CHAPTER 16
Early Attachment and Later Development
Familiar Questions, New Answers
ROSS A. THOMPSON
How does the child foreshadow the adult-to-be? methods for assessing the security of attachment
Philosophers, spiritualists, playwrights, and most for infants and young children, an enormous re-
recently behavioral scientists have sought to un- search literature emerged concerning the origins,
derstand how early dispositions and influences correlates, and consequences of secure and inse-
provide a foundation for adult personality. Among cure relationships. Guided by a general expecta-
the answers they have offered is the influence of tion that a secure attachment would predict better
early, close relationships. This view was eventu- later functioning, developmental researchers have
ally crystallized in Freud’s (1940/1963, p. 45) fa- explored the association between early security
mous dictum that the infant–mother relationship and later relations with parents, peers, friends, and
is “unique, without parallel, established unalter- other social partners, as well as with self-concept,
ably for a whole lifetime as the first and stron- competence in preschool and kindergarten, per-
gest love-object and as the prototype of all later sonality development, social cognition, behavior
love-relations.” Drawing on this psychoanalytic problems, and indicators of emergent psychopa-
heritage, Bowlby (1969/1982, 1973, 1980) also thology. But researchers have also broadened their
enlisted formulations from evolutionary biology, inquiry to explore how security predicts later cog-
developmental psychology, and control systems nitive and language development, exploration and
theory to argue that a warm and continuous rela- play, curiosity, ego resiliency, math achievement,
tionship with a caregiver promotes psychological and even political ideology, extending the range
health and well-being throughout life in a manner of predictive correlates far beyond what Bowlby
that accords with the adaptive requirements of the originally envisioned. As Belsky and Cassidy
human species. In collaboration with Ainsworth (1994) mused, one might wonder whether there is
(1967, 1973), he proposed that differences in the anything with which attachment security is not as-
security of infant–mother attachment have sig- sociated. Comprehending this research literature
nificant long-term implications for later intimate thus requires reconsidering how and why early at-
relationships, self-understanding, and even risk for tachment security should be associated with later
psychopathology. development, as well as alternative models for why
Bowlby’s conceptual integration was pro- security might predict later functioning in direct
vocative, and with the validation of reliable and indirect ways. Doing so is important for inter-
348
, 16. Early Attachment and Later Development 349
preting research findings in a theoretically coher- attachment relationships, continuity and change
ent manner—and equally important for highlight- in parent–child interaction, and the dynamics of
ing the research designs that are likely to be most personality growth. Beyond theoretical breadth, of
informative for future studies of early attachment course, is the fact that subsequent attachment re-
and its sequelae. searchers have had their own ideas about the influ-
This chapter begins, therefore, with consid- ence of early attachment security, which they have
eration of alternative explanations for why a se- sought to harmonize with Bowlby’s formulations.
cure attachment should be associated with later These are all signs of a vibrant, generative
behavior, with a focus on attachment security in theory. Indeed, it can be argued that today the
the early years. Following this is a review of the proper role of Bowlby’s theory is not as a source
research examining these associations in the de- of orthodoxy for attachment theorists (much as
velopmental domains that have been best studied: Freud’s theory was treated in the early decades of
parent–child relationships, close relationships psychoanalysis), but rather as a foundation for new
with peers and other partners, personality, emo- thinking about early parent–child relationships.
tion regulation, self-concept, emotion understand- The problem this presents for contemporary re-
ing, social cognition, memory, and conscience. In searchers, however, is the proliferation of concep-
a final section, these results are discussed in light tual explanations for why early attachment should
of what we can conclude about how attachment (and sometimes should not) be associated with
security influences later developmental function- later development. Beyond the casual post hoc
ing, and which research approaches are most likely explanations offered by researchers for unexpect-
to elucidate this association in future studies. ed empirical findings, in other words, there have
grown from the foundation of Bowlby’s theory
various attachment minitheories, with somewhat
CONCEPTUAL PERSPECTIVES different views of the nature of the developmen-
tal influences arising from secure or insecure early
To an observer, it might appear surprising that it relationships.
would be necessary to begin this discussion by sort- In this section, therefore, the goal is to sum-
ing through the various conceptual explanations marize and evaluate several alternative views of
in the literature for why early attachment security the developmental influence of attachment that
should be associated with later development. After have become significant in contemporary attach-
all, wasn’t Bowlby’s theory clear on this issue? ment research. These approaches are discussed
The challenge facing contemporary attach- with respect to certain key conceptual questions.
ment researchers is not only Bowlby’s theory, but For which developmental domains is early security
also its generativity. Attachment theory was for- likely to be most important, and at what ages? How
mulated decades ago, at a time when scientific much should the effects of early attachment be ex-
understanding of infancy and early childhood pected to endure, and what mediators might affect
underestimated the cognitive and behavioral so- its continuing influence? What are the conditions
phistication of the child and the dynamics of early in which attachment should most influence later
parent–child relationships. There have also been development? Although most of these attachment
significant advances in behavioral ecology and minitheories do not provide clear answers to all of
evolutionary biology. It is natural that in efforts these conceptual questions, the purpose in posing
to keep the theory current with advancing knowl- them is to clarify our thinking about why early at-
edge, Bowlby’s heirs would expand, elucidate, and tachment should be developmentally provocative.
update his formulations in ways that he could not
anticipate. Furthermore, as would be expected of a
Internal Working Models
conceptually innovative approach, Bowlby’s the-
ory provides a conceptual umbrella for broad and One of Bowlby’s most heuristically powerful for-
narrow constructions of the developmental impact mulations is the view that attachment security in-
of attachment relationships. Grossmann (1999), fluences psychological growth through children’s
for example, has identified at least two different developing mental representations, or “internal
conceptualizations of “internal working models” working models” (IWMs), of the social world.
in Bowlby’s theory, and the breadth of the theory IWMs are based on infants’ expectations for the
offers explanations for developmental influences accessibility and responsiveness of their caregivers;
related to the biologically adaptive qualities of these expectations develop into broader represen-
, 350 III. ATTACHMENT IN INFANCY AND CHILDHOOD
tations of themselves, their attachment figures, resentations that change over time with a child’s
interpretations of their relational experiences, and conceptual growth. Thus our knowledge of young
decision rules about how to interact with others. children’s developing conceptual skills could es-
These mental representations not only enable tablish parameters for what we would expect to be
immediate forecasts of the caregiver’s responsive- true of their IWMs of self and relationships, espe-
ness, but develop into interpretive filters through cially early in life. Bretherton (1993; Bretherton
which children (and adults) reconstruct their un- & Munholland, 1999 and Chapter 5, this volume)
derstanding of new relationships and experiences pioneered this developmental approach by con-
in ways that are consistent with past experiences ceptualizing IWMs in relation to mental schemas
and expectations arising from secure or insecure and constructive memory processes, underscoring
attachments. As a consequence, children choose that regardless of their unconscious influences,
new partners and behave with them in ways that IWMs are based on consciously accessible cog-
are consistent with, and thus help to confirm, the nitive processes that change with development.
expectations created from earlier attachments. Elsewhere (Thompson, 2000, 2006), I have built
IWMs thus constitute the bridge between an in- on this formulation by offering a developmental
fant’s experience of sensitive or insensitive care account that associates the growth of IWMs with
and the development of beliefs and expectations allied conceptual advances in early childhood—
that affect subsequent experience in close relation- such as implicit memory and social expectations
ships. Young children also internalize conceptions in infancy, the development of event representa-
of themselves from early relational experience tion and episodic memory in early childhood, the
that form the basis for developing self-concept emergence of theory of mind and autobiographi-
and other self-referential beliefs. This concept cal memory in the preschool years, and the devel-
has been theoretically generative: Bretherton and opment of specific social-cognitive skills later in
Munholland (1999 and Chapter 5, this volume), childhood. Each of these well-understood concep-
Crittenden (1990), Main (1991), Sroufe and tual advances contributes to the representational
Fleeson (1988), and I (Thompson, 2006) have capacities of Bowlby’s IWM construct: under-
each offered contemporary extensions of Bowlby’s standing other people (including attachment fig-
concept of IWMs. ures), representing experience (especially in close
In this formulation, therefore, IWMs would relationships), self-concept, and understanding
be expected to be most directly associated with the how to relate socially to others. When a develop-
child’s capacities to create and maintain successful mental understanding of IWMs is linked to these
close relationships (with parents, peers, teachers, conceptual advances in the encoding, representa-
and others), establish a positive self-image, and tion, and memory of social experience, the growth
perhaps also develop constructive social repre- of attachment-related working models can be con-
sentations of people and of relationships. How- ceptualized more precisely and studied in relation
ever, because of the imprecision of Bowlby’s por- to current understanding of children’s cognitive
trayal of IWMs (which is a conceptual metaphor growth.
rather than a well-defined theoretical construct), There are several other implications of this
this concept has assumed far greater explanatory developmental formulation (Thompson, 2000,
breadth in attachment research to account for a 2006). First, IWMs are likely to change in response
widening array of developmental outcomes, such to new relational experiences and also during peri-
as proneness to stress, theory of mind, and ideo- ods of representational advance, such as in the tran-
logical values. Its use by the field has caused some sition to the symbolic representational capacities
to question whether IWMs constitute a “catch- of early childhood and the emergence of abstract
all, post-hoc explanation” for almost anything to thought in adolescence. These transitions render
which a secure attachment is found to be associ- IWMs more susceptible to revision, as new modes
ated (Belsky & Cassidy, 1994, p. 384). The in- of understanding can alter earlier representations
clusiveness of the contemporary IWM construct of relational experience. Second, the security or
has tended to expand with every new empirical insecurity incorporated into IWMs may have their
finding that is “explained” with reference to it, greatest influence on social and personality capaci-
which is a problem for attachment theory and the ties during those periods when these capacities are
discriminant validity of the attachment construct maturing most significantly. The IWMs associ-
(Thompson & Raikes, 2003). ated with a secure attachment are likely to influ-
One solution to this problem of underspeci- ence self-concept or emotion understanding most
ficity is to understand IWMs as developing rep- strongly in early childhood, for example, when
Early Attachment and Later Development
Familiar Questions, New Answers
ROSS A. THOMPSON
How does the child foreshadow the adult-to-be? methods for assessing the security of attachment
Philosophers, spiritualists, playwrights, and most for infants and young children, an enormous re-
recently behavioral scientists have sought to un- search literature emerged concerning the origins,
derstand how early dispositions and influences correlates, and consequences of secure and inse-
provide a foundation for adult personality. Among cure relationships. Guided by a general expecta-
the answers they have offered is the influence of tion that a secure attachment would predict better
early, close relationships. This view was eventu- later functioning, developmental researchers have
ally crystallized in Freud’s (1940/1963, p. 45) fa- explored the association between early security
mous dictum that the infant–mother relationship and later relations with parents, peers, friends, and
is “unique, without parallel, established unalter- other social partners, as well as with self-concept,
ably for a whole lifetime as the first and stron- competence in preschool and kindergarten, per-
gest love-object and as the prototype of all later sonality development, social cognition, behavior
love-relations.” Drawing on this psychoanalytic problems, and indicators of emergent psychopa-
heritage, Bowlby (1969/1982, 1973, 1980) also thology. But researchers have also broadened their
enlisted formulations from evolutionary biology, inquiry to explore how security predicts later cog-
developmental psychology, and control systems nitive and language development, exploration and
theory to argue that a warm and continuous rela- play, curiosity, ego resiliency, math achievement,
tionship with a caregiver promotes psychological and even political ideology, extending the range
health and well-being throughout life in a manner of predictive correlates far beyond what Bowlby
that accords with the adaptive requirements of the originally envisioned. As Belsky and Cassidy
human species. In collaboration with Ainsworth (1994) mused, one might wonder whether there is
(1967, 1973), he proposed that differences in the anything with which attachment security is not as-
security of infant–mother attachment have sig- sociated. Comprehending this research literature
nificant long-term implications for later intimate thus requires reconsidering how and why early at-
relationships, self-understanding, and even risk for tachment security should be associated with later
psychopathology. development, as well as alternative models for why
Bowlby’s conceptual integration was pro- security might predict later functioning in direct
vocative, and with the validation of reliable and indirect ways. Doing so is important for inter-
348
, 16. Early Attachment and Later Development 349
preting research findings in a theoretically coher- attachment relationships, continuity and change
ent manner—and equally important for highlight- in parent–child interaction, and the dynamics of
ing the research designs that are likely to be most personality growth. Beyond theoretical breadth, of
informative for future studies of early attachment course, is the fact that subsequent attachment re-
and its sequelae. searchers have had their own ideas about the influ-
This chapter begins, therefore, with consid- ence of early attachment security, which they have
eration of alternative explanations for why a se- sought to harmonize with Bowlby’s formulations.
cure attachment should be associated with later These are all signs of a vibrant, generative
behavior, with a focus on attachment security in theory. Indeed, it can be argued that today the
the early years. Following this is a review of the proper role of Bowlby’s theory is not as a source
research examining these associations in the de- of orthodoxy for attachment theorists (much as
velopmental domains that have been best studied: Freud’s theory was treated in the early decades of
parent–child relationships, close relationships psychoanalysis), but rather as a foundation for new
with peers and other partners, personality, emo- thinking about early parent–child relationships.
tion regulation, self-concept, emotion understand- The problem this presents for contemporary re-
ing, social cognition, memory, and conscience. In searchers, however, is the proliferation of concep-
a final section, these results are discussed in light tual explanations for why early attachment should
of what we can conclude about how attachment (and sometimes should not) be associated with
security influences later developmental function- later development. Beyond the casual post hoc
ing, and which research approaches are most likely explanations offered by researchers for unexpect-
to elucidate this association in future studies. ed empirical findings, in other words, there have
grown from the foundation of Bowlby’s theory
various attachment minitheories, with somewhat
CONCEPTUAL PERSPECTIVES different views of the nature of the developmen-
tal influences arising from secure or insecure early
To an observer, it might appear surprising that it relationships.
would be necessary to begin this discussion by sort- In this section, therefore, the goal is to sum-
ing through the various conceptual explanations marize and evaluate several alternative views of
in the literature for why early attachment security the developmental influence of attachment that
should be associated with later development. After have become significant in contemporary attach-
all, wasn’t Bowlby’s theory clear on this issue? ment research. These approaches are discussed
The challenge facing contemporary attach- with respect to certain key conceptual questions.
ment researchers is not only Bowlby’s theory, but For which developmental domains is early security
also its generativity. Attachment theory was for- likely to be most important, and at what ages? How
mulated decades ago, at a time when scientific much should the effects of early attachment be ex-
understanding of infancy and early childhood pected to endure, and what mediators might affect
underestimated the cognitive and behavioral so- its continuing influence? What are the conditions
phistication of the child and the dynamics of early in which attachment should most influence later
parent–child relationships. There have also been development? Although most of these attachment
significant advances in behavioral ecology and minitheories do not provide clear answers to all of
evolutionary biology. It is natural that in efforts these conceptual questions, the purpose in posing
to keep the theory current with advancing knowl- them is to clarify our thinking about why early at-
edge, Bowlby’s heirs would expand, elucidate, and tachment should be developmentally provocative.
update his formulations in ways that he could not
anticipate. Furthermore, as would be expected of a
Internal Working Models
conceptually innovative approach, Bowlby’s the-
ory provides a conceptual umbrella for broad and One of Bowlby’s most heuristically powerful for-
narrow constructions of the developmental impact mulations is the view that attachment security in-
of attachment relationships. Grossmann (1999), fluences psychological growth through children’s
for example, has identified at least two different developing mental representations, or “internal
conceptualizations of “internal working models” working models” (IWMs), of the social world.
in Bowlby’s theory, and the breadth of the theory IWMs are based on infants’ expectations for the
offers explanations for developmental influences accessibility and responsiveness of their caregivers;
related to the biologically adaptive qualities of these expectations develop into broader represen-
, 350 III. ATTACHMENT IN INFANCY AND CHILDHOOD
tations of themselves, their attachment figures, resentations that change over time with a child’s
interpretations of their relational experiences, and conceptual growth. Thus our knowledge of young
decision rules about how to interact with others. children’s developing conceptual skills could es-
These mental representations not only enable tablish parameters for what we would expect to be
immediate forecasts of the caregiver’s responsive- true of their IWMs of self and relationships, espe-
ness, but develop into interpretive filters through cially early in life. Bretherton (1993; Bretherton
which children (and adults) reconstruct their un- & Munholland, 1999 and Chapter 5, this volume)
derstanding of new relationships and experiences pioneered this developmental approach by con-
in ways that are consistent with past experiences ceptualizing IWMs in relation to mental schemas
and expectations arising from secure or insecure and constructive memory processes, underscoring
attachments. As a consequence, children choose that regardless of their unconscious influences,
new partners and behave with them in ways that IWMs are based on consciously accessible cog-
are consistent with, and thus help to confirm, the nitive processes that change with development.
expectations created from earlier attachments. Elsewhere (Thompson, 2000, 2006), I have built
IWMs thus constitute the bridge between an in- on this formulation by offering a developmental
fant’s experience of sensitive or insensitive care account that associates the growth of IWMs with
and the development of beliefs and expectations allied conceptual advances in early childhood—
that affect subsequent experience in close relation- such as implicit memory and social expectations
ships. Young children also internalize conceptions in infancy, the development of event representa-
of themselves from early relational experience tion and episodic memory in early childhood, the
that form the basis for developing self-concept emergence of theory of mind and autobiographi-
and other self-referential beliefs. This concept cal memory in the preschool years, and the devel-
has been theoretically generative: Bretherton and opment of specific social-cognitive skills later in
Munholland (1999 and Chapter 5, this volume), childhood. Each of these well-understood concep-
Crittenden (1990), Main (1991), Sroufe and tual advances contributes to the representational
Fleeson (1988), and I (Thompson, 2006) have capacities of Bowlby’s IWM construct: under-
each offered contemporary extensions of Bowlby’s standing other people (including attachment fig-
concept of IWMs. ures), representing experience (especially in close
In this formulation, therefore, IWMs would relationships), self-concept, and understanding
be expected to be most directly associated with the how to relate socially to others. When a develop-
child’s capacities to create and maintain successful mental understanding of IWMs is linked to these
close relationships (with parents, peers, teachers, conceptual advances in the encoding, representa-
and others), establish a positive self-image, and tion, and memory of social experience, the growth
perhaps also develop constructive social repre- of attachment-related working models can be con-
sentations of people and of relationships. How- ceptualized more precisely and studied in relation
ever, because of the imprecision of Bowlby’s por- to current understanding of children’s cognitive
trayal of IWMs (which is a conceptual metaphor growth.
rather than a well-defined theoretical construct), There are several other implications of this
this concept has assumed far greater explanatory developmental formulation (Thompson, 2000,
breadth in attachment research to account for a 2006). First, IWMs are likely to change in response
widening array of developmental outcomes, such to new relational experiences and also during peri-
as proneness to stress, theory of mind, and ideo- ods of representational advance, such as in the tran-
logical values. Its use by the field has caused some sition to the symbolic representational capacities
to question whether IWMs constitute a “catch- of early childhood and the emergence of abstract
all, post-hoc explanation” for almost anything to thought in adolescence. These transitions render
which a secure attachment is found to be associ- IWMs more susceptible to revision, as new modes
ated (Belsky & Cassidy, 1994, p. 384). The in- of understanding can alter earlier representations
clusiveness of the contemporary IWM construct of relational experience. Second, the security or
has tended to expand with every new empirical insecurity incorporated into IWMs may have their
finding that is “explained” with reference to it, greatest influence on social and personality capaci-
which is a problem for attachment theory and the ties during those periods when these capacities are
discriminant validity of the attachment construct maturing most significantly. The IWMs associ-
(Thompson & Raikes, 2003). ated with a secure attachment are likely to influ-
One solution to this problem of underspeci- ence self-concept or emotion understanding most
ficity is to understand IWMs as developing rep- strongly in early childhood, for example, when
