Can Foster Carers Help Children Resolve Their Emotional and Behavioural Difficulties?

Although foster care is generally seen as providing a positive experience for the children and young people for whom it caters, it is rarely conceived of as a place where the children are helped to address their emotional difficulties and modify their often difficult behaviour. Yet research evidence suggests that some foster carers are consistently less likely to have placements which break down, and that foster carers who show particular skills in parenting can make a difference to successful outcomes. The article draws on a large longitudinal study of foster care to argue that it is possible to learn from what these foster carers do in order to develop these skills in others. A model of successful foster care. developed from the main statistical part of the study is first described. Two cases from the qualitative, case-studies component of the research are then analysed to demonstrate a quality of responsive parenting. The model is further developed within the framework of the dynamic of attachment and interest sharing proposed by Heard and Lake (1997), to show how this can be used as a basis for future approaches to working with foster placements.

T H I S A RT I C L E D R AW S on a large longitudinal study 1 to consider what we know about the strengths and limitations of foster care, and how this knowledge can be used to ensure better outcomes for the children and young people involved by bringing about changes in their often difficult and disturbed behaviour.
To anticipate the argument, one of the key findings of the study, undertaken by the present author with colleagues (Sinclair, Baker, Wilson, & Gibbs, 2005;Sinclair, Gibbs, & Wilson, 2004; is that although foster care is on the whole viewed as a benign activity by its key players, it is rarely seen as a place where change takes place. In keeping perhaps with this expectation, the study did not find major changes in the personalities and social performance of children over the year in which they were followed up. Yet our study also found that skilled foster care could make a difference. Some carers were more likely to have placements which were considered successful and did not break down. Foster carers who showed particular qualities of parenting and child orientation experienced fewer placement disruptions and contributed to successful outcomes in ways which led to some, albeit often small, improvements in the children's happiness, behaviour and relationships. 2 If this was true of some, is there a way of understanding and learning from what these foster carers do so that it can be true of more and so that more substantial changes could be achieved? We shall draw on the findings of the study, with an additional analysis within the framework of the dynamic of attachment and interest sharing, to suggest that this is indeed the case.
The evidence comes from a study of support for foster carers. This was carried out in seven local authorities: Two London boroughs, a metropolitan district council, two shire counties and two unitary authorities. In total 472 foster children were followed up at an interval of 14 months. The children's social workers, the family-placement social workers and the foster carers completed questionnaires and reported on disruptions, and the reasons for placement success or lack of it. These views combined with comments from the children led to hypotheses about the origins of successful placements. Detailed case studies were also undertaken involving 24 placements, 12 selected because they were judged to be successful, 12 as less successful, which were intended to check and also deepen the conclusions from our main surveys.
The findings of the study are in many ways very encouraging. The great majority of the foster children who answered our questionnaires said they were happy with their placements. Our case studies illustrated excellent practice. Our questionnaires gave ample evidence of the commitment of carers. Social workers judged that around 7 out of 10 placements were going or had gone very well, a proportion similar to those found by Rowe, Hundleby, and Garnett (1989) and Cliffe and Berridge (1991).
Despite these encouraging results the study identified a number of key problems. The most fundamental of these was the contrast between children's wish for a stable family life and the limited time most of them in fact spent in placements. Foster care did not seem to be conceived as treatment. (The social workers and foster carers in the study rarely saw the purpose of the placement as being for 'treatment' or to bring about change in the foster child.) It was infrequently seen as a preparation for return home, with under 1 in 10 of the placements intended to get the child and her or his parents back together again. It very rarely offered a long-term 'family for life', with fewer than 1 in 8 remaining in the same placement for more than 4 years. Essentially it provided in a phrase used by Thoburn et al. (2000) 'a port in a storm'.
These problems are perhaps a reflection of the prevailing fundamental approach to foster care. For children to go into 'care' is seen by many as a confession of failure. It is preferable, as guidance and legislation suggest, for children to remain with their parents. It is acceptable if they can be adopted. Parenting by the state is seen as expensive and ineffective -something which should only be undertaken as a last resort and which should be as temporary as possible. Given such beliefs it is not surprising that so many of the foster children in our sample were repeatedly tried at home, that very long stays in placements were uncommon and that so few of them stayed on beyond the age of 18 (Sinclair, Baker, Wilson, & Gibbs, 2005; This set of expectations does not easily provide a coherent framework for foster care. The old approach to foster care as a kind of quasi-adoption, where foster carers could bring children up as their own with limited or no contact with the children's birth families, is no longer accepted (Wilson & Petrie, 1998). As one foster carer in our pilot study wistfully commented, 'they expect you to bring the child up as your own, but not to get involved'. Yet if foster care is not a kind of up-bringing what is it? In contrast to what was formerly true of at least some forms of residential care it does not claim to provide 'therapy'. It is rarely (except in specialist 'treatment foster care') seen as involving treatment designed to change the child. Treatment, if provided at all, is given in the context of foster care, that is, by other professionals such as psychotherapists or psychologists giving help, rather than being an intrinsic part of it.

The need for a theory of foster care as an environment for change
Children who enter foster care and remain there for any length of time have generally experienced abuse, severe family conflict or both. This leaves many of them with difficulties in handling relationships. This in turn probably contributes to their difficulties at school. Difficult behaviour learnt at school or in other social settings can in turn exacerbate their problems in relationships with others. So although the foster children in our sample showed many admirable traits, such as kindness to others and resilience, many were also seen as difficult and to have a range of problems to do with relationship difficulties, disturbed behaviour within the home and antisocial behaviour outside the home. Our study found that children may change at the beginning of a placement and that there was probably improvement in quite a range of behaviour when the children were first placed. Thereafter we found little shift in either the absolute level of children's disturbance or social performance or their position in these respects relative to others. Could foster care do more to enable change to occur? If it could, then this in turn might ensure more stability and permanence for children in preventing the kinds of difficult behaviour which could lead to placement breakdown, and, in those situations where return home was possible, might do more to ensure this was successfully achieved.
We attempted tentatively to find answers to these problems by analysing factors in our sample which lead to a 'successful placement' and the avoidance of disruption. Our aim was to build a 'model of how placements work'. This means no more than that we try to identify key variables in the placement and show how, in combination, they lead to success or the reverse. We based this model on the views of foster carers, foster children, and social workers. We then tested its elements through statistical comparisons, showing that each of the key variables is associated with outcome. Finally, we put the variables together in a 'multivariate model'. We used a number of case studies to develop the model in a way which the more static statistical accounts cannot do. We elaborated the model through the concept of 'responsive parenting', taking the analysis further in a way which is detailed and precise, and, we suggest, has relevance to understanding the processes which govern changes over time and can be built on to bring about improvements in practice.
In the original model we drew on attachment concepts as well as those derived from learning theory, to suggest that carer responses of warmth and refusal to respond to bad behaviour with rejection is in keeping with attachment theory on the one hand, and that on the other, firmness and reinforcement of positive behaviour fit with ideas from learning theory. However, I shall argue here that Heard and Lake's (1997; see also Heard, 2003) elaboration, termed the attachment dynamic, of Bowlby's (1982) attachment theory to include a parenting component in care-giving behaviour, and three further goal-directed systems, provides a more complete explanatory framework for understanding the emotional needs of children with attachment difficulties and their relationships with those who care for and work with them.

A model of effective foster care
In developing the model, 3 success was judged by whether or not a placement had broken down, and whether it was seen as successful by the professionals concerned (i.e., foster carers, social workers and family-placement social workers). These views combined with comments from the children led to hypotheses about the origins of successful placements. Despite some differences in emphasis the three groups consistently identified three aspects of the placement as being important in foster care and contributing to success. The first of these was the children's characteristics. Children who wanted to be fostered, had attractive characteristics and low levels of disturbance did better. Second were the qualities of the foster carer. Placements with warm, child-oriented carers were more successful. The third element in our model related to matching and interaction. Social workers in particular were clear that similar children might elicit different reactions from different carers. The element included the degree to which the children matched the family and carer characteristics (e.g., the child was not likely to bully children already in the house and was of an age which the carer liked to take) and the 'chemistry' between child and carer, an unpredictable factor that led both sides to know that this placement would or would not work: 'Placements that do not have the matching potential criteria can sometimes have the best outcome because of attachments, commitments and personality. In short "just clicking"' (Sinclair & Wilson, 2003, p. 5). Malign or positive spirals of interaction could occur so that rejection led to rejection in return or love elicited love. These spirals were not, however, inevitable so that, for example, overtures for love could be met with an embarrassed response or difficult behaviour on the part of the child might not be interpreted as rejection but rather as a signal of distress: 'These carers have bags of patience and tenacity. They were able to interpret the child's behaviour as communication' (p. 5).
By putting these accounts together we developed our model. This was that outcome depends on: ■ the children (their motivation, attractiveness and difficulty); ■ the carers (their 'warmth', persistence and ability to 'set limits'); ■ the interaction between the two. We tested this model by a range of statistical measures.. These included measures of child's characteristics (e.g., motivation to be in placement, attractiveness/prosocial scores from Goodman's Strengths and Difficulties Questionnaire (Goodman & Scott, 1977)) and three foster-carer ratings, a carer-parenting score based on the views of social workers, a rejection score based on carer responses relating to their fondness/acceptance of their foster child and a child-orientation score, based on an instrument developed by Marjorie Smith at the Thomas Coram Research Unit. This essentially counts the number of different things that a parent might do with a child that the child might be expected to like (for example, reading a bedtime story, going to a football match). (These are discussed in detail in Sinclair & Wilson, 2003;Sinclair, Wilson, & Gibbs, 2005, pp. 188-192.) All our hypotheses were confirmed. First, in relation to measures relating to the child's characteristics, that is, motivation, prosocial scores and low measures of disturbance, placements were more likely to be successful and less likely to disrupt when the children were seen as wanting to stay in the placement and had high prosocial scores and low measures of disturbance. Table 1 relates our measures (grouped for the purposes of presentation) to our success measure and Table 2 to disruption.
Second, we found that our three statistical measures, the child-orientation score and the two parenting scores tap dimensions of carer behaviour which distinguish one carer from another and provide evidence that carers who are good with one child tend to be good with another. Tables 3, 4 and 5 show the associations of our parenting measures with one of our outcome measures, disruption. Our results suggest that carers who score well on these dimensions are more likely to have had successful placements. Some carers are consistently better than others (that is, warmer, clearer about what they expect, more empathic with the child, less easily put out by the child's behaviour and more concerned to do things with the child).
It remains possible of course that difficult children produce carers who parent less skilfully. Two findings, however, suggest that there is some influence from parent to child.  Notes. The percentages add up across the rows. Tables reproduced with kind permission from the publishers and from .
First, our measure of foster-carer orientation was not associated with child difficulty. So the child's difficulty does not appear to explain the negative association between child orientation and disruption. Second, our parenting scores were 'better' among carers whose previous foster placements were less likely to break down. There was no similar association between previous breakdowns and the difficulty of the current child so it is unlikely that 'good parenting' is simply a reflection of a 'good' child. In general, these findings suggest the crucial relevance of the way carers 'parent' to outcomes. In this respect it is highly consistent with the work of Farmer, Moyers, and Lipscombe (2004), Quinton, Rushton, Dance, and Mayes (1998), and Schofield, Beek, Sargent, and Thoburn (2000). They also support the need to pay close attention to children's views, and suggest that success depended in part on the match, 'chemistry' and developing interaction between parent and child. Other findings also highlight the potential importance of early intervention to prevent negative spirals of interaction between carer and child. In the context of the theme of this article, it is also worth noting that, with one exception, 4 we did not find any association between outcomes and different forms of special help (e.g., psychotherapy, play therapy, life-story work) provided to the children by other professionals.

The model elaborated
In order to deepen our understanding of the model developed from the questionnaires, we selected for face-to-face interviews with the foster carers and telephone interviews with their social workers 24 cases judged on the basis of carers' responses to the second questionnaire to be 'successful' or 'less successful' according to the study criteria. These were matched as far as possible on age, gender and length of time in placement although in the event 13 successful and 11 less successful placements were sampled. I now turn to discuss two of these cases, to show how we developed our more detailed model of successful and unsuccessful foster care in a way which is more precise and complete than is possible from our main surveys. I go on to extend the analysis using the framework of the attachment dynamic and interest sharing developed by Heard and Lake (1997). In outlining the model we distinguish between two kinds of variable: 1. Responsive parenting: The way in which the carer deals with the child; 2. Conditions: The prior conditions which make this kind of interaction more or less likely. In the model, we distinguish two types of conditions: (a) Central conditions to do with the characteristics of the child, the foster carer (for example, general skill, absence of strain) and the compatibility between them, and (b) those to do with the wider context of the placement.
Responsive parenting is a necessary condition of a successful placement. The various conditions are not necessary or sufficient for success but they do make it more, or less, likely. These conditions relate to the characteristics of the foster child (for example, attractiveness, difficult behaviour and motivation to be in the placement), particular attributes of the foster carer, for example her/his general skill and the presence/absence of a sense of strain, and the compatibility between the carer and the child in question. Other conditions reflect the wider context, including the care plans for the child, relationships with social worker, school and other professionals and the involvement of the birth family members and members of the foster carer's own family.

James 5
Our first case was seen by all those involved as successful. James was 9 years old at the time of interview and had a learning disability. He came into care when he was 5 years old, following his brother's disclosure of physical abuse and later sexual abuse at the hands of his father, and was placed with his paternal grandmother. From his birth-mother's subsequent encounters and behaviour towards James's foster carer during contact, we may hypothesize that her caretaking pattern was highly ambivalent and defensive, with a lack of concern for his needs when she felt herself to be threatened emotionally, for example when as happened James's behaviour appeared to threaten her relationship with a new partner. James's grandmother cared for him until she became unable to cope with his increasingly unruly, violent behaviour, which included physical attacks, and setting fire to his bedroom, and other small fires in the kitchen and living room. He was placed with his present foster family initially for an 8-week assessment, and had been there at the time of the interview for 2 1 ⁄ 2 years.
His foster family consists of a married couple, and five birth and foster children, all older than James and including a 16-year-old foster child whom they are in the process of adopting. Mrs Stanton is an experienced foster carer, who worked in residential child care before fostering, and now chiefly undertakes time-limited assessments of 'looked-after' children for the local authority. She comes across as a warm, articulate individual, very confident about what she and her family can offer the children they look after. She is clear about the support which she needs in order to work effectively with James. She and her husband have distinctive roles in caring for James, with her husband dealing with everyday issues and being there as a safe person and 'for cuddles and to play with or get computer games', while 'all the other work . . . you know, tackling his behaviour, the soiling, everything else, I dealt with'. She says that she 'loves James to bits', that he is part of the family, and sees the family as providing a permanent home for him. She describes an immediate mutual attraction between her and James. During the first 8 weeks, James's behaviour was extremely difficult: He was verbally abusive to the children in the household, he used to hurt them. He was very frightened, unsure what was happening to him. He soiled and smeared faeces . . . He weed everywhere. He ripped furniture and bedding . . . we took him away to our caravan and he set it on fire. He threw food, he wouldn't eat it. He got a knife to me at one point.
After weathering these problems during the first weeks with James, and seeing him beginning to settle and become trustful, neither she nor her family felt they could let him go to another family, and the social services department agreed, with some evident relief, that he should remain with them on a permanent basis. At follow-up over a year later, James was still living successfully with the Stantons.
In our original analysis of the foster placement, we explored the case first by describing the key conditions, which we saw as encompassing both central conditions to do with the characteristics of the child (e.g., attractiveness, difficult behaviour and motivation to be in the placement) and the foster carer, and the conditions to do with the wider context, including support from the foster family, school and other professionals.

The core conditions
In relation to the foster carer, we describe her as a skilful and experienced foster carer, whose qualities include general skill and motivation to foster, a realistic and clear-sighted commitment to the children she fosters, a sense of professionalism and a relative absence of strain. Her response to James is to see him as a challenge, which she would enjoy rather than feel threatened by. Although entirely committed to James, she is also realistic in her expectations of him, acknowledging that with James it is never going to be easy. Her behaviour is also characterized by an absence of strain, and the capacity to judge how much she can take on by herself, and what she needs in addition -for example, time off, or advice and support from a psychologist -in order to enable her to do her job properly. She has the ability to identify the help that she considered she needed to work with James, and assessed and used this appropriately.
In relation to another of the core conditions, the child, we show that, although it would have been difficult to describe James in general as an obviously attractive child, there seems to have been enough about him from the beginning to make him a rewarding one: Slowly he was giving more and more and he was starting to trust. And he started calling me mum. And I just couldn't let him go. James would always give you something back which made you feel good about the relationship you'd got . . . he'd come each day and stroke John [her husband]'s leg, and say, 'Are you all right, John?'

Responsive parenting
In our original analysis, we showed how the quality of responsive parenting was shown to a high degree in the way Mrs Stanton cares for James. This includes handling attachment sensitively, reinforcing socially acceptable self-esteem, and handling difficult CLINICAL CHILD PSYCHOLOGY AND PSYCHIATRY 11(4) behaviour appropriately. In handling attachment for example, she speaks of him as a loved child: 'We couldn't let him go', 'James would always give you something back which made you feel good about yourself'.
She uses attachment-sensitive times, such as moments when he is settling down in the evening, or after an upset, and notably, gets across to him a feeling of being loved and secure by handling situations in ways where she might be expected not to convey love and acceptance (e.g., when being physically attacked by him).
She provides a secure base, by demonstrating sensitivity, availability and consistency. She also shows an ability to promote a secure attachment by demonstrating understanding and accurate empathy. In reflecting in the interview, for example, on what prompts his outbursts, she suggests that the behaviour may come from a wish to be naughty, from a wish to hit out in anger, out of fear of having someone trying to control him (by implication because of his earlier experiences of punitive parental care), and from a wish to avoid the experience of failure by not trying to succeed in the first place: 'Because he wanted to be able to have, say, a dry bed. But he knew he'd have problems achieving that, so he'd rather kick off and not try than be a failure'.
Another characteristic of responsive parenting is the ability to handle difficult behaviour appropriately -providing predictable limits and thus lessening his anxieties and leaving him more available for personal engagement: I never altered the way I handled him, so that he got a feeling of knowing what I was going to do. For him to trust me, he had to learn to understand me and how I work. I kept my voice on an even keel all the way through dealing with him. It worked because it gave him a feeling of, well, not being in control of me as such, but of knowing that I wasn't going to hurt him.
A further characteristic, as we suggest in our original analysis, is that she develops his sense of efficacy, identity and self-esteem, by encouraging his intellectual development and achievements: 'And he's beginning to enjoy who he is, what he's done, little things, reading simple books and suchlike. And every now and again, we have a sit-down and a talk about all the positives that's happened to James'.

The dynamics of attachment and interest sharing
I want now to suggest that employing the formulation of the dynamics of attachment and interest sharing enables us to build a more complete model than the one we have so far developed. In doing this I draw on Heard and Lake's (1997) formulation which includes: An expanded system of parenting; the identification of an exploratory interestsharing system with peers; and a self-defence system, which once activated may interrupt the operation of other goal-directed systems, such as caregiving.
Attachment theory and research have expanded rapidly from Bowlby and Ainsworth's original work to encompass adult-attachment styles, the more complex organization of attachment responses in older children, intergenerational transmission of attachment styles, and new approaches to working with attachment difficulties. However, the theory has also been criticized for its predominant focus on care-seeking behaviour. It fails to account for other distinct interpersonal forms of relating, supportive and companionable interaction in particular. And it is incomplete, in that it does not integrate other intrapersonal systems such as creativity (highlighted by Winnicott (1965Winnicott ( , 1971; also cited by Heard, 2003) in his seminal work) or, perhaps most importantly, psychosexuality.
Heard and Lake (1997) address these gaps and imbalances by adding three other goaldirected systems to Bowlby's original model. All these systems are instinctive, goal corrected (i.e., needing a match to be reached between an inbuilt representation of what is required and what is obtained) and intrinsically motivated. Two of the three have interpersonal goals:

■ The system for parenting: This includes Bowlby's original concept of care giving, but
Heard and Lake consider that this has to be seen as representing one of two components of a parental system -the first, which is as defined by Bowlby, provides protection, comfort and satisfaction of physical needs; the second promotes and sustains the development of the psychological capacities to be autonomous, exploratory and able to share mutual interests with peers. The care-giving component of the parenting system is active in a fragmentary way from early childhood. The development (or 'educative') component does not mature fully until early adult life. The individual whose parental system is fully functional moves instinctively from one component of parenting to the other, and gives empathic responses to an infant, child or adult who is either seeking care in circumstances defined by Bowlby, or seeking help to cope with some situation developed through their own efforts. ■ The care-seeking system: The system which Bowlby discussed in detail. Heard also points out, following Trewarthen's studies of intersubjectivity (cited in Heard, 2003) that no one is able to seek both components of parenting simultaneously. Seekers oscillate from seeking care giving to seeking help for their own interests after careseeking needs are satisfied. ■ The exploratory system, extended from Bowlby's original formulation to include exploratory interest-sharing and companionable interaction with peers during development and adulthood. This is the social component of the individual exploratory system and affected in the same way (i.e., liable to be overridden by the arousal of care seeking). ■ The system for affectional sexuality: Heard and Lake (1997) consider sexuality to be in fact two systems -one offshoot of interest sharing with peers, the other as one expression of the system for personal defence. ■ Personal defence system: Activated either when fear of abandonment, shaming and or dismissive and angry care are experienced or when care giving is perceived as insufficiently nurturing and protective. The system is active from birth. It includes all the instinctive knee-jerk reactions to danger invoked by fear. It also activates attachment behaviour.
The attachment dynamic therefore provides a more complete theoretical integration of systems for caring for others, of exploration and interest sharing with peers, and of selfcare within intimate relationships which is helpful in analysing the interpersonal dynamics in complex child-care cases. Applying the dynamics of attachment and interest sharing to Mrs Stanton's responses, we seem here to witness the smooth, interrelated functioning of all the five systems posited by Heard and Lake in adults.
To take first the parenting system, Heard and Lake's (1997) development of the additional component within the parenting system allows us to encompass both those care-giving responses which handle attachment, and also those responses based on social learning theory which are directed towards handling difficult behaviour, and which can be seen as making up the growth and development component.
In the first, care-giving component, we see how James is enabled to experience secure attachment through Mrs Stanton's care-giving responses: Consistency, availability, sensitivity and empathy. The second, 'educative' development component of the foster carer's parenting system is also very evident: She encourages his small achievements, as in the example of his beginning to read books, earlier, or identifies and develops clear behavioural strategies for dealing with his difficult behaviour, as in the following example: So he took that on board, and I put it to the test a couple of days later, because I wasn't here when he came home from school. What I tend to do is if I go through a situation like that with James [she had been out when he returned from a trip and he had thrown a tantrum] I test it out fairly soon afterwards before he's forgotten about it, so I can acknowledge to him how proud I am he's coped and also work with him more if I need to, around that issue, if it's failed.
And at another point in the interview she was asked about her way of handling difficult behaviour, in this case, getting him to clear up a mess that he had deliberately made: I'd sit in a corner, hold my back to the wall, and I'd do the crossword. I'd be there with him . . . keep on reminding him that I was there, and that when he was ready, I was ready. And then he'd sort of realized that nothing was going to alter until he'd done whatever I'd requested him to do.
She also helps him to understand his feelings and develop strategies for dealing with times when he is becoming upset or angry: 'And he can sort of say, "well, I'm going up to me room, Mum, because I'm jumpy inside"'.
In the above example, her handling of the tantrum may both serve to help James learn not to repeat this behaviour, and also at the same time meet the needs of his attachment system for a secure base. It is clear that the same behaviour on the part of the carer may serve multiple ends. For example, by speaking of the child as a 'loved child' a carer may simultaneously reduce the child's fear of rejection, increase her/his self-esteem and reduce the anger and anxiety which may lie behind some of the difficult behaviour. Heard and Lake's (1997) division of the parenting system into two components does, I am suggesting, provide a framework which enables us to explain and describe this more accurately.
The other components of the attachment dynamic to be illustrated from Mrs Stanton are those of the exploratory interest sharing system with peers and the care-seeking system. Mrs Stanton is ready to seek outside help and support, recognizing that 'with James it's never going to be easy'. She acknowledges her need for professional advice, and seems willing to accept it: I know that there's no way I would cope without knowing that I've got a back-up system. I have an awful lot of experience, through friends, through children placed with me to know that the children I foster have immense problems. And I can't tackle these problems on my own. They need support, and I need support.
She uses respite and a range of other supports and activities in order to address her own needs for rest and recovery. From her account, too of the way in which she and her husband, John, divide the fostering roles, we can surmise that the couple's relationship is supportive and, from this snapshot picture at least, companionable: All the other work that was, you know, tackling his behaviour, the soiling, I dealt with. And we felt that it was important that James had a safe person . . . and so John has . . . to encourage him to turn to John, and get a positive feeling, you know, of a father figure. And as I say, we maintained that, and John would be the one that he would go to afterwards, and he would be there for him.
Overall the interview suggests that the systems for adult attachment and interest sharing are running smoothly.
Finally, Mrs Stanton does not see James's care-seeking behaviour, or his difficult behaviour when his self-defence system is highly activated, as a threat, so that in turn her own self-defence system is not activated and her parenting system lowered. Thus her first reaction to his somewhat defiant behaviour at their first meeting is to see him as a challenge, which she would enjoy -'Ooh, I've got my hands full here, and I love a challenge' -rather than to respond defensively. She is alert to the need to address at once James's mildly oppositional response -'So I said, look at me James, you know that's not right' -in other words, her parenting system remains active.
In our original account of the following example, we suggested that her secure adult response demonstrated her reflective capacity, recognizing the behaviour as in part a product of their relationship, but not personalizing the behaviour as an attack on her: 'He's not actually hitting out at me. He's hitting out at his anger. And if he thinks he's hurt me, he's devastated. And that's the big contrast to where he first came, where he'd actually charge at me and go out to hurt me'.
A more complete explanation is to see this process as having the potential for the lowering of Mrs Stanton's parenting system in the face of what someone else very understandably might see as a threat. As we suggested earlier, it would be easy to react to James's highly defensive, out-of-control behaviour, with a self-defence system which in turn becomes activated (so that the individual experiences high levels of anxiety, anger, frustration at being rejected, threatened and so on). In fact, as the above quotation arguably demonstrates, her nonconfrontational and nondefensive response enabled James to experience secure caregiving. This in turn, although she did not in fact in the interview go on to describe it, would allow James's self-defence system to become deactivated, and his other exploratory and care-seeking systems to revert to their original prethreat levels.

Giles
I want now further to illustrate the attachment dynamic by analysing the case of Giles which we matched as a 'less successful' case with James, in terms of age, sex and length of time in placement, in developing our original model. For reasons of confidentiality, since we were analysing the case as an unsuccessful placement, we did not reproduce the narrative in the detail which we did for James. Giles at the time of interview was a boy of 10 years with moderate learning difficulties who had experienced severe sexual and physical abuse. He was placed initially as a short-term placement with relatively inexperienced foster carers, where he has remained, despite serious behavioural difficulties, for 3 years. Giles has a violent temper and has physically attacked his foster mother and his foster brother. He is destructive, kicking through the partition between his bedroom and that of the children of the foster family and nearly destroying the interior of the caravan on a family holiday. He masturbates at times obsessively, urinates and defecates in his bedroom and in drawers, and exposes himself at the bedroom window. This behaviour shows no sign of diminishing. His foster mother seemed overwhelmed by and somewhat fearful of the difficulties he poses, and to feel that if he continues as he is, he will become too strong for her to handle. Although there are signs of some attachment to his foster mother he remains troubled and difficult to handle.
Mrs Lowe had had limited experience as a foster carer before having Giles, and this inexperience, and a period of ill-health, had put her under considerable strain. She presented in interview as somewhat isolated, and seemed to have a marital relationship in which the fostering task was tacitly understood to be hers rather than a shared one. She is critical of the social work support which she receives, suspecting that she was not given the complete story of Giles's background when he was placed with her. She does not consider the relationship with social services to be a good one, and has difficulty in carrying out suggestions for handling Giles.
Analysing this within the framework of attachment dynamic and interest sharing, in contrast to Mrs Stanton it does not seem that the goals of the carer's care-seeking and interest-sharing systems are being met harmoniously. The carer's professional relationships do not provide enough support for them; instead Mrs Lowe, perhaps similarly to Giles, seems to feel 'abandoned' to deal with Giles on her own. She does not appear to trust the social worker: 'They never tell you the full story, do they? He [the social worker] seemed a bit cagey'. She does not consider the relationship between herself and the social workers to be a good one, and finds the help offered inadequate or worse: 'They always seem to be in a rush. And they're never there when you need them'. She has had some intermittent support from a psychologist, but this is 'not much help. He doesn't really understand how difficult Giles is and what I have to put up with'.
It is likely that this lack of satisfactory professional supervision or help, that is, the failure to respond to Mrs Lowe's care-seeking behaviour, served to accentuate her feelings of inadequacy and hence her own self-defensive responses towards Giles. Equally, although of course we can only hypothesize, it is possible that the professionals themselves were experiencing some contagion of emotional responses, in trying to work with a clearly complex and emotionally troubled foster carer-child relationship. Heard and Lake (1999) argue that insecure care givers with high levels of self-defence and low exploration and interest sharing with peers, are prone to care seeking themselves. If insecure care givers see their children's care-seeking behaviours as a threat, their care giving lowers as their self-defence system becomes more highly activated. Feelings of incompetence, anxiety and failure too are likely to activate the self-defence system and lead to the inability satisfactorily to maintain care-giving behaviours. In the case of this foster carer, she appeared frequently to have her self-defence systems rather than her caregiving systems aroused when Giles's care-seeking needs were highly activated. In the following account, for example, she finds it difficult to empathize with him when she is feeling upset, and seems unable cognitively or emotionally to allow his feelings or needs to intrude upon her own needs for comfort: 'And he's coming for a cuddle within 10 minutes of exploding. And you're still sort of . . . you know, angry and wanting a break yourself, and sort of thinking to yourself, what on earth does he think he's doing? So it's very difficult to deal with him afterwards'. Or in the following, the tone of angry frustration and the absence of either a care-giving or 'educative' response suggests that her parenting system is unavailable: 'He doesn't often say sorry for anything . . . not for the right things. If he wants something. He broke his roller blades at the weekend. We went through a lot of sorry then'.
She responds in a negative way to behaviour which, although understandably irritating, seems to reflect emotional neediness, requiring some empathy and discussion with him. For example, she sees it as 'disgusting' when in the middle of the night he has raided the fridge and consumed the children's weekly allowance of Mars Bars. Instead of trying to explore with him the reasons for his difficult behaviour, or of helping him find alternative strategies for outbursts of temper, or setting clear and predictable limits, the strategies adopted seem to be largely mechanical ones, for example of putting locks on the doors and alarming his bedroom, or punishment, such as being sent to his room or into the garden. She concludes: 'We've had to change so much to keep him here . . . and if he doesn't learn to calm his temper down, then it will eventually break down, because I won't be able to handle him here'.

Discussion
Our two case studies of successful and less successful placements amplify the model developed from our statistical measures. They show that the foster carer needs to be sufficiently skilled in working with attachment and handling difficult behaviour, that the child needs to offer something to the carers, even if this only evolves over time, and that the professionalism and absence of strain and other supportive relationships on the carer's part may help create conditions in which the positive interaction between carer and child can occur. Placements can, if the context and expectations of the placement are right and children and carers are sufficiently well matched, enable change to take place. This is attributable both to the conditions of the placement, including the match, expectations and support for the carer, and also the quality of responsive parenting on the part of the carer.
Of course, one possible explanation of the differential success of the two placements, as one of the helpful reviewers of this article commented, is that James, perhaps by reason of his early experiences of care giving by his grandmother, was capable of greater developmental progress than Giles, who might have evoked more negative parenting responses and shown as little change wherever he was placed. Do we know if it was Mrs Stanton's parenting style which had the positive effect, and would her skill and success in meeting James's needs have been as apparent were Giles to have been placed with her? Could Mrs Lowe have become Mrs Stanton if the children were exchanged?
Our model, with its emphasis on the characteristics of the child and the 'fit' between carer and child might arguably suggest that Mrs Stanton would have been less successful with Giles. As she herself acknowledged, 'in all honesty, there's some you bond with and some you don't. James, I instantly bonded with'. However, the statistical model described earlier suggested that the characteristics of the foster carers are very important in predicting outcomes, even when the difficulty of the child is allowed for, and that some carers were more successful than others. This, coupled with the evidence from the detail of the interview, would seem to lend support to our hypothesis, that the quality of parenting was critical to achieving success. Given the very considerable tantrums displayed by James at the outset, which seemed at least to equal those of Giles in seriousness, other limitations in the core conditions, and the relative lack of responsive parenting displayed by Mrs Lowe, it seems reasonable to hypothesize that she would have been no more successful with James than she was with Giles.
At the beginning of the article, it was argued that, although it occurred rarely after the initial period, foster placements could and should be expected to bring about greater positive change in individual children. The analysis of two cases provides a framework for showing how this might be achieved. The examples suggest the importance and usefulness of using the dynamics of attachment and interest sharing both with children who have serious attachment difficulties and complex needs, and with the key adults who work with them. As Ryan (2004) has argued, the theoretical framework seems particularly clinically useful because of its internal coherence and breadth, and because it allows an assessment to encompass the complex internal and interpersonal functioning of children, carers and other key professionals.
First, it offers a more complete explanation of the foster carer-child interactions which we have called 'responsive parenting' and by identifying two components of parenting, emphasizes the need to confront difficult behaviour while at the same time showing empathic sensitivity and enabling the child to feel secure and confident of not being rejected. Since this may be one of the most challenging tasks for those caring for these troubled children, we suggest that the ideas proposed by Heard and Lake allow us to develop our analysis and understanding of 'responsive parenting' in a way which provides a helpful theory about parenting. The framework can help us distinguish those foster-carer approaches which make the difference between the success of one case and the failure of another, and also provide an interpretive framework as an ongoing basis for training, and for clinical supervision of foster carers.
Second, and crucially, it draws attention within the attachment dynamic, to the needs and attributes of foster carers and how they interact with the children in their care. Our first case provided an example of a placement where the foster child's behaviour did change and showed some of the detail of the processes involved.
Both cases discussed suggest that if adult carers' self-defence systems are not to be activated and their capacity for care giving, growth and development, so vital in working with children with complex needs such as these, is not to be impaired, there is a need to ensure that their care seeking and their system for exploratory interest sharing with peers are acknowledged and attended to. Both cases show the ways in which the instinctive nature of the five systems operate in the interpersonal relationships between carers and their children, and that the dynamics of attachment and interest sharing may then be reflected in the interactions with their wider network (including schools, social workers, and child therapists.) The framework emphasizes the importance of ongoing clinical supervision and the need for a coherent system of support for foster carers, to provide the experience of containment postulated by Bion (1959) and described in Odegard (2005). It also has the potential to provide clearer direction to the initial assessment of the case, planning for intervention and more effective supervision of the therapeutic interventions being undertaken by carers.

Conclusion
In the first part of this article we examined findings from a large longitudinal study on foster care (Sinclair et al., 2004;Sinclair, Baker, et al., 2005; which suggested that children coming into foster care are likely to have, albeit to varying extents, difficulties relating to attachment, emotional and behavioural disturbance and social problems. As already described, the study found that, although their disturbed behaviour did seem to improve at the beginning of the placement, afterwards it was very persistent, changing little over the year the children were followed up. It was argued that the failure to bring about much behavioural change, coupled with the limited amount of permanency offered to these children, may in part be a reflection of the low expectations which all concerned have of foster care. Nonetheless, our questionnaires, statistical material and qualitative data showed how, despite difficult behaviour, skilled committed foster carers can nevertheless succeed, as Mrs Stanton did, in bringing about changes in their children's wellbeing, self-esteem and behaviour. There are compelling arguments for changing perceptions of foster care so that it is conceived of as an environment in which children's difficult behaviour, poor social skills and problems with relationships can be addressed. Not only is this desirable in itself, but ameliorating these difficulties is likely to prevent some of the negative spirals (where difficult behaviour prompts rejection and rejection further difficult behaviour) which we identified as a powerful contributing factor in placement breakdown. Furthermore, if foster carers can themselves help to change their children's emotional and behavioural responses, then these children's very limited coping resources and fragile attachment relationships can be sustained and strengthened within a secure setting, rather than being further taxed by therapeutic help from sources external to the environment of the foster placement, which our study in any case found to be largely ineffective in its impact on outcomes. Formulating placements within the framework of the dynamics of attachment and interest sharing could encourage agencies to view foster care as the arena for treatment, while at the same time building in provision for proper assessments, treatment plans, and training, supervision and support adequate to meet the interest-sharing and care-seeking needs of foster carers. Notes 1. A further follow-up of the placement cohort was conducted in 2001, and is reported in a third book (Sinclair, Baker, Wilson, & Gibbs, 2005). The data discussed in this article are taken from the first two surveys and case studies, conducted between 1998 and 1999. 2. Although some carers were more likely to have 'successful' placements, our statistical measure identified little change over the year in which the children were followed up, despite some evidence of change early on in the placement. There was however some evidence of less easily measurable changes, and of changes which had already occurred before we contacted the placement. 3. A longer account of this part of the study is given in , and of the statistical model in Sinclair and Wilson (2003). 4. The exception to this rule of no obvious effect was provided by contact with an educational psychologist. We explore possible explanations for this finding in Sinclair, Wilson, and Gibbs (2005, pp. 214-219). The positive effect of educational psychologists on outcomes was not however replicated in the final stages of our study (Sinclair, Baker, Wilson, & Gibbs, 2005.) Evidence from our case studies suggests the sometimes positive impact of consultations offered to foster carers by other professionals, for example based within CAMHS, but numbers of these were insufficient to be tested statistically. 5. Names and some details of all those discussed in the cases have been changed to ensure anonymity.