States, traits and meaningless debates
The standard way of thinking about the relationship between psycho-pathology and time requires making a distinction between states and traits. Like many concepts widely employed by psychopathologists, this distinction seems commonsensical when first encountered but, on closer examination, appears more suspect. A state is said to be some characteristic of the individual (a type of cognitive organization or a facet of personality) that is present during episodes of illness. A trait, on the other hand, is a more enduring characteristic that seems to precede the onset of complaints, and which is often thought to be aetiologically important. This distinction is explicitly evoked in many of the theories we have encountered earlier. The models proposed by Paul Meehl13 and Gordon Claridge14 both suggest that schizotypal personality traits predispose people to develop schizophrenic symptoms, an idea that is mirrored in the stress-vulnerability models of psychosis developed by researchers such as Keith Nuechterlein and his colleagues in the USA.15 The distinction between states and traits is also important in some theories of depression. For example, as we have seen, Martin Seligman, Lyn Abramson and others have argued that a pessimistic attributional style predisposes people to become depressed when they are exposed to unpleasant events.16
There are two problems with this distinction. First, and most obviously, it is doubtful whether states and traits are two distinct types of psychological phenomena. Psychological characteristics probably vary in their mutability from highly unstable and state-like to highly stable and trait-like. Indeed, we have already seen in Chapter 10 that individuals vary in the extent to which at least one important characteristic – global self-esteem – is trait-like. Michael Kernis found that, in some people, global self-esteem is relatively stable over time whereas, in others, it fluctuates dramatically.17 He also found that individuals whose self-esteem fluctuated greatly tended to make extreme attributions about hassles they encountered in their lives.18 Of course, observations of this kind are not too difficult to accommodate so long as we realize that we are talking about gradations of psychological flexibility, rather than two different kinds of processes.
A deeper and more sinister difficulty concerns the inferences that are sometimes made on the basis of this distinction. It is often assumed that state-like processes are mere epiphenomena of symptoms, and therefore causally unimportant. According to this way of thinking, unless the psychological characteristics we observe in patients can be shown to be present prior to episodes, they can play no causal role. The considerable efforts that have been directed towards determining whether psychological processes implicated in complaints are trait-like or state-like have often been motivated by this assumption.
The despair sometimes experienced by psychologists on discovering that their favoured psychological mechanisms are not trait-like may be something of an over-reaction. There is no reason why we should not attribute a causal role to psychological processes that vary alongside complaints. To see why this is the case, it will be helpful to recall the model of the attribution–self-representation cycle that I introduced in Chapter 10 (p. 254–62) According to that model, beliefs about the self influence appraisals of events, which in turn influence future beliefs about the self. We saw that one property of a non-linear system of this kind is that, under certain circumstances, the positive feedback loop between one process and another allows marked shifts in both over time. They are like neatly balanced weights that topple together when given a push of sufficient magnitude. The person making an internal attribution for a negative event feels more miserable about herself as a consequence, and this increases the probability that she will make an internal attribution for negative events in the future. Over a period of time, the psychologist observing this process sees someone whose attributions become more self-blaming at the same time as she becomes more depressed. However, this does not mean that her attributions were causally unimportant in this process.
Of course, when two people react to exactly the same event in different ways – one with stoicism and the other with distress – we can be sure that there must be something different about them to begin with. However, these differences need not be dramatic at the outset in order for them to spiral into extreme and more easily detected differences later on. Given the inherently unstable nature of psychiatric complaints, and the complex functional relationships that lie behind them, I suspect that direct or indirect feedback loops of this kind affect many of the psychological processes involved in mental illness.
Chaos and complexity
If the analysis I have just given is even half correct, fluctuations in psychotic complaints might provide a valuable source of information about the psychological processes that give rise to psychiatric problems. Unfortunately, these kinds of fluctuations, although widely recognized by clinicians, have rarely been studied in detail. A handful of studies that are exceptions to this rule are worth looking at in some detail.
American psychiatrists Allan Gottschalk, Mark Bauer and Peter Whybrow studied day-to-day fluctuations in mood in a small group of bipolar patients over periods ranging between one year and two and a half years. Using complex mathematical techniques, they searched their data for patterns but found no evidence of cyclic variations between positive and negative moods. Indeed, Gottschalk and his colleagues concluded that the mood changes in their patients appeared to be formally ‘chaotic’ (see Figure 16.5).19
Although these findings show that patients’ moods do not follow a regular rhythm, they do not imply that mood changes are the products of random mechanisms. Although in popular language the term ‘chaos’ implies disorder, in the mathematical sciences it is used to describe the complex and unpatterned behaviour that often arises out of quite simple systems of interacting processes. Chaotic systems are deterministic, which is to say that each element in the system influences the others in ways that can be precisely specified as functional relationships (usually in the form of mathematical equations). Nonetheless, because of the cumulative effects of these influences, the overall behaviour of a chaotic system is highly unpredictable even over short periods of time (changes in the weather, and long-term variations in populations, tend to be chaotic in this formal sense).20 Despite this unpredictability, chaotic systems tend to exhibit a degree of self-organization or
Figure 16.5 Daily mood variations (ranging from ‘best I have ever felt’ to ‘worst I have ever felt’) for seven bipolar patients (from Gottschalk et al., 1995).
homeostasis. (For example, although annual population changes in some species are chaotic, under most conditions the population levels remain between certain minimum and maximum levels; similarly the average rainfall in a particular region and in a particular season is reasonably predictable.)
A more recent series of studies, conducted by Inez Myin-Germeys and her colleagues at Maastricht, in Holland, has revealed something of the mechanisms involved in these kinds of mood fluctuations.21 Myin-Germeys used portable electronic bleepers to cue participants to fill in simple questionnaires about their experiences at random points in time. (In Chapter 9 we saw that this experience-sampling method has been used to demonstrate that patients with flat affect experience subjective emotional states that are as strong as those of ordinary people.) Her participants were remitted psychotic patients, first-degree relatives of patients, and ordinary people. When the bleepers went off, the participants were asked to record their current mood, to describe the most important event that had happened to them since the last time the bleeper had sounded, and to answer a series of simple questions about how stressed they were feeling. In all of the groups, it was found that experiencing subjectively stressful events was associated with increases in negative affect and decreases in positive affect. However, the patients reported less positive affect and more negative affect than their relatives or ordinary people, made more extreme appraisals of the events they had experienced, and reacted to the experiences with more extreme shifts in mood. On Myin-Germeys’s
own account, these findings are evidence of extreme sensitivity to stress in psychotic patients. This, of course, is true, but they are more specifically consistent with the idea that psychotic patients make abnormal attributions for negative events, and with the model of the attribution– self-representation cycle that I developed in earlier chapters.
In an interesting further study, Myin-Germeys found that sensitivity to everyday stressors, as assessed by her experience-sampling method, was unrelated to neurocognitive functioning, as measured with the kinds of tests favoured by American neuropsychologists. She therefore argued that neurocognitive deficits and stress sensitivity separately and independently contribute towards vulnerability to psychosis.22 Of course, this is exactly what we would expect from the research we considered earlier in this book (especially in Chapter 8, where we saw that neurocognitive deficits appear to play little or no role in positive symptoms).
Understanding relapse
The account I have just given of the ups and downs of mental health has been fairly speculative – it is more an agenda for further research than a polished theory. It also has at least two important limitations. First, I have said very little about the kinds of external events that buffet the individual, bringing about changes in mood and symptoms. Second, I have not considered the more global changes that occur as patients move from periods of remission, in which they are relatively well, to periods of illness. In the remaining part of this chapter I will attempt to show how the account I have offered can be extended still further to overcome these limitations.
Relapse is rarely a simple step-like event. Rather, the transition from remission to active psychosis is usually gradual, often taking a number of weeks, during which time the patient often shows evidence of the kind of chaotic symptom fluctuations that we have just considered.23 Two clues suggest that social cognition may play a crucial role in this process. First, as we saw in Chapter 9, studies of the prodromal symptoms that precede relapses have shown that these mainly consist of depression, anxiety, irritability, tension and sleeplessness.24 (Episodes of mania appear to be an exception to this rule, as they are usually preceded by periods of excitement, energy and sleeplessness.25 However, we saw in Chapter 11 that fear of incipient dysphoria may nevertheless play a crucial role in the onset of manic symptoms.) Although it is possible, as some psychologists have argued,26 that these mood symptoms are brought about by the patient’s perception that another episode of illness is imminent, the absence of frank psychotic symptoms during the early stages of relapse suggests that this is usually not the case. It seems much more likely that changes in patients’ attributions, their beliefs about themselves, their perceptions of other people and related processes bring about initial changes in mood, which later cascade into full-blown psychosis.
The second clue concerns the kinds of events that are known to influence the likelihood that a relapse will occur. Many types of stressors can make vulnerable individuals suffer an exacerbation of their psychotic complaints. Some, for example the taking of illicit drugs or abstaining from sleep, can be thought of as biological stressors, because they have obvious effects on the individual’s physical as well as psychological well-being. However, the best-understood stressors are social, and are precisely the kinds of events that are likely to challenge the individual’s social-cognitive system.
Why Stressful Relationships are Damaging to Mental Health
Most of us intuitively recognize that emotionally charged relationships, even those in which the dominant feeling is love, can make us feel miserable. The idea that such relationships can have especially adverse effects on people who are vulnerable to psychiatric symptoms has a long history. In a later chapter, we will see that theories that implicate dysfunctional family relationships in the onset of psychosis were popular in the 1950s and 1960s, but have usually been discounted in recent years. However, the idea that families can have a negative influence on the future well-being of people who are already ill is now widely accepted. Studies of this effect began with the work of a small group of investigators based at the Institute of Psychiatry in London in the late 1950s. Initially, this group consisted of Jim Birley, Morris Carstairs and Gillian Topping. However, the person who was to have most influence on how the work progressed was George Brown, an anthropology graduate who was later to become Professor of Sociology at Birkbeck College.
According to his own account, Brown arrived at the Institute knowing very little about serious mental illness:
A… dominant memory of those days was my emotional revulsion on reading some of the standard psychiatric accounts of the condition [schizophrenia]. Was it all so clear-cut? Were the core symptoms and handicaps so clearly linked to underlying endogenous processes? My feelings were so strong that it was several years before I could force myself to finish the account of schizophrenia in Mayer-Gross, Slater and Roth’s (1954) Clinical Psychiatry.27
Brown was interested in the fate of those patients who, in an era that preceded care in the community, were able to achieve discharge from psychiatric hospitals. Unlike Ronald Laing, who was beginning his work at about this time, Brown did not feel that he had any particular skill at communicating with mad people. Instead, therefore, he decided to focus on the experiences of their parents and other relatives.
An early study conducted by Brown, Carstairs and Topping published in 1958 produced an unexpected and, at the time, quite unsettling result. Following up 229 recently discharged men, most of whom had a diagnosis of schizophrenia, they found that those leaving hospital to live with wives or parents were less likely to remain out of hospital, and hence more likely to relapse, than those leaving to live in lodgings or with brothers or sisters. Among those who had been married, those who were widowed, separated or divorced were more likely to remain well than those who left hospital to live with their spouses. Among those who went to live with their parents, those who spent most of their days with their mothers were particularly likely to do badly. Although various interpretations of these findings were considered, it was impossible to escape the conclusion that close family relationships could be hazardous to patients who were recovering from a psychotic illness. This result was quite opposite to the researchers’ expectations.28
The next step was to discover which aspects of family relationships had this toxic effect. In a prospective study in which relatives were interviewed as patients were discharged from hospital, Brown observed that patients returning to live in families characterized as having a high level of ‘emotional involvement’ had especially high rates of relapse. This risk was reduced in those families in which the mother worked, limiting the amount of time that she spent in face-to-face contact with her mentally ill son or daughter.29
Soon afterwards, Michael Rutter joined the team and an attempt was made to develop a measure of family atmosphere that would adequately capture the crucial emotional characteristics. After two years of development work, the team came up with the Camberwell Family Interview (CFI), a structured assessment in which a close relative is asked about the patient’s behaviour and its impact on the household.30 Based on the relative’s responses, five scores are obtained from the interview. Critical comments, which are simply counted, are remarks that express a negative emotional attitude towards things that the patient has done. Hostility, on the other hand, is rated on a five-point scale and refers to a general negative attitude towards the patient’s personality. Emotional over-involvement, also rated on a five-point scale, refers to extreme emotional distress experienced by the relative accompanied by self-sacrificing and over-protective behaviour towards the patient. Positive comments, which are counted, are remarks which praise or express a positive attitude about things the patient has done, and warmth, rated on a five-point scale, measures general positive attitudes towards the patient as a person.
In practice, the positive comments and warmth scales are rarely used, as they have been found to have little or no predictive value. However, scores on the critical comments, hostility
and emotional over-involvement scales were found to be associated with a high risk that patients would relapse. It is easy to picture the emotional tone associated with high scores on each of these scales – anger at the patient for being ill in the case of critical comments and hostility, and guilt about the origins of the illness in the case of emotional over-involvement. In the years following Brown’s work it has become commonplace to describe a relative who scores highly on any one or more of these scales as exhibiting high expressed emotion (or EE). It is unfortunate that the label ‘high EE relative’ is now occasionally used pejoratively by mental health professionals, in a way that gives the impression that there is something nasty or unpleasant about those who are so categorized. Paradoxically, perhaps, high EE parents often seem to be very caring people, who have been emotionally overwhelmed by the trauma of seeing their sons or daughters afflicted by a condition that appears to be severely disabling, but which has no obvious physical cause.
Madness Explained Page 48