COGNITIVE AND BRAIN DEVELOPMENT

THE ADOLESCENT MIND: BUILDING AN INTEGRATED MODEL
The 1958 publication in North America of Inhelder and Piaget’s classic work, The
Growth of Logical Thinking from Childhood to Adolescence (following its 1955 publication
in France as “De la logique de l’enfant à la logique de l’adolescent: Essai sur la
construction des structures opératoires formelles”) is widely recognized as having
launched the systematic study of adolescent cognitive development. The introduction
to an English-language audience of a comprehensive model for understanding the cognitive
transition from childhood to adolescence inspired a generation of researchers to
undertake intensive investigations. The echoes reverberate still, even though direct assessment
of Piaget’s theoretical claims regarding formal operations has become less frequent
(Bond, 1998; Keating, 1990a; Moshman, 1998).
Some appreciation of the impact can be gleaned from a simple count of publications
on the topic before and after this watershed event. Using a comprehensive search engine
(PsychInfo, Cambridge Scientific Abstracts) and crossing the terms adolescent or
adolescence with cognition, cognitive development, or thinking, fewer than a dozen empirical
articles from 1872 (the earliest date for indexing) to 1960 were identified. In succeeding
decades, an exponential growth pattern in such publications is apparent, from
the low hundreds in the 1960s up to the low thousands in the 1990s. In contrast, this
new emphasis virtually displaced publications on psychometric research specific to
adolescence, a trend astutely noted at its cusp by Ausubel and Ausubel (1966). Crossing
adolescent or adolescence with intelligence or mental ability yielded over 1,000 titles
in the earliest period, but fewer than 100 in the 1990s.
The shift from a psychometric focus on intelligence and mental ability toward underlying
cognitive structure and process reflected larger trends in psychology generally and
cognitive science in particular from the early 1960s onward. In addition to the primary
goal of rescuing cognitive activity as a legitimate area of inquiry from the behaviorist
hegemony of the earlier period, a major scientific goal of the cognitive revolution
(Gardner, 1985; Johnson & Erneling, 1997; Pribram, 1986) has been to understand the
underpinnings of complex cognitive performance in the most elementary terms possible.
The study of cognitive development in adolescence fit comfortably within this larger
trend. The search for the essential elements of cognitive activity in adolescents yielded
several prime candidates, each of which spawned a substantial line of research. There
45
were several commonalities among these models. Each sought to identify the driver of
adolescent cognitive activity, producing what can be viewed as single-device accounts
(Keating, 2001a). In so doing, each also sought to answer the perennial question, “What
develops?” (Siegler, 1978). Each also sought to demonstrate that there were specifically
adolescent features of cognitive development.
Three major theoretical approaches to the core of adolescent cognitive development
during this initial efflorescence can be identified: cognition as reasoning, cognition as
processing, and cognition as expertise. These major accounts were not unrelated and
indeed arose sequentially in response to each other (Keating, 1980, 1990a). In computational
terms, one can view these as focusing, respectively, on developments in the design
of the operating system, on additions to the speed and/or capacity of the system,
or on changes in the size or structure of the database that the system has available.
A major theme of this review is that claims for independent developments in these
various aspects of adolescent cognition have not been supported by the weight of evidence,
even though important developments are clearly evident in each aspect of the
overall cognitive system. For example, Demetriou, Christou, Spanoudis, and Platsidou
(2002), using individual growth curve modeling from late childhood to adolescence,
found that more advanced levels of reasoning arise in part from bottom-up changes in
processing efficiency and working memory, but they also found that these lower order
elements are reciprocally affected from the top down. This interdependence has complicated
the search for the core of what develops—that is, identifying the fundamental
driver of adolescent cognitive development independent of other potentially confounding
cognitive shifts.
A Central Paradox
The inability to identify a single dominant source that explains what develops in adolescent
cognition has contributed to a central paradox. Despite the absolute growth in
the number of studies dealing with adolescent cognitive development just noted, and
despite the centrality of the questions that animated the major approaches, there is a
paradoxical (but widespread) view that there has been relatively little progress on these
central themes: “The study of cognitive development in adolescence has been moribund
for some time now, replaced by studies of adolescent decision making and judgment”
(Steinberg & Morris, 2001, p. 101). Compounding the paradox is an equally
widespread view that there are fundamental shifts in adolescent cognition and that
these changes are readily observable in important, broadly defined aspects of adolescent
thinking (Keating, 1990a; Moshman, 1998). The power of this paradox is further
reinforced by the emerging understanding of the dramatic nature and long reach of
pubertal events (Angold, Costello, Erkanli, & Worthman, 1999; Angold, Costello, &
Worthman, 1998; Susman & Rogol, this volume), which sustains the belief that there
should be large and readily identifiable shifts in both cognitive and brain development.
Can this paradox be resolved? To begin to answer this difficult challenge, we need to
attend to the subsequent trajectories of research that originated in the different theoretical
perspectives noted earlier (Keating, 2001a). This is of more than historical interest.
As these approaches have pursued their largely separate paths, they have en-
46 Cognitive and Brain Development
countered highly similar dilemmas. Indeed, they are similar not only by analogy, but
also as homology. Each has confronted, I argue, the core problem of all single-device
accounts. The greater the refinement of parameters to measure the theoretically targeted
developmental mechanism, the less these parameters are able to account for
broad changes that the theories were designed to explain (Keating, 1996b; Keating,
List, & Merriman, 1985).
In short, researchers have long sought to explain the readily apparent and broad
changes in adolescent thinking by identifying the underlying mechanism of “what develops”
(Siegler, 1978). In pursuing this agenda, it is logical to identify and measure parameters
of the targeted mechanism as precisely as possible: pure logic, pure processing
efficiency, pure capacity, and so on. As the evidence reviewed in this chapter demonstrates,
this research agenda has yielded much new knowledge about adolescent cognition
but has encountered significant problems in answering the original question.
To the extent that the strategy has succeeded in purifying the parameters of specific
underlying mechanisms, it has been at the cost of explaining variance in the broader
cognitive shifts. Moreover, in studies where multiple mechanisms have been examined,
especially in longitudinal designs (e.g., Demetriou et al., 2002), there is strong evidence
for interdependence among them. It seems increasingly unlikely that any single-device
explanation of the major shifts in adolescent cognition will suffice. Confronting these
recurring limitations, lines of research have bifurcated and branched, yielding a burgeoning
complexity of questions and paradigms that focus on adolescent cognition.
As Steinberg and Morris (2001) accurately noted, this has redirected much of the research
attention away from the original questions about the core of adolescent cognitive
development toward a search for answers to specific practical questions. These have
been more typically addressed by an applied cognitive science with a more contextspecific
focus on pathways to expertise, on the role of teaching and learning in the acquisition
of expertise, and on the development of judgment and decision making (e.g.,
Keating, 1990b; Klaczynski, Byrnes, & Jacobs, 2001; Kokis, Macpherson, Toplak, West,
& Stanovich, 2002; Kuhn, Garcia-Mila, Zohar, & Andersen, 1995; Kuhn & Pearsall,
2000; Kwon & Lawson, 2000; Kwon, Lawson, Chung, & Kim, 2000; Lawson et al., 2000).
In pursuing these more applied questions, the dilemmas of interdependence and insufficiency
of single-device accounts reemerge for the same reasons and in the same way.
More provocatively, this paradox—of major shifts that are not explained by any
single developmental mechanism—can be viewed as a strong indication that a more integrative
theoretical account is needed. If one adopts this position, as this review does,
we then need to ask what shape such an integrative account might take, what evidence
may be available to support such an account, and what would constitute an effective research
agenda to probe an integrated model. The prospects for an integrative account
have been substantially enhanced by recent major advances in the neurosciences (e.g.,
Casey, Giedd, & Thomas, 2000; Giedd et al., 1999; Johnson, 2001; Luna et al., 2001;
Nelson, 1999; Paus et al., 1999; Sowell, Delis, Stiles, & Jernigan, 2001; Sowell, Trauner,
Gamst, & Jernigan, 2002; Steingard et al., 2002), in comparative neuroanatomy across
closely related primate species that illuminate core issues of human cognitive evolution
(Donald, 2001; Rilling & Insel, 1999), in the increasing sophistication of analyses of longitudinal
data and individual growth curves (Demetriou et al., 2002; McArdle, Ferrer-
The Adolescent Mind: Building an Integrated Model 47
Caja, Hamagami, & Woodcock, 2002), and in a deepened understanding of the critical
role of culture and context in the shaping of cognitive and brain development (Donald,
2001; Francis, Diorio, Plotsky, & Meaney, 2002).
An Emergent Resolution
Drawing on these various sources, the nature of an integrated account has begun to
emerge. It is sketched in this introductory section and elaborated in subsequent sections.
What lies at the core of adolescent cognitive development is not likely to be any
single device that drives it. Rather, it is the attainment of a more fully conscious, selfdirected,
and self-regulating mind that characterizes the adolescent transition. This is
achieved principally through the assembly of an advanced executive suite of capabilities
(Donald, 2001), rather than through specific advancement in any one of the constituent
elements. This represents a major shift in prevailing views of cognition, going
beyond the search for underlying elements (or hidden “demons”; Dennett, 1991) that
are formed and operate largely outside awareness. Such coordination of cognitive elements
resonates with contemporary work on metacognitive and metastrategic development.
In describing the core difference between young and mature scientists, for
example, Kuhn and Pearsall (2000) noted that both seek to coordinate theory and evidence,
but that in the latter, “the coordination of theory and evidence is carried out under
a high degree of conscious control” (p. 115).
Recent advances in developmental neuroscience, propelled by dramatic advances in
imaging technologies, point toward a similar resolution. A broad outline of the most
significant and cognitively relevant developments in the adolescent brain has begun to
coalesce very recently (Casey et al., 2000; Johnson, 2001). Much of the underlying action
is focused on specific developments in the prefrontal cortex, but with an equally
significant role for rapidly expanding linkages to the whole brain (Donald, 2001; Luna
et al., 2001; Newman & Grace, 1999). This complex process of assembly is supported
by increasingly rapid connectivity (through continued myelination of nerve fibers), particularly
in communication among different brain regions, and by significant and localized
synaptic pruning, especially in frontal areas that are crucial to executive functioning
(Giedd et al., 1999; Sowell et al., 2001, 2002; Steingard et al., 2002).
It is noteworthy that the most marked differences between adult humans and nonhuman
primates (Rilling & Insell, 1999) occur precisely in those features of brain development
that emerge most strikingly during adolescence: differentially greater increases
in neocortical volume (beyond the expected increase owing to larger brain volume overall);
differentially greater gyrification of the prefrontal cortex, indicating a more convoluted
design that affords both more capacity for central coordination and more rapid
communication; and a greater relative increase in cerebral white matter relative to neocortical
gray matter, “suggesting that axonal connections between neocortical neurons
may increase faster than the number of neurons as brain size increases” (p. 222). The
prefrontal cortex is thus among the latest brain systems to develop, both phylogenetically
and ontogenetically, especially the dorsolateral prefrontal cortex (Fuster, 2000).
Its importance for adolescent cognitive development can be inferred from this convergence
but is more directly evident from its integrative functions.
The prefrontal lobes have long been seen as central to the coordination of cognitive
48 Cognitive and Brain Development
activity (Case, 1992; Stuss, 1992), whose function is to sustain “many high-level metacognitive
operations, such as self-evaluation, long-term planning, prioritizing values,
maintaining fluency, and the production of appropriate social behavior” (Donald,
2001, p. 198). The reach and complexity of the prefrontal cortex has been further emphasized
in recent work identifying important details of its structure and function (Barbas
& Hilgetag, 2002; Fuster, 2000; Fuster, Van Hoesen, Morecraft, & Semendeferi,
2000; Watanabe, 2002). In particular, the temporal integration of retrospective memory
and preparatory set in the dorsolateral prefrontal cortex (Fuster, 2000) creates the
conditions for a broader, more fully conscious control of cognition and behavior. In
other words, the narrow window of active awareness that is the focus of much of experimental
cognitive science, which seems to undermine assertions of a broader consciousness
with significant agency in decision making and planning, captures too little
of the important cognitive action (Donald, 2001). In addition, the close connection and
communication among regions of the prefrontal cortex that are more cognitive with
those that have important linkages with emotional processing, especially the orbitofrontal
cortex (Barbas & Hilgetag, 2002), support the view of the prefrontal cortex as a
more general synthesizer of experience and governor of action. Thus the role of the prefrontal
cortex as not only an integrator of cognitive functions, but also a governing regulator
of emotion, attention, and behavior, takes on special importance during adolescence
in three ways.
First, patterns of individual differences in how cognition, emotion, and behavior become
integrated during adolescence may well have a long reach with respect to the development
of psychopathology (Steinberg et al., in press) and to normative habits of
mind (Keating, 1996b, 1996c) that influence trajectories of competence and coping.
Second, the pubertal influences on many hormonal and neuroendocrine systems are
dramatic (Angold et al., 1998, 1999; Susman & Rogol, this volume), entailing the cascading
reorganization of body and brain systems. Third, recent evidence from animal
models demonstrated the partial reversibility of damage acquired during early development,
at both the behavioral and the physiological levels, as a function of enriched
environments during puberty (Francis et al., 2002). In combination, this evidence
points strongly toward both enduring (but not limitless) neural plasticity and the critical
role of developmental experience in shaping future developmental trajectories in
cognition and behavior (Nelson, 1999; Nelson, Bloom, Cameron, Amaral, Dahl, &
Pine, 2002).
Recent research directions thus point toward the distinct possibility that adolescence
may be a sensitive or critical developmental period, much like early development
in its ability to shape future trajectories and in the biological embedding of developmental
experiences as the principal method through which this occurs (Boyce & Keating,
in press; Keating & Hertzman, 1999; Meaney, 2001). The major changes during
adolescence just outlined form the essential substrate for adolescence to function as a
critical developmental period: the interdependence and developmental coordination of
numerous cognitive elements, and of cognition with emotion and behavior; the role of
puberty in a fundamental restructuring of many body systems; the apparent concentration
of changes in the adolescent brain in the prefrontal cortex (which serves as a
governor of cognition and action) together with the enhanced interregional communication
between the prefrontal cortex and other brain regions; and in the evidence for
The Adolescent Mind: Building an Integrated Model 49
substantial synaptic pruning and for nontrivial physiological reversibility of behavioral
and neuroendocrine patterns arising from early developmental experiences.
Brain, Culture, and the Centrality of Consciousness
The reemergence of consciousness and its development as a legitimate area of scientific
inquiry (Chalmers, 1996; Dennett, 1996; Donald, 2001; Ferrari, Pinard, & Runions,
2001; Searle, 1997) has increased the probability that adolescence may be a critical developmental
period for cognitive development and for the myriad ways that cognition
plays a role in emotion and action. The late development of the prefrontal cortex in phylogeny
and ontogeny, as well as its central integrative function, provides a crucial starting
point for understanding the distinctive features of human cognition generally and
adolescent cognition specifically.
As McGinn (1999) noted, “Consciousness is hard to miss but easy to avoid, theoretically
speaking” (p. 44). The classic subject-object and mind-body problems in
epistemology (and philosophy more generally) drove consciousness out of scientific
scrutiny in two directions, either by explaining it away through reductionist accounts
that allowed it to exist only as an epiphenomenon resting on the actions of more fundamental
cognitive automata (Dennett, 1991), or, alternately, by awarding it a privileged
position beyond objectifying scientific inquiry, to be understood by largely phenomenological
methods. But as McGinn (1999) observed, consciousness just won’t go
away, posing again the question: “How does consciousness fit into the scientific world
picture. . . ?” (p. 44).
Ferrari et al. (2001) argued that Piaget (for whom, they noted, the scientific study of
consciousness never became unrespectable) proposed a solution that should be revisited.
Specifically, the physiological “rules” that enable cognitive activity and the (psycho)
logical “rules” that govern conscious awareness coevolved in a “sophisticated form
of parallelism” (p. 195) that exhibits substantial isomorphism and leads toward an “integrative
monism” (p. 195) that may never actually arrive.
A Piagetian anecdote about his encounter with Soviet psychologists (in a new translation
by Ferrari et al., 2001, p. 199), while revealing heavy ideological baggage, emphasizes
the convoluted epistemological and philosophical argumentation that the scientific
study of consciousness confronts. In the anecdote, Piaget is cleared of the charge
of being an idealist on the grounds that he grants primacy to the preexistence of objects
before a knower, even though he maintains a strong stance that the origin of knowing
lies in the first actions upon objects. What the Soviets were probing is the essential (but
creative) tension at the heart of the Piagetian project (Keating, 1990d, pp. 312–315), between
the specification of closed structures with the definable end point of formal reasoning
(an idealist position despite the Soviet acquittal), and the identification of an
open-system, constructivist process that codevelops iteratively through action on the
world (sufficiently objectivist to satisfy his interlocutors on that occasion). As I have
previously argued (Keating, 1990d), it is the former position, the closed structure with
defined end points, that was initially more amenable to cognitive developmental investigation,
and hence defined the limits of much of the research on adolescent cognition.
The open systems view, more compatible with much intervening theoretical work, sup-
50 Cognitive and Brain Development
ports the identification of Piaget as a root theorist of the development of consciousness
(Ferrari et al., 2001).
Relying more heavily on recent evidence in human evolution and neuroscience, and
thus eschewing many of the ideological issues, Donald (2001) proposed a framework
that is compatible with these Piagetian arguments that are based in the method of genetic
epistemology. Donald’s (2001) core argument is that the evolution of the human
brain, especially as it branched from the broader primate lineage, was quintessentially
social: “The key to understanding the human intellect is not so much the design of the
individual brain as the synergy of many brains. We have evolved an adaptation for living
in culture, and our exceptional powers as a species derive from the curious fact that
we have broken out of one of the most critical limitations of traditional nervous systems—
their loneliness, or solipsism” (p. xiii).
It is important to distinguish this notion of the fundamentally social nature of human
mind from the line of research in organizational behavior launched by Janis’s
(1972) seminal work on groupthink. That theoretical construct has focused on the
mechanisms by which ineffective or poor decision making can occur in the context of
(usually small) group dynamics (Flippen, 1999; Jones & Roelofsma, 2000; Park, 2000;
Turner & Pratkanis, 1998), such as failure of monitoring, premature concurrence, failure
to consider unconventional options, or perceived risk of excluding oneself from the
group through disagreement (Baumeister & Leary, 1995). The fundamentally social nature
of cognition and consciousness may be a necessary condition for groupthink, but
the multifaceted nature of social or collective mind cannot be accurately characterized
as groupthink. It is interesting to note, however, that despite a substantial line of research
on how peers may influence individual decision making during adolescence
(Ormond, Luszcz, Mann, & Beswick, 1991; Payne, 2002; Steinberg & Silverberg, 1986),
there is little research on groupthink among adolescents. This is somewhat surprising
given the group context that often underlies immaturity of decisions (Cauffman &
Steinberg, 1995; Dishion, Eddy, Haas, Li, & Spracklen, 1997) and the salience of the
sense of belonging that is an important aspect of the adolescent self (Marshall, 2001).
There is a real tension between the formative role of society in essential cognitive development
and the need to develop an autonomously critical habit of mind (Keating,
1996c), and clearly this tension is a key dynamic in adolescent cognition.
Donald (2001) proposed three major transitions as levels of consciousness beyond
the episodic awareness that we share with our nonhuman primate cousins. Each rests
on continuing development and refinement of the “executive suite” that is concentrated
in the prefrontal cortex (including its interregional connectivity): “(1) more precise and
self-conscious control of action in mimesis; (2) richer and faster accumulation of cultural
knowledge, in speech; and (3) much more powerful and reflective cultures, driven
by symbolic technology” (p. 262). Note that each of these proceeds as a coevolution of
brain and culture; indeed, the signal species characteristic of Homo sapiens can be
thought of as cultural mind sharing that activates individual minds (whose brains have
been designed, as it were, by evolution to participate in just such activity). In this light,
there is an essential concordance on the centrality of interactive constructionism between
Vygotsky’s (1979) cognitive socialization and Piaget’s (1965) “internal interactionism”
(Ferrari et al., 2001, p. 198).
The Adolescent Mind: Building an Integrated Model 51
The implication of these convergences is that there exist potentially homologous
qualitative transitions in the evolution of the human brain, the nature of primate group
interaction (episodic, mimetic, mythic, and theoretic; Donald, 2001), and adolescent
cognitive and brain development. The essence of the homology lies in the nature of human
consciousness, including its phylogeny, ontogeny, and inseparability from culture
and context. Fully attained human consciousness is thus potentiated by key developments
in the brain that are late arrivals in evolutionary history and become fully available
for assembly in individual ontogeny only during the adolescent transition. But this
potential assembly becomes actual only in close interplay with the surrounding cultural
and cognitive web, in which the individual adolescent experiences culture not passively
as an external entity, but rather as an active force (Swidler, 1986) that can be used to
both define and achieve goals, and serves simultaneously as a fundamental coconstructor
of cognition and consciousness.
Understanding how adolescence may function as a critical developmental period
will be a major challenge. The necessity of incorporating the multiple interactions of
brain-biology, behavior-cognition, and culture-context implies a level of complexity
that is daunting. This poses two risks to a productive research agenda that can advance
our understanding. First, the complexity may be overwhelming, leading instead toward
continued fragmentation and inattention to core questions—the central paradox described
earlier. Second, the need for coherence may lead to an overly facile acceptance
that the mere proposal of a global homology centered on the emergence of consciousness
fully resolves the paradox. It does not, because consciousness emerges (phylogenetically,
historically, and ontogenetically) not as one thing, but as the assembly of
many things (Donald, 2001). Each of its many subsystems retains some degree of ongoing
modular independence in the context of increasing systemic interdependence.
Their coordination into a more centrally governed metasystem is unlikely to be smooth
developmentally or uniform in execution. Indeed, the individual and developmental
anomalies of assembly are readily apparent as performance or procedural intrusions
that muddy the waters of what might otherwise be observed as a clear progression to
fundamental logical competence (Overton, 1990).
To summarize, the initial stages of research on adolescent cognitive development focused
logically and necessarily on isolating key cognitive components. The recent emergence
(or reemergence) of consciousness as a legitimate and necessary focus of cognitive
research (Donald, 2001) has been driven by a need to understand how those components
get assembled into a complex, coordinated cognitive system. The accumulating
evidence is particularly relevant to adolescent cognitive development, reflecting the
broader movement in cognitive science as it comes to terms with cognitive, cognitiveaffective,
and developmental neuroscience. In this light, understanding adolescent cognitive
development will have to play a central role in understanding human consciousness
more generally, as research in a number of related fields points with increasing
specificity to this period of development as the one in which the processes of assembly
are at their peak.
Contemporary evidence also supports the view that adolescents are biologically susceptible
to the embedding of experience in brain and body systems. Thus, a comprehensive
research agenda will require a much more detailed understanding of the ways
in which context organizes the salient experiences of adolescents (see chapters in Part
52 Cognitive and Brain Development
II of this volume, especially that by Larson & Wilson). If the adolescent mind is coconstructed
by the potentiation of developing cognitive capacities through active engagement
with culturally embedded knowledge and meaning systems, then the specific
ways in which adolescents engage their cognitive universe become crucial. In contrast,
identifying the most salient aspects of early development in order to understand how
they may become biologically embedded in the infant and young child may be relatively
easier than in the adolescent, with the increased complexity of the relevant contexts and
the increased ability for individual choice in the selection of contexts.
The principal goal of this review is thus to examine closely (but not exhaustively,
given the breadth of the extant literatures) the major research efforts on adolescent cognitive
and brain development, to identify common themes and findings that cut across
the disparate research agendas, and to situate those themes with respect to an emergent
theoretical perspective that integrates them. To move this research agenda forward productively,
emerging methodologies that support interdisciplinary cross-fertilization
will be essential. Wilson (1998) argued for an innovative methodological approach that
seeks consilience of evidence among biological, social, and human sciences. Although
hard to achieve, the understanding of adolescent cognitive development requires such
an integration, and the evidence reviewed in this chapter suggests that the prospects are
promising.
Following this first section, in which the overall argument has been outlined, the balance
of the review is organized as follows. In the second part, a more detailed consideration
of the major lines of research on adolescent cognitive development is provided,
and their theoretical and empirical implications for an integrated model are identified.
In the third part, the prospects for grounding adolescent cognitive development in
underlying biological and brain development are discussed further, with a particular focus
on emerging evidence from the neurosciences that supports (or even requires) an integrated
model. In the fourth and final part, the role of context in adolescent cognitive
development is addressed in light of these newly emerging perspectives. In contrast to
conventional approaches that sought to partial out or control for context in order to
identify purely cognitive parameters, arguments for the role of enculturation as an active,
integral, and endogenous feature of adolescent thinking are reviewed (Donald,
2001; Keating, 1996b, 1996c; Keating & Sasse, 1996).