Attachment, the development of the self, and its pathology in personality disorders

Peter Fonagy

1. Introduction

Attachment theory concerns the nature of early experiences of children and the impact of these
experiences on aspects of later functioning of particular relevance to personality disorder. The
question we attempt to address here is how deprivation, in particular early trauma, comes to affect the
individual’s propensity to personality disorder. As part of this question we are naturally also
concerned to understand how such adverse consequences may be avoided. The key assumption made
by the invoking of attachment theory is that individual social behaviour may be understood in terms
of generic mental models of social relationships constructed by the individual. These models,
although constantly evolving and subject to modification, are strongly influenced by the child’s
experiences with the primary caregivers. Let us now turn to the details of the theory.

2. The nature of the attachment system

Attachment theory, developed by John Bowlby (1969, 1973, 1980), postulates a universal human
need to form close affectional bonds. It is a normative theory of how the ‘attachment system’
functions in all humans. Bowlby described attachment as a special type of social relationship,
paradigmatically between infant and caregiver, involving an affective bond. More significantly, it
may also be seen as the context within which the human infant learns to regulate emotion (Sroufe,
1990).

The stability of early childhood attachment patterns is well demonstrated. Mary Ainsworth and her
colleagues (Ainsworth, 1985; Ainsworth & Wittig, 1969; Ainsworth et al., 1978) developed a
procedure commonly known as the Strange Situation, which classifies infants and toddlers into one
of four attachment categories. Secure infants explore readily in the presence of the primary caregiver,
are anxious in the presence of the stranger, are distressed by their caregivers departure and brief
absence, rapidly seek contact with the caregiver following a brief period of separation, and are
reassured by renewed contact. The recovery from an over-aroused disorganised state is smooth and
carried to completion in the sense that the infant returns to exploration and play.

Some infants, who are usually made less anxious by separation, do not automatically seek proximity
with the caregiver on her return following separation and may show no preference for the caregiver
over the stranger; these infants are designated ‘Anxious/Avoidant’. A third category, the
‘Anxious/Resistant’ infant manifest impoverished exploration and play, tend to be highly distressed
by separation from the caregiver, but have great difficulty in settling after reunion showing
struggling, stiffness, or continued crying, or fuss in a passive way. The caregiver’s presence or
attempts at comforting fails to offer reassurance and their anxiety and anger appears to interfere with
their attempts to derive comfort through proximity. Both these insecure groups appear to be coping
with arousal and ambivalence through a precautious over-control of affect because they appear to be
uncertain in their expectation that the caregiver will do his or her part to modulate their emotional
arousal (Main & Weston, 1981; Sroufe, 1990).
It is generally held that the patterning of attachment related behaviour is underpinned by different
strategies adopted by children to regulate their emotional reactions. As affect regulation is acquired
with the help of the child’s primary caregiver, the child’s strategy will be inevitably a reflection of the
caregiver’s behaviour towards him/her. Secure infants’ behaviour is based on the experience of well
co-ordinated, positive interactions where the caregiver is rarely over-arousing and is able to
restabilise the child’s spontaneously emerging disorganising emotional responses. Therefore, they
remain relatively organised in stressful situations. Negative emotions are not seen as threatening in
and of themselves but are regarded by the infant as serving a communicative function (Grossman et
al., 1986; Sroufe, 1979, 1996).

By contrast, Anxious/Avoidantly attached children are presumed to have experiences when their
emotional arousal was not restabilised by the caregiver because of personal or social pressures on the
caregiver and an associated mild neglect or even resentment of the child. The same expectations may
arise in children who were over aroused through intrusive parenting, therefore they over-regulate
their affect and steer away from situations that are likely to be emotionally arousing.
Anxious/Resistantly attached children under-regulate, heightening their expression of distress
possibly in an effort to elicit the expectable response of the caregiver. These children have low
thresholds for threat and may become preoccupied with having contact with the caregiver, but show
signs of frustration regarding this contact even when it is available (Sroufe, 1996).

A fourth group of infants appear to exhibit a range of seemingly undirected behavioural responses
giving the impression of disorganisation and disorientation (Main & Solomon, 1990). Infants who
manifest freezing, handclapping, headbanging, the wish to escape from the situation even in the
presence of the caregiver, are referred to as ‘Disorganised/ Disoriented’. It is generally held that for
such infants the caregiver has served both as a source of fear and as a source of reassurance, thus the
arousal of the attachment behavioural system produces strong conflicting motivations. Not
surprisingly, a history of severe neglect or physical or sexual abuse is often associated with the
manifestation of this pattern (Cicchetti & Beeghly, 1987; Main & Hesse 1990).

2.1 Determinants of attachment security

It is beyond the scope of this paper to consider in detail the rich literature on determinants of infant
security. There are many excellent reviews available, notably by Belsky (Belsky et al., 1995). Clearly
genetic transmission may account for some component of the prediction from parental attachment
status to the child’s security of attachment (van Ijzendoorn, 1992). The influence of temperament on
attachment security is controversial, but the balance of the evidence is now against the
appropriateness of a temperamental account (Kagan, 1982; Lamb et al., 1984). There is little
evidence that distress-prone infants become anxious-resistant babies (van den Boom, 1990).
Temperament changes in the first year of life (Belsky et al., 1991) and the attachment pattern of a
child to his two parents is often inconsistent (Fox et al., 1991) and appears to be dependent on the
internal working model of each parent (Steele et al., 1996).

The quality of maternal care has been repeatedly shown to predict infant security. The sensitive
responsiveness of the parent is traditionally regarded as the most important determinant of attachment
security in the infant (Isabella, 1993; Isabella & Belsky, 1991). The parameters assessed include:
ratings of maternal sensitivity (e.g. Cox et al., 1992; Isabella, 1993), prompt responsiveness to
distress (Del Carmen et al., 1993), moderate stimulation (Belsky et al., 1984), non-intrusiveness
(Malatesta et al., 1986), interactional synchrony (Isabella et al., 1989), warmth, involvement and
responsiveness (O'Connor et al., 1992). These associations have been strengthened by findings from
experimental studies, where the enhancement of maternal sensitivity has been shown to increase the
proportion of secure infants in high-risk populations (van den Boom, 1995). Similar parameters have
been predictive for fathers (Cox et al., 1992) and for professional caregivers (Goosens & van
Ijzendoorn, 1990).

Negative parental personality traits are associated with insecurity in many studies, although by no
means all (Zeanah et al., 1993). This has been shown for anxiety (Del Carmen et al., 1993),
aggression (Maslin & Bates, 1983) and suspicion (Egeland and Farber, 1984). Parental
psychopathology is also found to be a risk factor in some studies (Campbell et al., 1993). Of the
contextual factors, support from the partner (Goldberg and Easterbrooks, 1984) and from others in the
mother's environment (Crnic et al., 1983) appear important. The strength of these associations is
reinforced by experimental studies where social support was systematically manipulated (Lyons-Ruth
et al., 1990; Jacobson & Frye, 1991; Lieberman et al., 1991).

These predictors of infant security are correlated to one another and are all likely to be unequally
distributed across socio-economic groups. It is known that socio-economic status and other indicators
of social deprivation are linked to both infant and adult classifications (e.g. Ward & Carlson, 1995;
van Ijzendoorn & Kroonenberg, 1988; Crittenden et al., 1991; Zeanah et al., 1993). Poor parenting
skills and the maltreatment of children are more common in families suffering economic hardship
(Gabarino, 1992). Insecure classification is more common in deprived groups. Maltreatment of
children, strongly associated with economic deprivation (Belsky, 1993) is most likely to be associated
with the disorganised/disoriented pattern of infant attachment.

2.2 The continuity of patterns of attachment

Bowlby proposed that the quality of childhood relationships with the caregivers results in internal
representations or working models of the self and others that provide prototypes for later social
relations. Internal working models are mental schemata, where expectations about the behaviour of a
particular individual toward the self are aggregated. The expectations are themselves abstractions
based on repeated interactions of specific types with that individual. If the child’s physical injury is
quickly dealt with, sources of unhappiness are rapidly addressed, the child will develop the legitimate
expectation that, with that person at least, his distress is likely to be met by reassurance and
comforting. The internal working model is the result of a natural process of abstraction of the
invariant features from diverse social situations with a particular individual (Stern, 1985, 1994).

Such internal models of attachment remain relatively stable across the lifespan (Collins & Read,
1994). Secure children, with the benefit of well-regulated caregiver-infant relationships behind them,
are expected to evolve positive expectations concerning their exploratory competence, to achieve a
reliable capacity for modulation of arousal, a good capacity for communication within relationships
and, above all, confidence in the ongoing availability of the caregiver. Early experiences of flexible
access to feelings is regarded as formative by attachment theorists, enabling secure children both to
maximise the opportunities presented to them by the environment and draw on socially supportive
relationships. The autonomous sense of self emerges fully from secure parent-infant relationships
(Emde & Buchsbaum, 1990; Lieberman and Pawl, 1990; Fonagy et al., 1995). The increased control
of the secure child permits it to move towards the ownership of inner experience and come to
recognise the self as competent in eliciting regulatory assistance, to develop metacognitive control
and to achieve an understanding of self and others as intentional agents whose behaviour is organised
by mental states, thoughts, feelings, beliefs and desires (Sroufe, 1990; Fonagy et al., 1995).

2.3 Prediction from adult attachment measures

The stability of these attachment assessments are dramatically illustrated by longitudinal studies of
infants assessed with the Strange Situation and followed up in adolescence or young adulthood with
the Adult Attachment Interview (AAI) (George et al., 1985). The AAI asks subjects about childhood
attachment relationships and the meaning which an individual currently gives to attachment
experiences. The instrument is rated according to the scoring system developed by Main & Goldwyn
(1994) which classifies individuals into Secure/Autonomous, Insecure/Dismissing or Avoidant,
Insecure/Preoccupied or Resistant or Unresolved/Disorganised with respect to loss or trauma,
categories according to the structural qualities of their narratives of early experiences. (While
autonomous individuals clearly value attachment relationships and regard these as formative, insecure
individuals are poor at integrating memories of experience with their assessment of the meaning of
that experience. Those dismissing of attachment deny or devalue early relationships. Preoccupied
individuals tend to be confused, and angry or passive in their current relationships with their parents
and others). Two studies (Hamilton, 1994; Waters, et al., 1995) have shown a 68-75%
correspondence between attachment classifications in infancy and classifications in adulthood. This
work speaks to the remarkable stability of attachment classifications across the lifespan. Similar
findings are beginning to emerge using other measures of attachment in adults (Hazan & Zeifman,
1994).

3. The transgenerational transmission of attachment patterns

There is further important evidence that attachment relationships may play a key role in the
transgenerational transmission of hardship and deprivation. Individuals categorised as secure are 3 or
4 times more likely to have children who are securely attached to them (van Ijzendoorn, 1995). This
turns out to be true even in prospective studies where parental attachment is assessed before the birth
of the child (Benoit & Parker, 1994; Fonagy et al., 1991; Radojevic, 1992; Steele et al., 1996; Ward
& Carlson,1995). These findings also emphasise the importance of quality of parenting in
determining the child’s attachment classification. The findings from our lab suggest that parental
attachment patterns predict variance in addition to temperament measures or contextual factors, such
as experience, social support, marital relationship, psychopathology and personality (Steele, Steele &
Fonagy, in preparation). If attachment is linked to personality disorder we may anticipate a
substantial overlap between determinants of infant security and long-term predictors of criminality.

3.1 Attachment and mentalising

A compelling model for the transmission of secure attachment, which has moved the field beyond a
simple view of caregiver sensitivity, was suggested by Mary Main (1991) in her seminal chapter on
metacognitive monitoring and singular versus multiple models of attachment. Main (1991) showed
that the absence of metacognitive capacity, the inability to "understand the merely representational
nature of their own (and others') thinking" (p. 128), makes infants and toddlers vulnerable to the
inconsistency of the caregiver's behaviour. They are unable to step beyond the immediate reality of
experience and grasp the distinction between immediate experience and the mental state which might
underpin it. Main drew our attention to the development in the child of the mental state that Dennett
(1987) called "the intentional stance". Dennett stressed that human beings are perhaps unique in
trying to understand each other in terms of mental states: thoughts, feelings, desires, beliefs, in order
to make sense of and, even more important, to anticipate each other’s actions. It is self-evident that by
attributing an emotional or cognitive state to others we make their behaviour explicable to ourselves.
If the child is able to attribute a withdrawing, non-responsive mother's apparently rejecting behaviour
to her emotional state of depression, rather than to himself as bad and unstimulating, the child is
protected from, perhaps permanent, narcissistic injury. Perhaps even more central is the child’s
capacity to develop representations of the mental states, emotional and cognitive, which organise
his/her behaviour toward the caregiver.

We attempted to operationalise individual differences in adults' metacognitive capacities which we

believe might help to fill the 'transmission gap'. We were curious to know if the extent of self-
reflective observations about the mental states of self and others in Adult Attachment Interview (AAI)
narratives could predict infant security. We chose the term "reflective self-scale" (as opposed to "self-
reflection scale") to underscore that we were concerned about the clarity of the individual's
representation of the mental states of others as well as the representation of their own mental state.

Consistent with our expectation, reflective-self ratings were reliable (intraclass r = .8 and above) and
provided a good pre-natal prediction of the Strange Situation behaviour of the child. Both fathers and
mothers who were rated to be high in this capacity were three or four times more likely to have
secure children than parents whose reflective capacity was poor (Fonagy et al., 1991).

The capacity for metacognitive control may be particularly important when the child is exposed to
unfavourable interaction patterns, in the extreme, abuse or trauma. For example, in the absence of the
capacity to represent ideas as ideas, the child is forced to accept the implication of parental rejection,
and adopt a negative view of himself. A child who has the capacity to conceive of the mental state of
the other can also conceive of the possibility that the parent's rejection of him or her may be based on
false beliefs, and therefore is able to moderate the impact of negative experience.

We examined this issue by administering a brief structured interview to parents in our sample, 18
months after they had completed the Adult Attachment Interview, concerning a number of simple
indicators of family stress and deprivation which had been reported in past studies to increase
dramatically the probability of adverse outcome, including, in a recent study, the likelihood of
insecure infant attachment. These indicators included: single parent families residing separately,
overcrowding, paternal unemployment, etc. We divided our sample into those who had reported
significant experience of deprivation (more than 2 items) and those who had not. Our prediction was
that mothers in the deprived group would be far more likely to have children securely attached to
them if their reflective-self rating (metacognitive capacity) was high.

10 out of 10 of the mothers in the deprived group with high reflective-self ratings had children who
were secure with them, whereas only 1 out of 17 of deprived mothers with low ratings did so.
Reflective-self function seemed to be a far less important predictor for the non-deprived group. Our
findings imply that the intergenerational replication of early negative experiences may be aborted, the
cycle of disadvantage interrupted, if the caregiver acquires a capacity to fully represent and reflect on
mental experience (Fonagy et al., 1994).

3.2 Metacognitive monitoring and the development of the self

Metacognitive monitoring completes one aspect of the intergenerational cycle. Not only are parents
high in reflective capacity more likely to promote secure attachment in the child, particularly if their
own childhood experiences were adverse, but also secure attachment may be a key precursor of
robust reflective capacity (Fonagy et al., 1995).

In London, we have collected cross-sectional data from 3-5 year olds which appears to indicate a
strong correlation between security on a projective measure of attachment (the SAT) and the early
development of a theory of mind, using the belief-desire reasoning task. We found the children who
were rated Secure on the Separation/Anxiety test were more likely to pass the theory of mind tasks
(Fonagy et al., 1997). In a longitudinal study, we found that of 92 children, the 59 who had passed the
task at 5 years, 66% were secure at one year with their mother. Of the 29 who failed, only 31% had
been secure. Attachment security to father was less significantly associated with greater competence
at this task. There was clear indication that the reflective self function of mother was associated with
the child’s success. 80% of children whose mothers were above the median in reflective self function
passed, whereas only 56% of those whose mothers were below did so.

These results suggest that the parents' capacity to observe the child's mind facilitates the child's
general understanding of minds mediated by secure attachment. The availability of a reflective
caregiver increases the likelihood of the child's secure attachment which, in turn, facilitates the
development of theory of mind. Throughout these studies we assume that a secure attachment
relationship provides a congenial context for the child to explore the mind of the caregiver, and, as
the philosopher Hegel (1807) taught us, it is only through getting to know the mind of the other that
the child develops full appreciation of the nature of mental states. The process is intersubjective: the
child gets to know the caregiver's mind as the caregiver endeavours to understand and contain the
mental state of the child.

The child perceives in the caregiver's behaviour not only her stance of reflectiveness which he infers
in order to account for her behaviour, but also he perceives in the caregiver's stance an image of
himself as mentalizing, desiring and believing. He sees that the caregiver represents him as an
intentional being. It is this representation which is internalised to form the self. "I think therefore I
am" will not do as a psychodynamic model of the birth of the self; "She thinks of me as thinking and
therefore I exist as a thinker" comes perhaps closer to the truth.

If the caregiver's reflective capacity enabled her accurately to picture the infant's intentional stance,
the infant will have the opportunity to "find itself in the other" as a mentalizing individual. If the
caregiver's capacity is lacking in this regard, the version of itself that the infant will encounter will be
an individual conceived of as thinking in terms of physical reality rather than mental states.

4. Some speculations about pathological development based on the dialectic model

The fundamental need of every infant is to find his mind, his intentional state, in the mind of the
object . For the infant, internalisation of this image performs the function of "containment", which
Winnicott has written of as "giving back to the baby the baby's own self" (Winnicott, 1967, p. 33) .
Failure of this function leads to a desperate search for alternative ways of containing thoughts and the
intense feelings they engender.

The search for alternative ways of mental containment may, we suggest, give rise to many
pathological solutions, including taking the mind of the other, with its distorted, absent or malign
picture of the child, as part of the child's own sense of identity. Winnicott (1967) wrote: "What does
the baby see when he or she looks at the mother's face? ...ordinarily, the mother is looking at the baby
and what she looks like is related to what she sees there... [but what of] the baby whose mother
reflects her own mood or, worse still, the rigidity of her own defences... They look and they do not
see themselves... what is seen is the mother's face" (p. 27).

This picture then becomes the germ of a potentially persecutory object which is lodged in the self, but
is alien and unassimilable. There will be a desperate wish for separation in the hope of establishing an
autonomous identity or existence. However, tragically, this identity is centred around a mental state
which cannot reflect the changing emotional and cognitive states of the individual, because it is based
on an archaic representation of the other, rather than the thinking and feeling self as seen by the other.
Paradoxically, where the child's search for mirroring or containment has failed, the later striving for
separation will only produce a movement towards fusion. The more the person attempts to become
himself, the closer he moves towards becoming his object, because the latter is part of the self-
structure. This in our view accounts for the familiar oscillation of borderline patients, between the
struggle for independence and the terrifying wish for extreme closeness and fantasised union.
Developmentally, a crisis arises when the external demand for separateness becomes irresistible, in
late adolescence and early adulthood. At this time, self-destructive and (in the extreme) suicidal
behaviour is perceived as the only feasible solution to an insoluble dilemma: the freeing of the self
from the other through the destruction of the other within the self.

In some individuals, for whom separateness is a chronic problem, we assume that the experience of
self-hood can only be achieved through finding a physical other onto whom the other within the self
can be projected. Naturally, this increases the individual’s need for the physical presence of the
object. Thus, many such individuals experience considerable difficulty in leaving home and if they
finally achieve physical separation, they can only do so by finding an alternative and comparable
figure onto whom the other within the self may be projected. If the other dies, or abandons the
individual, a pathological mourning process may be initiated whereby the person feels compelled to
maintain a live picture of the other, in order to retain the integrity of the self.

Another possible outcome of poor development of the psychological self, with consequent conflicts
over separation, is that the body may be used to contain and enact mental states. In these cases the
child's own body comes to serve the function of metarepresentation of feelings, ideas and wishes.
Violence towards the body of the self (e.g. self cutting) or that of the other (apparently unprovoked
aggression or ‘mindless violence’ may be a ways of ‘controlling’ mental states which are invested in
bodily states (e.g. the mother seen as part of ones own body) or destroying ‘ideas’ experienced as
within the body of the other. In other young children the search for the psychological self in the other
may lead to the physical image of the object being internalised as part of the child's identity. In
extreme cases, this may result in gender identity disorder (Fonagy & Target, 1995).

If the child finds no alternative interpersonal context where he is conceived of as mentalizing his
potential in this regard will not be fulfilled. In cases of abusive, hostile or simply totally vacuous
relationship with the caregiver, the infant may deliberately turn away from the mentalizing object
because the contemplation of the object's mind is overwhelming as it harbours frankly hostile
intentions toward the infant's self. This may lead to a widespread avoidance of mental states which
further reduces the chance of identifying and establishing intimate links with an understanding object.

As studies of resilient children suggest, even a single secure/understanding relationship may be

sufficient for the development of reflective processes and may 'save' the child. Metacognitive
monitoring is biologically prepared and will spontaneously emerge unless it's development is
inhibited by the dual disadvantage of the absence of a safe relationship and the experience of
maltreatment in the context of an intimate relationship. We do not anticipate that trauma outside of
the context of an attachment bond would have pervasive inhibitory effects on mentalizing. It is
because the theory of mind or, more broadly, and reflective self function evolve in the context of
intense interpersonal relationships, that the fear of the mind of another can have such devastating
consequences on the emergence of social understanding. To illustrate the clinical relevance of this
model, it may be helpful to consider borderline personality disorder from the point of view of
attachment theory.

4.1 A transgenerational model of borderline personality disorder

Although accurate figures are hard to come by and vary across studies, considerable evidence has
accumulated to support the contention that child abuse is transmitted across generations. Oliver
(1993), in his recent review of 60 studies, mainly from the United States and the UK, concluded that
approximately one third of child victims of abuse grow up to continue a pattern of seriously inept,
neglectful or abusive child-rearing as parents. Research has documented that a specific link exists
between the history of childhood maltreatment and borderline personality disorder and sexual abuse
is especially implicated. In brief, as infants and children, borderline individuals frequently have
caretakers who are themselves within the so called 'borderline spectrum' of severely personality
disordered individuals. The social inheritance aspect of BPD may be an important clue in our
understanding of the disorder.

George Moran, Mary Target and I (Fonagy et al., 1993) have put forward an attachment theory
formulation of severe narcissistic and borderline states based on epidemiological findings of the
association of severe personality disorder and a history of childhood maltreatment and sexual abuse.
We proposed that borderline individuals are those victims of childhood (sexual) abuse who coped by
refusing to conceive of the contents of their caregiver's mind and thus successfully avoided having to
think about their caregiver's wish to harm them. They go on to defensively disrupt their capacity to
depict feelings and thoughts in themselves and in others. This leaves them to operate upon inaccurate
and schematic impressions of thoughts and feelings and they are thus immensely vulnerable in all
intimate relationships.

Many of the symptoms of BPD individuals may be understood in terms of a defensive strategy of
disabling mentalizing or metacognitive capacity:

1. Their failure to take into consideration the listener's current mental state makes their associations
hard to follow.

2. The absence of concern for the other which may manifest as extreme violence and cruelty, arises
because of the lack of a compelling representation of suffering in the mind of the other. A key
moderator of aggression is therefore absent. The lack of reflective capacity in conjunction with a
hostile world view may predispose individuals to child maltreatment but such inhibition may be a
necessary component of all violence against persons. Military training has the apparent and explicit
aim of fashioning men into machines and the enemy into an inanimate or sub-human object. Seeing
the other as imbued with thought and feeling is very likely imposes a break.

3. Their fragile sense of self (identity diffusion to use Kernberg's term), may be a consequence of
their failure to represent their own feelings, beliefs and desires with sufficient clarity to provide them
with a core sense of themselves as a functioning mental entity. This leaves them with overwhelming
fears of mental disintegration and a desperately fragile sense of self.

4. Such patient's mental image of object remains at the immediate context dependent level of primary
representations - he/she will need the object as they are and will experience substantial difficulties
when confronted with change.

5. Absence of prominence 'as if' in the transference requires meta representations, the capacity to
entertain a belief whilst at the same time knowing it to be false. Psychotherapy requires such pretence
and it's absence manifests as so called 'acting out' of the transference.

4.2 The association of attachment status and borderline states

In an ongoing study (Fonagy et al., 1996) we administered AAI's to a sample of 85 consecutively
admitted non-psychotic inpatients at the Cassell Hospital in London, which is run along the principles
of a psychoanalytic therapeutic community. About 40% of the patients met diagnostic criteria for
borderline personality disorder (BPD) on the basis of a structured interview (SCID-II).

The distribution of AAI classifications arrived at totally independently of the diagnostic process, did
not distinguish well Borderline Personality Disorder (BPD) from other personality disorder diagnoses
but the number of entangled (particularly E3) classifications were well above the number which
would be expected by chance (75%).

Borderline patients' interviews were, however, differentiated by a combination of 3 characteristics:

(1) higher prevalence of sexual abuse reported in the AAI narratives,

(2) significantly lower ratings on the reflective self-function scale,

(3) a significantly higher rating on the lack of resolution of abuse, but not loss scale of the AAI.

Further, there was a significant interaction between abuse and RSF: individuals with experience of
abuse who had low RSF were very likely to have a diagnosis of BPD.

These findings are consistent with our assumption that individuals with experience of severe
maltreatment in childhood who respond to this experience by an inhibition of reflective self function
are less likely to resolve this abuse, and are more likely to manifest borderline psychopathology.

Childhood maltreatment may or may not have long term sequelae and the determinants of the
outcome are only partially understood. Here we propose that if children are maltreated but they have
access to a meaningful attachment relationship which provides the intersubjective bases for the
development of mentalising capacity, they will be able to resolve (work through) their experience and
the outcome of the abuse will not be one of severe personality disorder. We do not expect that their
reflective processes will protect them from episodic psychiatric disorder, such as depression, and
epidemiological data suggests that victims of childhood maltreatment are at an elevated risk for many
forms of [Axis-I] disorder.

However, if the maltreated child has no social support of sufficient strength and intensity for an
attachment bond to develop which could provide the context for the acquisition of a reliable capacity
to envisage the psychological state of the other in intense interpersonal relationships, then the
experience of abuse will not be reflected on or resolved. Naturally, the unresolved experience of
abuse diminishes the likelihood of meaningful relationships which, in a self-perpetuating way, further
reduces the likelihood of a satisfactory resolution of the disturbing experience through the use of
reflective processes. In fact a pattern may be established whereby suspicion and distrust generalises
and leads to a turning away from the mental state of most significant objects and an apparent
'decoupling' of the 'mentalizing module' leaving the person bereft of human contact. This may
account for the 'neediness' of borderline personality disordered individuals; yet no sooner do they
become involved with another then the malfunctioning of their inhibited mentalising capacity leads
them into terrifying interpersonal confusion and chaos. Within intense relationships their inadequate
mentalising function rapidly fails them, they regress to the intersubjective state of the development of
mental representation and they are no longer able to differentiate their own mental representations
from those of others and both of these from actuality. These processes combine and they become
terrorised by their own thoughts about the other experienced (via projection) in the other, particularly
their aggressive impulses and fantasies ; these become crippling and most commonly they reject or
arrange to be rejected by their object. Psychoanalysis or psychotherapy can break the vicious cycle by
reinforcing reflective capacity.

5. Crime, violence and attachment

As with borderline patients, a history of maltreatment is present in 80-90% of juvenile offenders and
approximately a quarter of those with histories of severe maltreatment are likely to have criminal
convictions (e.g. Taylor, 1986). We have suggested that attachment to individuals as well as social
institutions may be critical in reducing the risk of delinquency and adjustment processes are severely
disrupted by childhood maltreatment. More specifically, if attachment to the primary caregiver is
intimately linked to the acquisition of reflective capacity (see section on moral development, above),
the latter may be a key mediator in predisposing an individual to criminality, particularly to violent
offences. We may suppose that those individuals, who were never exposed to interpersonal
relationships where the acquisition of a reflective capacity would have been facilitated, or who were
exposed to caregiving environments where their only route to adaptation was the inhibition of
mentalizing, are most likely to develop insecure attachments and manifest low reflective capacities,
thus removing essential inhibitions on criminal activities. The capacity to envision the mental state of
the potential victim may be essential in preventing us from deliberately harming other members of
our social group (or species).

To put these ideas to a test, Levinson & Fonagy (in preparation) collected AAI’s from 22 prisoners
(convicted or on remand with diagnosable psychiatric disorder) and matched them with 2 control
groups on age, gender, social class and IQ: 1) a psychiatric inpatient control group matched for
diagnoses (Axis I/II) and 2) a normal control group recruited from a medical outpatient department.
The findings may be summarised as follows:

(1) There was significantly more secure attachments in the normal control group but the two clinical
groups did not differ in terms of overall level of security.

(2) 36% of the prison group vs. 14% of the psychiatric group were classified as ‘Dismissing’ with
normal controls in between (23%).

(3) 45% of the prisoners vs. 64% of psychiatric controls were classified as ‘Preoccupied’ with only
14% of non-criminal controls receiving this classification.

(4) 82% of psychiatric patients but only 36% of prisoners and 0% of non-clinical controls received
‘Unresolved’ classifications.

(5) 82% of prisoners and only 36% of psychiatric patients were rated as having been abused with
only 4% of normal controls. (2/3 of abuse was physical, 1/3 sexual in both clinical groups).

(6) Neglect was more prevalent in the prison group but rejection was more frequently reported by
psychiatric patients.

(7) Current anger with attachment figures was dominant in psychiatric patients but relatively more
among prisoners.
(8) Prisoners had significantly lower ratings on the reflective function scale (RSF) than either
psychiatric patients or those from the non-clinical group, but RSF ratings of normals was still
significantly higher than those of psychiatric patients.

(9) When the prison group was split into those with violent index offences (murder, malicious
wounding, GBH, armed robbery, indecent assault to child), vs. non violent ones (possession,
importation, obtaining property by deception, theft, handling stolen goods) the rating on
reflectiveness of the former group was found to be significantly lower than the latter.

This pattern of results is consistent with our assumption that criminality arises in the context of weak
bonding with individuals and social institutions and the relatively ready dismissal of attachment
objects. Criminal behaviour may be seen as a socially maladaptive form of resolving trauma and
abuse (which was almost ubiquitous in our small sample). Violent acts are committed in place of
experienced anger concerning neglect, rejection and maltreatment. Committing antisocial acts is
facilitated by a non-reflective stance of the victim which may be of particular significance in cases
where the victim is clearly identifiable as in violent acts against another person.

This is only a pilot investigation, but the results are promising to the extent that they link attachment
related narratives to the nature of the offence committed. Naturally, an important alternative account
to the one proposed here may be that it was these crimes which caused the disorganisation of the
attachment system and it was the psychological impact of crime which permeated the interviews of
the violent group. The less serious offences may have made less impact on the representation of
relationships.

6. Psychotherapy and mentalising

Clinical psychoanalysis inevitably deals with individuals whose past experience has left them
vulnerable to current stress and the repetition of adverse early experiences. The treatment imposes a
non-pragmatic elaborative, mentalistic stance. This enhances the development of reflective self
function and may in the long run enhance the psychic resilience of individuals in a generic way,
providing them with improved control over their system of representation of relationships. It equips
them with a kind of self-righting capacity where through being able to operate on their
representational models, the latter can become an object of review and change. Such gradual and
constant adjustments facilitate the development of an internal world where the behaviour of others
may be experienced as understandable, meaningful, predictable and characteristically human. This
reduces the need for splitting of frightening and incoherent mental representations of mental states,
and new experiences of other minds can more readily be integrated into the framework of past
relationship representations.

The abused child, evading the mental world, never acquires adequate meta-control over the
representational world of internal working models. Unhelpful models of relationship patterns emerge
frequently and the internal world of the child and adult comes to be dominated by negative affect.
The individual's enhanced suspiciousness of human motives reinforces his/her strategy to forego
mentalizing, thus further distorting the normal development of a reflective function. Caught in a
vicious cycle of paranoid anxiety and exaggerated defensive manoeuvres, the individual becomes
inextricably entangled into an internal world dominated by dangerous, evil and above all mindless
objects. He has abnegated the very process which could extract him from his predicament, the
capacity to reflect on mental states.
Psychotherapeutic treatment in general, and psychoanalytic treatment in particular, compels the
patient's mind to focus on the mental state of a benevolent other, that of the therapist. The frequent
and consistent interpretation of the mental state of both analyst and patient (i.e. the interpretation of
the transference in the broadest sense) is then desirable, if not essential, if the inhibition on this aspect
of mental function is to be lifted. Over a prolonged time period, diverse interpretations concerning the
patient's perception of the analytic relationship would enable him to attempt to create a mental
representation both of himself and of his analyst, as thinking and feeling. This could then form the
core of a sense of himself with a capacity to represent ideas and meanings, and create the basis for the
bond that ultimately permits independent existence.

7. Summary

The paper puts forward an attachment theory model of severe personality disturbance. It is suggested
that
(1) secure attachment is the basis of the acquisition of metacognitive or mentalizing capacity;
(2) the caregiver’s capacity to mentalize may foster the child’s bonding with the parent;
(3) maltreatment may undermine the acquisition of a mentalizing capacity;
(4) symptoms of borderline personality disorder may arise as a consequence of inhibited mentalizing;
(5) violent crime and anti-social personality disorder may be possible because the capacity to reflect
upon the mental states of the victim is compromised;
(6) psychotherapeutic work may facilitate the reactivation of this inhibited capacity.

References

Ainsworth M.D.S. & Wittig B.A. (1969). Attachment and exploratory behavior of one-year-olds in a
strange situation. In B.M. Foss (Ed.), Determinants of infant behavior (pp. 113-136). London:
Methuen.

Ainsworth M.D.S., (1985). Attachments across the lifespan. Bulletin of the New York Academy of
Medicine, 61, 792-812.

Ainsworth M.D.S., Blehar M.C., Waters E. & Wall S. (1978). Patterns of attachment: A
psychological study of the Strange Situation. Hillsdale, NJ: Erlbaum.

Belsky J. (1993). Etiology of child maltreatment: A developmental-ecological analysis.

Psychological Bulletin, 114, 413-434.

Belsky J., Fish M. & Isabella R. (1991). Continuity and discontinuity in infant negative and positive
emotionality: Family antecedent and attachment consequences. Developmental Psychology, 27, 421-
431.

Belsky J., Rosenberger K. & Crnic C. (1995). The origins of attachment security: "Classical" and
contextual determinants. In S. Goldberg, R. Muir, & J. Kerr (Eds.), John Bowlby's Attachment
Theory: Historical, Clinical and Social Significance (pp. 153-184). Hillsdale, NJ: Analytic Press.

Belsky J., Rovine M. & Taylor D. G. (1984). The Pennsylvania Infant and Family Development
Project, III: The origins of individual differences in infant-mother attachment: Maternal and infant
contributions. Child Development, 55, 718-728.
Benoit D. & Parker K.C.H. (1994). Stability and transmission of attachment across three generations.
Child Development, 65, 1444-1457.

Bowlby J. (1969). Attachment and Loss. Vol. 1: Attachment. London: Hogarth Press (2nd ed.: New
York: Viking Penguin, 1984) (trad. it.: Attaccamento e perdita. Vol. 1: L'attaccamento alla madre.
Torino: Boringhieri, 1976 [I ed.], 1989 [II ed.]).

Bowlby J. (1973). Attachment and Loss. Vol. 2: Separation: Anxiety and Anger. London: Hogarth
Press (trad. it.: Attaccamento e perdita. Vol. 2: La separazione dalla madre. Torino: Boringhieri,
1978).

Bowlby J. (1980). Attachment and Loss. Vol. 3: Loss: Sadness and Depression. London: Hogarth
Press (trad. it.: Attaccamento e perdita. Vol. 3: La perdita della madre. Torino: Boringhieri, 1983).

Campbell S.B., Cohn J.F., Meyers T.A., Ross S. & Flanagan C. (1993). Chronicity of maternal
depression and mother-infant interaction. Presented at Biennial meeting of the Society for Research
in Child Development, New Orleans.

Cicchetti D. & Beeghly M. (1987). Symbolic development in maltreated youngsters: An

organizational perspective. In D. Cicchetti & M. Beeghly (Eds.), Atypical symbolic development. New
Directions for Child Development (Vol. 36, pp. 5-29). San Francisco: Jossey-Bass.

Collins N.L., & Read S.J. (1994). Representations of attachment: The structure and function of
working models. In K. Bartholomew & D. Perlman (Eds.), Advances in Personal Relationships Vol
5: Attachment Process in Adulthood (pp. 53-90). London: Jessica Kingsley Publishers.

Cox M., Owen T., Henderson V. & Margand N. (1992). Prediction of infant-father and infant-mother
attachment. Developmental Psychology, 28, 474-483.

Crittenden P.M., Partridge M.F. & Claussen A.H. (1991). Family patterns of relationship in
normative and dysfunctional families. Development and Psychopathology, 3, 491-449.

Crnic K.A., Greenberg M.T., Ragozin A.S., Robinson N.M. & Basham R.B. (1983). Effects of stress
and social support on mothers and premature and full-term infants. Child Development, 54, 209-217.

Del Carmen R., Pedersen F., Huffman L. & Bryan Y. (1993). Dyadic distress management predicts
security of attachment. Infant Behavior and Development, 16, 131-147.

Dennett D.C. (1978). Brainstorm. Cambridge MA: MIT Press.

Dennett D.C. (1987). The intentional stance. Cambridge, MA: MIT Press.

Egeland, B., & Farber, E. A. (1984). Infant-mother attachment: Factors related to its development and
change over time. Child Development, 55, 753-771.

Emde, R., & Buchsbaum, H. (1990). "Didn't you hear my mommy?" Autonomy with connectedness
in moral self-emergence. In D. Cicchetti & M. Beeghly (Eds.), The self in transition (pp. 35-60).
Chicago: University of Chicago Press.
Fonagy P. (1991). L'approccio evolutivo per la comprensione del transfert nel trattamento di pazienti
con distubi borderline della personalità. In: Zabonati A., Migone P. & Maschietto G. (a cura di), La
validazione scientifica delle psicoterapie psicoanalitiche. Mestre (VE): IPAR, 1993.

Fonagy P., & Target M. (1995). Understanding the violent patient: The use of the body and the role
of the father. International Journal of Psycho-Analysis, 76, 487-502.

Fonagy P., Leigh T., Steele M., Steele H., Kennedy R., Mattoon G., Target M. & Gerber A. (1996).
The relation of attachment status, psychiatric classification, and response to psychotherapy. Journal
of Consulting and Clinical Psychology, 64, 22-31.

Fonagy P., Moran G. S. & Target M. (1993). Aggression and the psychological self. International
Journal of Psycho-Analysis, 74, 471-485.

Fonagy P., Redfern S., & Charman T. (1997). The relationship between belief-desire reasoning and a
projective measure of attachment security (SAT). British Journal of Developmental Psychology, 15,
51-61.

Fonagy P., Redfern S. & Charman T. (in press). Individual differences in theory of mind acquisition:
The role of attachment security". British Journal of Developmental Psychology.

Fonagy P., Steele M., Moran G., Steele M. & Higgitt A.C. (1991). The capacity for understanding
mental states: The reflective self in parent and child and its significance for security of attachment.
Infant Mental Health Journal, 13, 200-216.

Fonagy P., Steele M., Steele H., Higgit, A. & Target M. (1994). Theory and practice of resilience.
Journal of Child Psychology and Psychiatry, 35, 231-257.

Fonagy P., Steele M., Steele H., Leigh T., Kennedy R., Mattoon G. & Target M. (1995a). The
predictive validity of Mary Main's Adult Attachment Interview: A psychoanalytic and developmental
perspective on the transgenerational transmission of attachment and borderline states, in Goldeberg
S., Muir R. & Kerr J. (editors), Attachment Theory: Social, Developmental and Clinical Perspectives,
Hillsdale, NJ: The Analytic Press, 1995, pp. 233-278.

Fonagy P. & Target M. (2001). Attaccamento e funzione riflessiva. Milano: Cortina.

Fox N. A., Kimmerly N. L. & Schafer W. D. (1991). Attachment to mother/attachment to father: A

meta-analysis. Child Development, 62, 210-225.

Gabarino J. (1992). The meaning of poverty in the world of children. American Behavioral Scientist,
35, 220-237.

George C., Kaplan N., & Main M. (1985). The Adult Attachment Interview. Unpublished manuscript,
Department of Psychology, University of California at Berkeley.

Goldberg W. A. & Easterbrooks M. A. (1984). The role of marital quality in toddler development.
Developmental Psychololgy, 20, 504-514.

Goosens F., & van Ijzendoorn, M. (1990). Quality of infants' attachment to professional caregivers.
Child Development, 61, 832-837.
Grossmann K. E., Grossmann K. & Schwan A. (1986). Capturing the wider view of attachment: A
reanalysis of Ainsworth’s Strange Situation. In C.E. Izard & P.B. Read (Eds.), Measuring Emotions
in Infants and Children, volume 2 (pp. 124-171). New York: Cambridge University Press.

Hamilton C. (1994). Continuity and discontinuity of attachment from infancy through adolescence.
Unpublished doctoral dissertation, UC-Los Angeles.

Hazan C. & Zeifman D. (1994). Sex and the psychological tether. In K. Bartholomew & D. Perlman
(Eds.), Advances in Personal Relationships, volume 5 (pp. 151-178). London: Jessica Kingsley
Publishers Limited.

Hegel G. (1807). The Phenomenology of Spirit, translated by A.V. Miller. Oxford: Oxford University
Press, 1977 (trad. it.: La fenomenologia dello spirito. Firenze: La Nuova Italia, 1949).

Isabella R.A. (1993). Origins of attachment: Maternal interactive behaviour across the first year.
Child Development, 64, 605-621.

Isabella R.A. & Belsky J. (1991). Interactional synchrony and the origins of mother-infant
attachment: A replication study. Child Development, 62, 373-384.

Isabella R.A., Belsky J. & von Eye A. (1989). Origins of infant-mother attachment: An examination
of interactional synchrony during the infant's first year. Developmental Psychology, 25, 12-21.

Jacobson S. W., & Frye, K. F. (1991). Effect of maternal social support on attachment: Experimental
evidence. Child Development, 62, 572-582.

Kagan J. (1982). Psychological Research on the Human Infant: An Evaluative Summary. New York:
Wiley.

Lamb M. E., Thompson R. A., Gardner W. P., Charnov E. L.,& Estes D. (1984). Security of infantile
attachment as assessed in the "strange situation": Its study and biological interpretation. The
Behavioral and Brain Sciences, 7, 127-147.

Levinson A. & Fonagy P. (submitted). Attachment classification in prisoners and psychiatric patients.

Lewis M. & Feiring C. (1989). Infant, mother, and mother-infant interaction behavior and subsequent
attachment. Child Development, 60, 831-837.

Lieberman A. F. & Pawl J. H. (1990). Disorders of attachment and secure base behavior in the second
year of life: Conceptual issues and clinical intervention. In M. T. Greenberg, D. Cicchetti & E. M.
Cummings (Eds.), Attachment in the Preschool Years (pp. 375-398). Chicago: University of Chicago
Press.

Lieberman A. F., Weston D.R. & Pawl J.H. (1991). Preventive intervention and outcome with
anxiously attached dyads. Child Development, 62, 199-209.

Lyons-Ruth K., Connell D. B. & Grunebaum H. U. (1990). Infants at social risk: Maternal depression
and family support services as mediators of infant development and security of attachment. Child
Development, 61, 85-98.
Main M. & Goldwyn R. (1994). Adult Attachment Rating and Classification System, Manual in
Draft: Version 6.0. Unpublished manuscript, University of California at Berkeley.

Main M. & Hesse E. (1990). Parents' unresolved traumatic experiences are related to infant
disorganized attachment status: Is frightened and/or frightening parental behavior the linking
mechanism? In D.C.M. Greenberg & E.M. Cummings (Eds), Attachment in the preschool years:
Theory, research and intervention (pp. 161-182). Chicago: University of Chicago Press.

Main M. & Solomon E. (1990). Procedures for identifying infants as disorganized/ disoriented during
the Ainsworth Strange Situation. In D.C.M. Greenberg & E.M. Cummings (Eds), Attachment in the
preschool years: Theory, research and intervention (pp. 121-160). Chicago: University of Chicago
Press.

Main M. (1991). Metacognitive knowledge, metacognive monitoring, and singular (coherent) vs.
multiple (incoherent) models of attachment: Findings and directions for future research, in: P. Harris,
J. Stevenson-Hinde & C. Parkes, editors, Attachment Across the Life Cycle. New York: Routledge,
1991, pp. 127-159.

Main M., & Weston D. (1981). The quality of the toddler's relationship to mother and to father:
Related to conflict behavior and the readiness to establish new relationships. Child Development, 52,
932-940.

Malatesta C. Z., Grigoryev P., Lamb C., Albin M. & Culver C. (1986). Emotional socialisation and
expressive development in pre-term and full-term infants. Child Development, 57, 316-330.

Maslin C.A. and Bates J.E. (1983). Precursors of anxious and secure attachments: A multivariant
model at age 6 months. Paper presented at the biennial meeting of the Society for Research in Child
Development: Detroit, MI.

O'Connor M., Sigman, M. & Kasari, C. (1992). Attachment behavior of infants exposed prenatally to
alcohol: Mediating effects of infant affect and mother-infant interaction. Development and
Psychopathology, 4, 243-256.

Oliver J.E. (1993). Intergenerational transmission of child abuse: Rates, research, and clinical
implications. American Journal of Psychiatry, 150, 1315-1324.

Radojevic M. (1992). Predicting quality of infant attachment to father at 15 months from pre-natal
paternal representations of attachment: an Australian contribution. Paper presented at the 25th
International Congress of Psychology: Brussels, Belgium.

Sroufe L.A. (1979). Socioemotional development. In J. Osofsky (Ed.), Handbook of Infant

Development (pp. 462-516). New York: Wiley.

Sroufe L.A. (1990). An organizational perspective on the self. In D. Cicchetti & M. Beeghly (Eds),
The self in transition: Infancy to childhood (pp. 281-307). Chicago: University of Chicago Press.

Sroufe L.A. (1996). Emotional development: The organization of emotional life in the early years.
New York: Cambridge University Press.
Steele H., Steele M. & Fonagy P. (1996). Associations among attachment classifications of mothers,
fathers, and their infants: Evidence for a relationship-specific perspective. Child Development, 67,
541-555.

Steele H., Steele M. & Fonagy P. (in preparation). A path-analytic model of determinants of infant-
parent attachment: limited rather than multiple pathways.

Stern D. (1985). The Interpersonal World of the Infant. New York: Basic Books (trad. it.: Il mondo
interpersonale del bambino. Torino: Bollati Boringhieri, 1987).

Stern D. (1994). One way to build a clinically relevant baby. Infant Mental Health Journal, 15, 36-
54.

Taylor P.J. (1986). Psychiatric disorder in London's life sentenced prisoners. British Journal of
Criminology, 26, 63-78.

van den Boom, D.C (1990). Preventive intervention and the quality of mother-infant interaction and
infant exploration in irritable infants. In W.Koops & et al. (editors), Developmental Psychology
Behind the Dikes (pp. 249-270). Amsterdam: Eburon.

van den Boom D.C. (1995). Do first-year intervention effects endure? Follow-up during toddlerhood
of a sample of Dutch irritable infants. Child Development, 66, 1798-1816.

van IJzendoorn M.H. (1992). Intergenerational transmission of parenting: A review of studies in non-
clinical populations. Developmental Review, 12, 76-99.

van IJzendoorn M.H. (1995). Adult attachment representations, parental responsiveness, and infant
attachment: A meta-analysis on the predictive validity of the Adult Attachment Interview.
Psychological Bulletin, 117, 387-403.

van Ijzendoorn M. & Kroonenberg P. M. (1988). Cross-cultural patterns of attachment. A meta-

analysis of the Strange Situation. Child Development, 59, 147-156.

Ward M. J. & Carlson E.A. (1995). Associations among Adult Attachment representations, maternal
sensitivity, and infant-mother attachment in a sample of adolescent mothers. Child Development, 66,
69-79.

Waters E., Merrick S., Albersheim L., Treboux D. & Crowell, J. (1995). From the Strange Situation
to the Adult Attachment Interview: A 20-year Longitudinal Study of Attachment Security in Infancy
and Early Adulthood. Presented at meeting of the Society for Research in Child Development:
Indianapolis, IN.

Winnicott D. W. (1967). Mirror-role of the mother and family in child development. In P. Lomas
(editor), The Predicament of the Family: A Psycho-Analytical Symposium. London: Hogarth, pp. 26-
33.

Zeanah C. H., Benoit D., Barton M., Regan C., Hirshberg L.M. & Lipsitt L.P. (1993).
Representations of attachment in mothers and their one-year-old infants. Journal of the American
Academy of Child and Adolescent Psychiatry, 32, 278-286.