On the Incomplete Architecture
of Human Ontogeny
Selection, Optimization, and Compensation as Foundation
of Developmental Theory
Paul B. Baltes
Max Planck Institute for Human Development and Education, Berlin
Drawing on both evolutionary and ontogenetic perspec-
tives, the basic biological-genetic and social-cultural
architecture of human development is outlined. Three
principles are involved. First, evolutionary selection
pressure predicts a negative age correlation, and, there-
fore, genome-based plasticity and biological potential
decrease with age. Second, for growth aspects of human
development to extend further into the life span, culture-
based resources are required at ever-increasing levels.
Third, because of age-related losses in biological plastic-
ity, the efficiency of culture is reduced as life span devel-
opment unfolds. Joint application of these principles sug-
gests that the life span architecture becomes more and
more incomplete with age.
Degree of completeness
can
be defined as the ratio between gains and losses in func-
tioning. Two examples illustrate the implications of the
life span architecture proposed. The first is a general
theory of development involving the orchestration of 3
component processes: selection, optimization, and com-
pensation. The second considers the task of completing
the life course in the sense of achieving a positive bal-
ance between gains and losses for all age levels. This
goal is increasingly more difficult to attain as human
development is extended into advanced old age.
D
uring the last decade, we have witnessed a vigor-
Ious effort to strengthen the link between evolu-
tionary and ontogenetic perspectives in the study
of human behavior. In this spirit, the purpose of this
article is to offer a general framework of the biological
and cultural architecture of human development across
the life span. With this approach, which considers both
evolutionary and ontogenetic arguments, I hope to iden-
tify the foundational structure that any general theory of
human development must have.
Many of the arguments presented owe their line of
reasoning to theoretical propositions associated with life
span developmental psychology (R B. Baltes, 1979,
1987; P. B. Baltes, Lindenberger, & Staudinger, in press;
Elder, in press; Featherman, 1983; Labouvie-Vief, 1982).
The arguments are also consistent with more recent theo-
retical efforts claiming that ontogenesis is inherently a
system of adaptive change involving as foundational ele-
ments the orchestration of three subprocesses: selection,
optimization, and compensation (M. M. Baltes & Cars-
tensen, 1996; P. B. Baltes & Baltes, 1980, 1990; Heck-
hausen & Schulz, 1995; Marsiske, Lang, Baltes, &
Baltes, 1995; Nesselroade & Jones, 1991).
Especially relevant for the present article is the no-
tion that since the classical work of Tetens (1777), life
span scholars proceeded in their theoretical efforts from
the basic assumption that human development essentially
is incomplete. In this article, I contend that this incom-
pleteness of what I call the biological and cultural archi-
tecture of life span development is less promising than
an unfinished Schubert symphony. The situation is more
like an ill-designed building in which inherent vulnerabil-
Editor's note.
Cheryl B. Travis served as action editor for this article.
Articles based on APA award addresses are given special consideration
in the
American Psychologist's
editorial selection process.
A version of this article was originally presented as part of an
Award for Distinguished Contributions to the International Advance-
ment of Psychology address at the 104th Annual Convention of the
American Psychological Association, Toronto, Ontario, Canada, August
1996.
Author's note.
I gratefully acknowledge the many valuable discussions
with and contributions by Margret Baltes, Laura Carstensen, Caleb
Finch, Alexandra Freund, Ulman Lindenberger, Michael Marsiske, John
Nesselroade, Jacqui Smith, and Ursula Staudinger. In addition, I thank
the MacArthur Research Network on Successful Midlife Development
(O. G. Brim, director) for their generous intellectual and financial
support.
Correspondence concerning this article should be addressed to
Paul B. Baltes, Lentzeallee 94, Max Planck Institute for Human Devel-
opment and Education, 14195 Berlin, Germany. Electronic mail may
be sent via Internet to sekbaltes@mpib-berlin.mpg.de.
366 April 1997
•
American Psychologist
Copyright 1997 by the American Psychological Association, Inc. 0003-066X/97/$2.00
Voh 52, No. 4, 366-380
ities, as old age is reached, become more and more
manifest.
The incompleteness of life span human development
results primarily from two conditions. Incompleteness
results first from the fact that biological and cultural
coevolution (Durham, 1991) has not come to a standstill
but is an ongoing process. Second, and most important,
incompleteness results from the fact that the biological
and cultural architecture of human ontogeny is relatively
undeveloped for the second part of the life span (P. B.
Baltes, 1991; P. B. Baltes & Graf, 1996). Neither biologi-
cal nor cultural evolution has had sufficient opportunity
to evolve a full and optimizing scaffolding (architecture)
for the later phases of life. A seeming paradox exists:
Historically speaking, old age is young.
To explore this incompleteness argument and its im-
plications for the future potentials of human development,
life span researchers have focused their work on search-
ing for methods to study age-related changes in plasticity
(potential) and for conceptualizations that permit the
definition of successful or effective human development.
One general approach to this topic has been to define
successful development
as the relative maximization of
gains and the minimization of losses (M. M. Baltes &
Carstensen, 1996; E B. Baltes, 1987; P.B. Baltes &
Baltes, 1990; Brandtstiidter & Wentura, 1995; Marsiske
et al., 1995).
Such a gain-loss approach also permits the defini-
tion of degrees of completeness or incompleteness of the
life span. Using the ratio between achieved gains and
losses as a criterion for evaluation, the life span architec-
ture would be the more complete, the more, in all age
periods of the life course, individuals were to show rela-
tively more gains than losses in functioning. Instead of
gains and losses, it would be possible also to use desir-
able and undesirable states as criteria. Currently, as de-
scribed below in more detail, this pattern of relative com-
pleteness does not exist for all phases of life. Beginning
in late adulthood and certainly in old age, losses outnum-
ber gains, and with age the balance becomes less positive
(P. B. Baltes, 1987).
The determination of what is a gain or a loss in
ontogenetic change is a topic of theoretical as well as
empirical inquiry (see also M. M. Baltes & Carstensen,
1996; Brandtst~idter, 1984; Hobfoll, 1989; Kahneman &
Tversky, 1984; Labouvie-Vief, 1982; Schulz & Heck-
hausen, 1996). Suffice it here to mention that the nature
of what is considered a gain or a loss can change with
age; involves objective in addition to subjective criteria;
and is conditioned by theoretical predilection, standards
of comparison, cultural and historical context, as well as
by criteria of functional fitness or adaptivity.
TAhrchitechJre
of Ufe
Span
Development:
e Frame Resulting From Biological and
Cultural Coevolution
What is the role of cultural and biological factors in
ontogenesis across the life span--how do they interact
and condition each other, for instance, in the sense of
biocultural coevolution (Durham, 1991)? What is the
zone of development, the norm of reaction, and the modi-
fiability or plasticity (P. B. Baltes, 1987; P. B. Baltes &
Schaie, 1976; Lerner, 1984, 1986; Magnusson, 1996) that
one can expect to operate during ontogenesis? On the
basis of genetic- and evolution-based factors and avail-
ability of cultural structures, only certain pathways can
be implemented during ontogenesis, and some of these
are more likely to be realized than others. Despite the
sizable plasticity of
Homo sapiens
and the dynamic qual-
ity of such conceptions as norms of reaction, zone of
development, or plasticity, not everything is possible in
ontogenetic development. Although open, development
inherently is also limited.
Figure 1 illustrates the main lines of argument that
I submit as being the three foundational (constraining)
principles of the life span architecture of human ontogeny
(P. B. Baltes, 1994; P.B. Baltes et al., in press). Note
first that the specific form (level, shape) of the three life
span functions depicted in Figure 1 is not critical. What
is critical are the overall direction and reciprocal interac-
tions between the functions displayed.
Evolutionary Selection Benefits Decrease
With Age
The first foundational principle of my view of the life
span architecture states that the benefits resulting from
evolutionary selection evince a negative age correlation
(Finch, 1990, 1996; Finch & Rose, 1995; Martin, Aus-
tad, & Johnson, 1996; Osiewacz, 1995; Rose, 1991;
Yates & Benton, 1995). As a consequence, the human
genome in older ages is predicted to contain an increas-
ingly larger number of deleterious genes and dysfunc-
tional gene expressions than in younger ages.
Why did whatever happened later in the life span
benefit less from the optimizing power of evolutionary
selection pressure? The primary reason is that reproduc-
tive fitness, the essential component of natural selection,
involved the transmission of genes in the context of fertil-
ity and parenting behavior, events and processes that typi-
Figure 1
Schematic Representation of Three Principles
Governing the DFnamics Between Biology and Culture
Across the Life Span
April 1997 • American Psychologist 367
cally extend from conception to earlier adulthood. As a
consequence, over evolutionary history, selection oper-
ated more strongly on the first half of life. Moreover,
given the much shorter life span in early human evolution,
selection pressure could not operate as frequently to be-
gin with when it came to the second half of life. Most
individuals died before possible negative genetic attri-
butes were activated or their possible negative conse-
quences could become manifest.
One concrete illustration of this aging-based weak-
ening of evolutionary selection benefits is the existence
of late-life illnesses such as Alzheimer's dementia (for
other examples, see Martin et al., 1996). This disease
typically does not become manifest until age 70. After age
70, however, dementia of the Alzheimer's type increases
markedly in prevalence (for specific data, see below).
Alzheimer's dementia is at least in part a late-life disease
because reproductive fitness-based evolutionary pressure
was unable to select against it. Geneticists (e.g., Martin
et al., 1996) call such a situation selection neutrality.
There are other aspects of a biology of aging that,
together and separately, imply an age-associated loss in
biological potential, and whose operation amplifies the
evolutionary neglect of old age (e.g., Finch, 1996; Martin
et al., 1996). Many of these age-related biological losses
are associated with the mechanisms of ontogenesis itself.
Among the prevalent explanations for biological aging
losses are wear-and-tear theories, entropy-based concep-
tions, as well as interpretations, related to the sources of
age-accumulated increases in mutations.
These various considerations about the role of ge-
netic and biological factors converge into an unequivocal
conclusion regarding the dynamics of biological factors
in life span development. Where evolutionary selection
and the ontogenetic biology of aging are concerned, the
life span of humans displays an unfinished architecture,
and in this instance, and certainly after physical maturity,
the consequences of this incompleteness are essentially
negative or dysfunctional. With age, the genetic material,
associated genetic mechanisms, and genetic expressions
become less effective and less able to generate or main-
tain high levels of functioning. Evolution and biology
are not good friends of old age.
A ge-R?lah~l Increase in Need (Demand)
for Culture
The second cornerstone of a life span architecture of
human ontogenesis states that there is an age-related in-
crease in the need or demand for culture (middle part of
Figure 1). Culture in this context refers to the entirety
of psychological, social, material, and symbolic (knowl-
edge-based) resources that humans have generated over
the millenia, and which as they are transmitted across
generations, make human development possible as we
know it today (Boesch, 1991; Cole, 1996; D'Andrade,
1995; Durham, 1991; Shweder, 1991). For human onto-
genesis to achieve increasingly higher levels of function-
ing, for instance, to live longer or to be able to read and
write, there had to be a conjoint evolutionary increase
in the content and dissemination of culture. And the fur-
ther we expect human ontogenesis to extend itself into
adult life and old age, the more it will be necessary for
particular cultural factors and resources to emerge and
operate to make this possible.
To appreciate the power of the evolution of such
culture-based resources, consider what happened to aver-
age life expectancy in industrialized countries during the
20th century. It was not the genetic make-up of the popu-
lation that evinced marked changes during this time. On
the contrary, it was economic and technological innova-
tions that produced significant additions to average life
expectancy, from an average of about 45 years in 1900
to about 75 years in 1995. Similarly, the dramatic increase
in literacy rates over the last centuries in industrialized
nations was not the result of a change in the genome
(that requisite evolution took place at a much earlier time
many millenia ago; e.g., Klix, 1993) but, above all, a
change in environmental contexts, cultural resources, and
strategies of education.
There is a second argument for the proposition that,
with age, the need for the supportive and enriching role of
culture increases. The demand for culture also increases
because, as individuals reach old age, their biological
potentials decline (left part of Figure 1). The older in age
individuals are, the more they are in need of culture-
based compensations (e.g., material, technical, social,
economic, psychological) to generate and maintain high
levels of functioning. This view of "culture as compensa-
tion" is a major tenet of many evolutionary theories in
cultural anthropology (P. B. Baltes, 1991; Brandtst~idter,
in press; Dixon & B~ickman, 1995; Durham, 1991; Mar-
siske et al., 1995). Consider the evolution of clothing in
its many variations as an example. At the evolutionary
base is a biological lack. Humans wear cloth in part
because of a thermoregulation deficit.
Age-Related Decrease in Efficiency of Culture
The right panel of Figure 1 illustrates the third corner-
stone of the overall architecture of the life course. This
foundational principle states that there is an age-related
loss in the effectiveness or efficiency of cultural factors
and resources. With age, and conditioned primarily by
the negative biological trajectory of the life course, the
relative power (effectiveness) of psychological, social,
material, and cultural interventions wanes. In summary,
although there continues to be plasticity in the second half
of life--and its extent may even be larger than typically
believed (M. M. Baltes & Carstensen, 1996; P. B. Baltes,
1987; Lerner, 1984; Willis, 1990)rathe scope of plasticity
of the human organism declines with age.
Take cognitive learning in old age as an example
(P. B. Baltes, 1993; Birren & Schaie, 1996; Craik &
Salthouse, 1992; Lindenberger & Baltes, 1995; Salt-
house, 1991). The older the adult is, the more time, prac-
tice, and cognitive support it takes to attain the same
learning gains. Moreover, when it comes to high levels
of performance, older adults may never be able to reach
the same levels of functioning as younger adults even
368 April 1997 • American Psychologist
after extensive training (P. B. Baltes & Kliegl, 1992;
Ericsson & Smith, 1991; Kliegl, Mayr, & Krampe, 1994;
Kliegl, Smith, & Baltes, 1989). Similar conclusions apply
to life span changes in neuronal plasticity (Magnusson,
1996). Neuronal plasticity continues to exist across the
life span, but with age it is reduced in scope and
efficiency.
This third principle of an age-related reduction in
cultural efficiency is likely to raise objections in social
science circles. Two reasons are central. A first objection
is the notion that the specifics of cultural systems, namely
their symbolic form, may follow different mechanisms
of efficiency. For instance, the life span developmental
entropy costs of symbolic systems may be more favorable
than those observed for basic biological processes (P. B.
Baltes & Graf, 1996). Second, some social scientists ar-
gue that the concept of efficiency contains assumptions
about human functioning that are inherently opposed to
phenomena such as meaning of life, a sense of religion,
or an understanding of one's finitude (P. B. Baltes et al.,
in press; Dittmann-Kohli, 1995; Rosenmayr, 1990). These
are serious lines of argument. However, I submit that
such perspectives, important and critical as they are for
an understanding of human development, do not alter the
general direction of the life span function outlined. In
other words, the primary impact of symbolic systems and
related meaning systems is on level, rate, and the life span
extension of cultural efficiency, not on its ontogenetic
directional course.
I submit that the three propositions and trajectories
outlined in Figure 1 form a robust architecture of the
life span dynamics between biology and culture. This
architecture, including its growing incompleteness as the
life course unfolds, represents the most general frame
within which developmental theory is embedded. What-
ever the specific content and form of a given psychologi-
cal theory of human ontogeny, they need to be consistent
with this architectural frame. For instance, any theory of
life span development that were to posit "general" posi-
tive advances across broad domains of functioning in
later adulthood can be judged to be false. Similarly false
would be the view that development at any age could
consist of pure gain. In the past, as argued for instance
by Hetherington and Baltes (1988) and Labouvie-Vief
(1982), such a pure-gain view of development was often
held by cognitive child developmentalists.
Why is a pure-gain view of ontogenesis false? As is
true for evolution, ontogenetic adaptivity and ontogenetic
attainments are always local, that is, context, space and
time (age) bound. In addition, development always in-
volves selection and, therefore, a trade-off between alter-
native pathways and success-failure constellations. This
recognition has led life span theorists such as Jochen
Brandts~'dter, Gisela Labouvie-Vief, and myself to reject
any conception of development that is unilinear and based
solely on the notion of growth as a unidimensional ad-
vance in quantity and quality of functioning. On the con-
trary, multicausality, multidimensionality, multidirection-
ality, and multifunctionality reign supreme in ontogenesis
at all stages of the life course (M. M. Baltes, 1996; P. B.
Baltes, 1987; P. B. Baltes et al., in press; Brandtst~idter,
in press; Labouvie-Vief, 1982; Marsiske et al., 1995;
Schulz & Heckhausen, 1996; Uttal & Perlmutter, 1989).
The changing life span dynamics in the ratio of
gains and losses are also evident at the level of subjective
beliefs about the overall developmental trajectory of
gains and losses across the life span (Heckhausen,
Dixon, & Baltes, 1989). For instance, when asked about
the typical adult developmental trajectory of a large num-
ber of attributes (such as intelligent, strong, anxious, sick,
etc.), persons report a script that involves a changing ratio
of gains and losses toward an increasingly less desirable
balance (see left part of Figure 2). With increasing age,
more and more of the expected changes involve losses
rather than gains, although there continue to be some
expectations of gains in old age, such as an increase in
dignity and wisdom. And with this changing ratio be-
tween gains and losses, there is a corollary change in the
expected proportion of success and failure experiences
(Schulz & Heckhausen, 1996).
This life span script about gains and losses has con-
sequences for the subjective experience of life span devel-
opment. Thus, when persons of different ages are asked
about what age they would like to be, they report an
increasing negative discrepancy between their actual age
and their desired age (Smith & Baltes, 1996). Seventy-
year-olds, for instance, would like to be on average about
10 years younger. For 90-year-olds, the discrepancy is
increased to about 25 years. These findings of an age-
related increase in the discrepancy between actual age
and desired age lend support to the reality of a growing
disjuncture between body and mind--between what is
desirable and what is possible. These findings also illus-
trate that the beliefs that individuals hold about the life
course reflect the biological and cultural incompleteness
of the life span and the resulting age-related increase in
less desirable outcomes.
Ufe Span Changes in theAIIocation of
Resources to Functions of Development
In the following section, I pursue a strategy where the
implications of this overall architectural landscape of hu-
man development are examined at levels of analysis that
step-by-step approximate the kind of questions that devel-
opmental psychologists study. A first step in this direc-
tion, and as illustrated in the right part of Figure 2, in-
volves consideration of three general functions or out-
comes of development: (a) the function of growth; (b) the
function of maintenance, including recovery (resilience);
and (c) the function of regulation of loss. These three
functions and their associated outcomes represent the sys-
temic whole of individual development (see also Stau-
dinger, Marsiske, & Baltes, 1995).
With the adaptive function of growth, I refer to be-
haviors involved in reaching higher levels of functioning
or adaptive capacity. Under the heading of maintenance
and recovery (resilience), I classify behaviors involved
in maintaining levels of functioning in the face of a new
April 1997 • American Psychologist 369
I
Figure 2
Life Span Script of Allocation of Resources to Functions of Development
Life Span
Note.
The left part shows subjective expectations about desirable (gains) and undesirable {losses) changes across adulthood on the basis of Heckhausen, Dixon,
and Baltes (1989). The right side illustrates life span changes in the allocation of resources into three functions
of development:
growth, maintenance (resilience), and
regulation of loss (Staudinger, Marsiske, & Baltes, 1995).
I I I I
contextual challenge or a loss in potential. Finally, regard-
ing regulation of loss or management, I mean behaviors
that organize functioning at lower levels when mainte-
nance or recovery (resilience) is no longer possible.
Because of the architecture outlined above, my col-
leagues and I argue that there is a systematic life span
shift in the relative allocation of resources to these three
functions (P. B. Baltes, 1994; P. B. Baltes et al., in press;
Marsiske et al., 1995; Staudinger et al., 1995). In child-
hood, the primary allocation is directed toward growth;
during adulthood, the predominant allocation is toward
maintenance and recovery (resilience). In old age, more
and more resources are directed toward regulation or
management of loss. Such a characterization of the life
span, of course, is an oversimplification, as individual,
functional (domain), contextual, and historical differ-
ences need to be taken into account. The life span script
is about relative probability and prevalence. Note in this
context that the reaUocation of resources toward mainte-
nance of functioning and regulation of loss is facilitated
by the tendency of individuals to prefer avoidance of loss
over enhancement of gains (Hobfoll, 1989; Kahneman &
Tversky, 1984).
The life span trajectories regarding investment of
resources into growth, resilience, and regulation of loss
also have implications for the dynamics involved in the
systemic and integrative coordination of these three func-
tions. A first telling research example is the strong inter-
est of adult development researchers in topics such as
selection of goals (Cantor & Fleeson, 1994) but espe-
cially compensation (Biickman & Dixon, 1992; P.B.
Baltes & Baltes, 1990; Dixon & B~ickman, 1995). And
when considering compensation, theoretical considera-
tions include the seemingly counterintuitive view that
deficits can breed advances through innovative efforts
(P. B. Baltes, 1987, 1991; Burghardt, 1984; Uttal & Per-
lmutter, 1989). Such a view, incidentally, is consistent with
evolutionary theory, where it is widely held that during
evolution, increases in adaptive capacity were enhanced
by conditions of stress and challenge (anagenesis).
Another telling example of the dynamics among the
functions of growth, resilience, and regulation of loss is
the life span study of the interplay between autonomy
and dependency in children and older adults (M. M.
Baltes, 1996). Whereas the primary focus of the first half
of life is the maximization of autonomy, in old age, the
productive and creative use of dependent behavior be-
comes critical. According to Margret Baltes, for older
adults to maintain autonomy in select domains of func-
tioning, the effective exercise and use of dependent be-
havior is a compensatory must. By invoking dependency
and support, resources are freed up for use in other do-
mains "selected" for personal efficacy and growth. Simi-
lar perspectives, of course, can be applied to functioning
in childhood, such as the nature of child care or motor
development (Hetherington & Baltes, 1988; Labouvie-
Vief, 1982; Thelen & Smith, 1994).
Selective Optimization With
Compensation: An Example of
a General (Sys_temic) Theory of
Life Span Development
In this section, I add another level of analysis and apply
the basic life span architectural frame to the formulation
of a general model of development. For this purpose, I
characterize a
metatheory of development,
selective opti-
mization with compensation (SOC), which, together with
several colleagues, Margret Baltes and I have developed
over the last decade (M. M. Baltes & Carstensen, 1996;
P. B. Baltes, 1987; P.B. Baltes & Baltes, 1980, 1990;
P.B. Baltes, Dittmann-Kohli, & Dixon, 1984; Cars-
tensen, Hanson, & Freund, 1995; Marsiske et al., 1995).
In our work on SOC, the original intellectual moti-
vation was the search for a general process of systemic
functioning that would serve as an effective strategy for
370 April 1997 • American Psychologist