea aphids. Nature, 328,797-
799.
Motto, J. A., & Bostrom, A. (1990). Empirical indica- Received: January 17, 1997
tors of near-term suicide risk.Crisis, 11, 52-59. Revision Accepted: December 24, 1997

You might also like