Introduction
Although at first glance melancholia and grief share some characteristics, the layperson of a century ago would have little difficulty distinguishing them. Paradoxically, the same layperson today might be confused, intuitively recognising a difference, yet forced perhaps to see both as forms of depression. Both have biological correlates and psychological parameters, but classical melancholia is immediately recognisable to an experienced clinician largely as a ‘biological disorder’, and grief, perhaps, as a relational process: a consequence of having been ‘attached’.
Current diagnostic systems focus largely on reliable measurement of objective criteria in the present, despite widespread doubts about the validity of such an approach, and considerable interest in the doctrines of multicausality and equifinality. Meyer's earlier psychobiology stressed the importance of understanding a person's psychopathology and presenting phenomenology through their life story. We would argue that a failure to appreciate the functional significance of phenomenology, and the current and developmental context in which it developed, risks generating erroneous conclusions about the nature of the problem and potentially inappropriate treatment. Various theories are available in which to couch such an argument. Our emphasis is on attachment theory, which will be summarised briefly before reviewing some ambiguous clinical situations that can be informed by this understanding.
Overview of attachment theory
Over the recent decades, attachment theory has become a key guiding theory in research in developmental psychology, psychopathology and neurobiology. It is the dominant model in child and adolescent psychiatry practice. Peter Fonagy and Jeremy Holmes especially have established its role in adult psychotherapy [e.g. Reference Fonagy(1,Reference Holmes2)] and it has made a substantial contribution to the understanding of developmental trauma Reference Glaser(3). Whilst individual practitioners have used the model in other fields (e.g. Koslowska's subtyping of somatoform presentations by attachment style) Reference Kozlowska and Williams(4), it rarely features in articles on diagnosis and treatment in many mainstream psychiatry journals, is ignored in much adult practice and is largely absent from the Diagnostic and Statistical Manual of Mental Disorders (DSM), despite considerable capacity to inform care.
Attachment theory arose out of the work of Bowlby Reference Bowlby(5). Although psychoanalytically trained, he broke with the tradition of the day in emphasising the importance of real relationships in an individual's developmental history over Freudian drives or Kleinian infantile fantasy. He was influenced by the emerging field of ethology, the study of animal behaviour: Harlow's research on the effects of separating infant monkeys from their mothers and Lorenz's work on imprinting and critical periods in goslings' connection to their mothers.
Bowlby's research in orphanages and paediatric wards led him to describe a characteristic process in the wake of parental loss: first protest, then despair and then detachment. In orphanages run along strict hygiene lines or with a perfunctory attitude of the staff, infants failed to thrive and frequently died, but in orphanages where staff were warmer and more maternal in their orientation, infants survived much better. Bowlby's work had an almost immediate effect on liberalising visiting practices in paediatric wards and initiated the move to deinstitutionalise orphanages.
Fish, reptiles and amphibians endure largely independent of parenting by producing excess offspring. Among birds, and especially among mammals, smaller numbers of offspring are nurtured to greater or lesser extent, for varying periods of time. These offspring need to master their environment but remain safe while doing so. There is some parallel between the duration of that nurturance and the social complexity of the species. Attachment theory posits that homo sapiens is a highly social species and human neurobiology is fashioned primarily towards the formation and maintenance of relationships. The principles are summarised here:
1. We have an evolved drive both to explore the environment and to link to an attachment figure (AF). This is most obvious in infancy.
2. The AF functions as a ‘secure base’ from which the infant can safely explore.
3. The relationship with AF is characterised by ‘proximity seeking behaviour’ evoked particularly by distress: separation, danger, fatigue, etc.
4. The infant develops a map or ‘internal working model’ (IWM) of the world, containing the whereabouts and likely responses of their AFs. This is modified throughout life, but earlier IWMs seem to shape later ones.
Central to attachment theory is the idea that a child's (or any person's) behaviour cannot be understood except in the context of their relationships. Hence, Winnicott's statement, ‘There is no such thing as a child’. Research with carer-infant dyads shows that newborns are primed for engagement Reference Meltzoff and Moore(6), and that mirroring and carer attunement is vital to normal development Reference Stern(7). Attuned, reflective parenting provides contingent, congruent but ‘marked’ responses (the self and infant clearly distinguished) Reference Fonagy, Gergely, Jurist and Target(8), similar to what we would expect of a good psychotherapist. It is in that inter-subjective space over time that narrative identity develops. The quality and predictability of parenting shapes biases in the child's behaviour and in their cognitive and emotional information processing, which are stored in (largely unconscious) procedural memory. We know them by many names: Bowlby's IWMs, neurotic habits, characteristic patterns, dispositional tendencies, schemata, etc.
Ainsworth Reference Ainsworth, Blehar, Waters and Wall(9) developed the Strange Situation Procedure (SSP) to assess toddlers' attachment to their parents. For attachment behaviour to be evoked, the child must be stressed. The SSP uses a series of separations and reunions between child, stranger and mother to do this, with particular analysis of the child's reunion behaviour. Her original work on normative samples identified three patterns:
1. ‘Type B’: Securely attached children protest on the parent's disappearance but are easily pacified on their return and resume exploratory play. Carers exhibit attuned parenting styles: predictably responsive to their child's needs and cues for arousal modulation, space or closeness.
2. ‘Type A’: Insecure-avoidant children protest little overtly, but on the parent's return hover warily or appear unconcerned. Play may or may not continue. Carers usually exhibit a dismissive parenting style: predictably less responsive especially to their child's need for nurturance.
3. ‘Type C’: Insecure-ambivalent children protest but cannot be pacified by the returning parent, clinging and resisting overtures to play. Carers usually exhibit an enmeshed parenting style: benevolent but inconsistent or unpredictable responses that lead the child to stay close but unsatisfied.
Even within Ainsworth's normative samples many children could not be classified, and within clinical samples this is strikingly so, some exhibiting strange and contradictory behaviours. From this observation Crittenden Reference Crittenden(10) developed the ‘A/C’ category and Main and Solomon Reference Main, Solomon, Yogman and Brazelton(11) the ‘D’ (Disorganised/ Disoriented) category, both of which showed strong correlations with histories of maltreatment. Debate continues as to whether these children are actually ‘disorganised’ (D), lapsing into dissociation in the face of fear Reference Hesse and Main(12), or whether in their desperation they are still strategic in deploying more extreme, complex and combined A and C behaviours Reference Crittenden(13).
An alternative perspective might be that the ‘D’ category confounds specifically trauma or loss related phenomena with the more general style of attachment. Trauma and loss occur more often in the environments that foster disturbed attachment and the later effects of both can seem similar. Just as in the clinical situation it is necessary to distinguish acute phenomenology from background disturbances in adaptive function, so the assessment of attachment classifications might be confounded by unresolved loss or trauma, depression and post-traumatic stress disorder (PTSD). This is a sort of Axis 1/Axis 2 distinction, with the potential for error in both research and clinical practice: disturbance by unresolved trauma can make an otherwise secure person seem preoccupied; denial of trauma can make an otherwise secure one seem dismissive on certain subjects.
A detailed comparison of attachment classification systems is beyond the scope of this article. Although there is overlap in the scoring systems, there is substantial discrepancy too (Reference Crittenden, Claussen and Kozlowska14–Reference Crittenden and Newman16). Those discrepancies may account for some of the inconsistencies in research findings.
Most of the studies have shown little or no correlation between attachment and temperament, and limited heritability Reference Fearon, Van Ijzendoorn, Fonagy, Bakermans-Kranenburg, Schuengel and Bokhorst(17). Rat cross-fostering studies make it quite clear that parenting behaviours are learnt, induce changes in gene expression in the offspring (epigenetics) and are passed on environmentally Reference Meaney(18,Reference Champagne and Curley19). Recent primate studies parallel Harlow's landmark studies while also revealing the effects of disrupted attachment on both neurobiology and gene expression Reference Sabatini, Ebert, Lewis, Levitt, Cameron and Mirnics(20).
Early work in infant attachment was followed by similar research across the age range. Most of the work has been done with Main and Goldwyn's Adult Attachment Interview (AAI) Reference Main and Goldwyn(21) to assess adult attachment representations from discourse analysis of structured interviews (Crittenden uses the same interview but scores it differently). There is significant correspondence between the attachment status of parents on the AAI, the attachment status of their offspring in infancy and the later attachment style of those infants in adulthood. The AAI also confirms what we know from clinical practice: it is not the fact alone of neglect, trauma or triangulation in childhood that distorts development, but the manner in which that experience is processed.
AAI ‘dismissives’ distance themselves from the emotional significance of experiences: narratives are curt, factual, lacking depth. This is the equivalent of infant insecure-avoidant. ‘Preoccupied’ adults by contrast provide lots of incoherent or contradictory facts, with a bias to emotional exaggeration, the equivalent of infant insecure-ambivalent. A balanced or secure adult provides a coherent, integrated narrative, acknowledging good and bad, fact and emotion, their own perspective and that of others: autonomous. The AAI also identifies a group who are ‘unresolved’ with respect to loss or trauma. This manifests as discontinuities and dissociative features. Originally seen as a discreet category (the adult equivalent of disorganised), they can usually still be ‘forced’ into one of the three major groups, or combinations thereof.
Finally, it is also worth noting that ‘secure’ attachment is not always adaptive. On average, the secure infant or the adult has access to a wider repertoire of coping strategies, but we learn the attachment style that works best in our developmental environment. For both adults and infants, there are extreme social and physical environments where either aloof distancing or intense enmeshment will better ensure survival.
Attachment and psychopathology: limited research findings
Dozier Reference Dozier, Chase Stovall-McClough, Albus and Cassidy, Shaver(22) provides a recent review of this area, emphasising that although a number of adult conditions show associations with insecure or disorganised attachment, proven predictive connections are scant. In depression, for instance, a role for early loss has long been accepted, but discreet associations with specific attachment types are largely unproven, and depression itself is very heterogeneous. Insecure attachment does predict depression in adolescence. There is a further suggestion that attachment at the ‘avoidant’ end of the spectrum, where children minimise the expression of negative affect and dependency needs, would be associated with a more ‘internalising’ type of depression, where more ‘ambivalent’ ones might develop a more hostile-externalising illness. One study supports the differential effectiveness of interpersonal psychotherapy as against cognitive behavioural therapy, using the ‘theory of opposites' to provide a more emotion focussed intervention for constrained internalisers and a more cognitive one for emotive externalisers. This is certainly consistent with Crittenden's model. Insecure-ambivalent/resistant attachment also shows an association with anxiety in 17-year-olds but does not predict discreet disorders. There are some softer associations with eating disorders and personality styles, an association between PTSD and ‘unresolved’ status on the AAI and a relationship between disorganised attachment and borderline PD. Dozier closes by stating ‘at this point, the only clear connections between infant attachment and adult psychopathology are between disorganised attachment and dissociative symptoms in adolescence and early adulthood, and between resistant attachment and anxiety disorders in adolescence’ (p. 736). Notwithstanding the paucity of applied research in the broader application of these ideas, the clinical usefulness of this conceptualisation is apparent.
Attachment and psychopathology: some clinical lessons
The concept of diagnosis remains challenging for psychiatry. Most of the major disorders have striking features and associated clinical paths that suggest clear-cut boundaries and easy definition. Psychosis, bipolar disorder, autism and attention deficit disorder seem at first glance far removed from ordinary human behaviour and difficult to comprehend as learned, adaptive strategies. Clinical experience suggests, however, that disturbances of adaptation can present in ways that overlap with these phenotypes and that a premature assumption of linear, biological causality is not necessarily accurate.
Personality disorder
Genetic factors account for about 50% of the variance in PD, but specific biological pathways or pharmacotherapy have yet to emerge. Most of the PD research focuses on borderline and antisocial patterns, so it is difficult to develop comprehensive models. There is a recognised association between these and disorganised attachment, but we have yet to seriously examine possible links between ‘softer’ forms of attachment bias and other PDs: the emotional inhibition and detachment of obsessive-compulsive disorder are similar to avoidant attachment; the exaggerated emotions of the histrionic suggest ambivalent attachment.
Therapy focussed on behaviour can work for symptoms (e.g. dialectical behaviour therapy), but enduring change in underlying identity seems to require time and relationships. Attachment theory, like psychoanalysis, stands to inform that process by looking at the meaning of the behaviour rather than the behaviour itself. From a descriptive perspective, the withdrawal of Avoidant and Schizoid PD are similar but they serve different functions: a clinician would rarely have difficulty distinguishing them, but scoring systems do. Therapy has tended to focus on maladaptive behaviour in the present, and can certainly be effective at that level, but to the extent that PD represents a working model developed from earlier relationships, it is more relevant to understand the positive purpose it serves: the danger from which it protects.
Depression
We used to believe the dictum that patients with melancholia would respond to medication or electroconvulsive therapy, and those without would not. The Diagnostic and Statistical Manual of Mental Disorders, 3rd Edition, of the American Psychiatric Association (DSM-III) discarded that distinction and the advent of SSRIs ushered in a period of wider prescribing, yet there seems some truth in the original premise. We are unable to confidently predict who will and will not respond (‘endogenous’ illnesses may involve psychological triggers, and some ‘exogenous’ cases respond to medication), but there are clearly differences. Indeed, most of the factors that predict antidepressant response are psychological and not biological Reference Dodd and Berk(23). It is unfortunate that a diverse collection of disturbances have come together under the ‘depression’ label Reference Parker(24) and that treatment algorithms focus largely on severity rather than formulation or comprehensive understanding.
Melancholia and bipolar aside, depression is frequently about loss: relationships, role, self-esteem and identity, etc. To understand the nature of that loss, we need to know the person's previous experiences of relationships and the basis of their own identity. Identity is formed in relationships with others, and our first relationship is a major determinant.
Psychosis
DSM classifications similarly struggle to capture the diversity of psychotic phenomenology. The distinction between hallucinations and pseudo-hallucinations, between delusions, referential thinking and over-valued ideas, is not as clear as we would have our trainees believe. The supposedly transient and reactive psychotic symptoms of borderline PD can be enduring, in the absence of any other evidence of discrete Axis I disorder. Clinicians new to adolescent psychiatry may find themselves diagnosing lots of schizophrenia because so many young people from traumatic backgrounds hear voices, and their apparently ‘dissociative’ experiences can be indistinguishable from the hallucinations of schizophrenia Reference Jessop, Scott and Nurcombe(25). There may be emotional awkwardness or blunting as well. While early intervention in psychosis is important, it is important to differentiate what may actually be a psychological device to minimise iatrogenic harm. Emotional detachment is typical of those operating at the ‘avoidant’ end of the attachment spectrum, and dissociative symptoms are common in those with prior disorganised attachment. Such people may require psychotherapy as primary treatment.
Bipolar disorder
As we move from classical disorder to ‘rapid cycling’, ‘ultra-rapid-cycling’ and ‘mixed states', the boundaries become blurred and the response to medication less predictable Reference Paris(26). Mood-stabilisers work for bipolar disorder; but their value in other conditions, especially borderline personality disorder (PD), PTSD and ‘ordinary’ moodiness, is more questionable. Nevertheless, some patients seek them and doctors can prescribe them, and the ‘Bipolar’ label may be used to endorse this process: it is possible to reframe an ordinary ‘downer’ into a depressive episode and to upgrade histrionic or attention-seeking behaviour, nervous laughter and rushed conversation to ‘hypomania’. Many individuals find personality diagnoses more stigmatising than medical ones, and, a medical diagnosis can externalise responsibility. ‘True’ Bipolar disorder can, of course, be sub-syndromal, and benefit from medication. It is equally true that a fragile sense of one's own identity and insecurity in past and present relationships can generate dramatic mood swings. The astute clinician should explore those common possibilities before relying on phenomenology alone to support a default diagnosis.
Paediatric bipolar disorder
There has been considerable debate about this disorder Reference Parry and Allison(27) and attachment issues may not be sufficiently considered Reference Parry(28). It is certainly clear that bipolar disorder often begins in adolescence (e.g. recall of ‘first symptoms' median age 17, recall of ‘first manic symptoms' median age 21 Reference Berk, Dodd and Callaly(29)), but pre-pubertal cases are rare. Our focus here is not on whether it exists, but on the dangers of failing to explore other possibilities. ‘Antisocial’ behaviour is normative for toddlers. Impulsivity and emotional volatility are characteristic of all children, and especially so of those from disrupted backgrounds. Children who cannot access a secure AF to foster emotional self-regulation develop other strategies. Severely insecure-avoidant children (Crittenden's ‘Compulsive As') inhibit negative affect until they explode. Severely insecure-ambivalent children (‘Coercive Cs') exaggerate negative affect to coerce marginal caregivers into providing what they need, oscillating widely from furious rage to pathetic helplessness. Both sorts of children can find themselves anergic, lowered in arousal, ‘depressed’; both can be raging, destructive demons, ‘high’. An exploration of the developmental story, and observation of dyadic behaviour, may provide a strategic understanding of the behaviour from which appropriate psychotherapy can evolve.
Attention deficit hyperactivity disorder
Here, again many children can be made to meet diagnostic criteria, and some will benefit dramatically, at least in the short term, from stimulants. Their long-term value is however less clear Reference Swanson, Arnold and Kraemer(30), as is whether response to stimulants is a marker of illness or an intrinsic pharmacological response to stimulants evident at a population level. Some attention deficit hyperactivity disorder (ADHD) appears ‘hard core’ biological, but in others it is better understood as emotional dysregulation in an awkward attachment relationship, perhaps with past trauma. Although earlier research cast doubts on the benefits of behavioural interventions, later reviews Reference Fabiano, Pelham, Coles, Gnagy, Chronis-Tuscano and O’Connor(31) have been more positive, especially where the focus is on disruptive behaviour. The emphasis has been on parent training. In the clinical situation, there is a subset of children whose ADHD behaviour is strikingly contingent on parental arousal and attunement. In dyadic therapy, these parents can develop an understanding of the strategic nature of their children's actions and be coached to respond more effectively. Therapy may explore, in the child's presence, the living history of the parent-child relationship, thus enhancing the child's knowledge of integrated emotional experience. Behaviour can settle at school and learning improves.
Asperger syndrome
Our understanding of autistic spectrum disorder (ASD) is limited, but the evidence for a genetic dimension is very strong. We have rightly discarded the parent-blaming ‘refrigerator mother’ models prominent decades ago. At first glance it seems hard to imagine any diagnostic confusion. However, the Romanian orphans cohort showed how sustained early attachment deprivation could lead to autistic spectrum features Reference Rutter, Anderson-Wood and Beckett(32). In the clinical situation, there is a group of children diagnosed with autism spectrum who ‘get better’ in a way inconsistent with that diagnosis. The diagnostic features of ASD emphasise impairment in social-emotional reciprocity and rigidity in habits and routines, but these features can come from different causes. Closer observation of some of these children shows subtle differences. Not the odd, off-target eye contact of autism, but gaze-avoidance or secret-looking. Not clumsiness with emotions but concealment of their own and noticeable but covert responsiveness to others' changes in play and proximity consistent with the material being discussed in the room.
Medication can be useful to ease arousal in ASD, but relational interventions may be more useful for some children. All children benefit from structure and clear communication, and greater awareness by the parent of how the child might be understanding the world. Meticulous attention to those issues helps autistic children to function better. In the non-autistic ‘look-alikes', the effects are transforming.
Conclusion
We are social animals. Our neurobiology evolved to enhance survival in relational contexts. Our developmental experience selects the patterns that will work in that setting: nature, nurture, niche; neurons, narrative, neighbourhood maybe. All are required to understand our behaviour.
The introduction of Axis II in DSM-III was an attempt to recognise the layered and complex nature of psychopathology, but the ‘objectification’ on both axes made that difficult. Denton Reference Denton(33) comments on the ‘notable and strikingly absent consideration of the role of relational processes and disorders in the development, maintenance and manifestations of mental disorders' within DSM-IV, and DSM5 is looking very similar. Augmenting a DSM diagnosis with a formulation-based understanding of the individual facilitates a more fine-grained conceptualisation of a person's presenting issues.
Ironically, mainstream psychiatry has become ‘detached from attachment' at the very time that neurobiology is beginning to make sense of it. An understanding of attachment concepts has the potential to enrich clinical practice. Diagnostic criteria are necessary but not sufficient for a sophisticated psychiatry.
Disclosure
Professor Michael Berk has received research grants and support from the Stanley Medical Research Institute, MBF, NHMRC, Beyond Blue, Geelong Medical Research Foundation, ARHRF, Bristol Myers Squibb, Eli Lilly, Glaxo SmithKline, Organon, Novartis, Mayne Pharma, Servier and Astra Zeneca. He has consulted for Astra Zeneca, Bristol Myers Squibb, Eli Lilly, Glaxo SmithKline, Janssen Cilag, Lundbeck, Pfizer and has been a speaker for Astra Zeneca, Bristol Myers Squibb, Eli Lilly, Glaxo SmithKline, Janssen Cilag, Lundbeck, Organon, Pfizer, Sanofi Synthelabo, Solvay and Wyeth.
