Anxiety and Cognition

A Unified Theory

Michael Eysenck

• Publisher: Psychology Press
• Year: 2014
• ISBN:
  9781317775034
• Edition: 1
• Topic: Psicología
• Subtopic: Historia y teoría en psicología

It is argued in this book that there are three major approaches to anxiety. First, there is anxiety as an emotional state. Second, there is trait anxiety as a dimension of personality. Third, there is anxiety as a set of anxiety disorders. What is attempted is to produce a unified theory of anxiety which integrates all these major approaches. According to this unified theory, there are four sources of information which influence the level of experienced anxiety: (1) experimental stimulation; (2) internal physiological activity; (3) internal cognitions, (e.g., worries); and (4) one's own behaviour. The unified theory is essentially based on a cognitive approach. More specifically, it is assumed that individual differences in experienced anxiety between those high and low in trait anxiety depend largely on cognitive biases. It is also assumed that the various anxiety disorders depend on cognitive biases, and that the main anxiety disorders differ in terms of the source of information most affected by such biases (e.g., social phobics have biased interpretation of their own behaviour). In sum, this book presents a general theory of anxiety from the cognitive perspective. It is intended that this theory will influence theory and research on emotion, personality, and the anxiety disorders.

Correction notice: Christos Halkiopoulos should have been credited for his role as the inventor of the Dot Probe Paradigm and for the design and execution of the experiment discussed in C. D. Spielberger, I. G. Sarason, Z. Kulczar, and J. Van Heck (Eds.), Stress and Emotion, Vol. 14. London: Hemisphere.

Chapter One

Theories of trait anxiety

Introduction

Anxiety has become an increasingly fashionable focus of research in recent years, and it may be wondered whether yet another book on anxiety is needed. However, most books on anxiety are rather limited in scope. More specifically, they generally focus on only one of three main areas of research in which anxiety is of major importance: emotion; personality; and clinical or abnormal psychology. In contrast, an attempt (ambitious or foolish) has been made in this book to consider anxiety in all three areas within a common theoretical framework. It is for this reason that the title of this book refers to a “unified theory of anxiety”.

There are numerous theories of emotion, most of which include a list or classification of the basic emotions. Anxiety (or fear) is regarded as one of the basic emotions in nearly all of these theories (e.g., Plutchik, 1980; Russell, 1991). As far as personality research is concerned, trait anxiety (or neuroticism) is included as one of the major dimensions of personality in most contemporary theories (see Digman, 1990, for a review). Finally, anxiety is of major concern in clinical psychology. For example, 12 different anxiety disorders are identified in DSM-IV (1994).

Why have most theorists concentrated on anxiety as an emotional state, or as an aspect of personality, or in connection with clinical disorders? One reason was identified by Eysenck (1992). As he pointed out, the goals of researchers in the clinical and personality areas have been rather different: “Those engaged in clinical research have focused primarily on cognitive factors involved in the aetiology and maintenance of clinical anxiety, whereas those investigating cognition in normal groups have concentrated on the ways in which anxiety affects performance” (p.l).

The starting point for the unified theory of anxiety was previous theory and research on trait anxiety in normal groups. Accordingly, relevant theories of trait anxiety are discussed in this chapter. In subsequent chapters, the new unified theory of anxiety is discussed with reference to an understanding of anxiety as an emotion, personality dimension, and clinical disorder. Four major issues provide recurring themes throughout the book. First, there is the issue of which factors determine the experience of anxiety. Second, there is the issue of the inter-relationships among the cognitive, physiological, and behavioural systems involved in anxiety. Third, there is the issue of why it is that some individuals experience much more anxiety than others. Fourth, there is the issue of the similarities and differences between anxiety in normals and in anxious patients.

Trait Anxiety

Personality theorists have spent several decades disagreeing among themselves about the number and nature of important personality factors or traits. Among the protagonists have been Cattell (e.g., Cattell, Eber, & Tatsouka, 1970), who argues that there are sixteen personality factors, and H. J. Eysenck (1967), who identifies only three. In recent years, however, there has been a developing consensus that there are five major personality factors, which are often called the Big Five. For example, Costa and McCrae (1985) identified the following five factors: extraversion; agreeableness; conscientiousness; neuroticism; and openness. There are minor quarrels about the nature of one or two of these factors (see Digman, 1990, for a review), but there is a large measure of agreement.

Although much has been accomplished at the level of personality description, this has not been matched by commensurate progress at the level of explanation. In other words, relatively little is known of the underlying causal factors and processes producing individual differences in these personality factors. What is missing from contemporary accounts was expressed in the following way by Block (1995):

The Big Five are often called ‘the five-factor model of personality’ and are referred to as providing an understanding of the ‘structure of personality’. As the term ‘model’ is used in conventional parlance among psychologists, it means a theoretically based, logically coherent, working representation or simulation that, in operation, attempts to generate psychological phenomena of interest. However, no identifiable hypotheses, theories, or models guided the emergence of or decision on this five-fold space (p. 188).

Virtually everyone agrees that one of the Big Five is a factor known variously as trait anxiety, neuroticism, or negative affectivity. Since much of this book is concerned with this personality factor, it is important to clarify the inter-relationships of these terms at the outset. Trait anxiety was defined by Spielberger, Gorsuch, and Lushene (1970) as “relatively stable individual differences in anxiety proneness” (p.3). Neuroticism closely resembles trait anxiety, with questionnaire measures of the two personality factors typically correlating approximately +0.70 with each other (Watson & Clark, 1984). However, there is one minor difference between them: trait anxiety correlates approximately -0.30 with extraversion, whereas neuroticism and extraversion are essentially uncorrelated (Watson & Clark, 1984).

In their article, Watson and Clark (1984) reviewed evidence indicating that a wide range of measures of neuroticism, trait anxiety, depression, and so on all inter-correlate highly with each other. They argued that these failures of discrimination indicate that it is inappropriate to assume that the various measures are actually assessing different personality factors. Their preferred solution is to assume that self-report questionnaires allegedly measuring trait anxiety, depression, and so on, are in fact assessing essentially the same personality factor of negative affectivity.

There have been various theoretical attempts to account for individual differences in trait anxiety or the closely related factors of neuroticism or negative affectivity. Some of these theories (e.g., H.J. Eysenck, 1967; Gray, 1982) have emphasised the role of heredity and individual differences in physiological activity. Other theories (e.g., M.W. Eysenck, 1992; Williams, Watts, MacLeod, & Mathews, 1988) have focused primarily on the role of the cognitive system in accounting for individual differences in trait anxiety.

Theories of H.J. Eysenck and Gray

Gray’s (1982) theory of trait anxiety and H.J. Eysenck’s (1967) theory of neuroticism both involved two major assumptions. The first assumption is that individual differences in the personality dimension of trait anxiety or neuroticism depend to a large extent on genetic factors. According to H.J. Eysenck (1967), “the evidence suggests fairly strongly that something like 50 per cent of individual differences in neuroticism and extraversion ... is accountable for in terms of hereditary influences” (p.210). Subsequently, he revised the figure upwards: “genetic factors contribute something like two-thirds of the variance in major personality dimensions” (Eysenck, 1982, p.28). According to Gray (1982), “studies of the personality traits of neuroticism and extraversion ... estimate the contribution of heredity to these conditions at about 50 per cent of the variance” (p.438).

The second major assumption is that heredity influences the level of trait anxiety or neuroticism via the physiological system. According to H.J. Eysenck (1967), individual differences in neuroticism depend upon the functioning of the so-called “visceral brain”. This consists of the hippocampus, amygdala, cingulum, septum, and hypothalamus. Somewhat similar structures were identified by Gray (1982). He argued that individual differences in trait anxiety depend upon the septo-hippocampal system, its neocortical projection in the frontal lobe, and its monoaminergic afferents from the brain-stem. He used the term “behavioural inhibition system” to refer to these structures.

Influence of heredity

The most important evidence relating to the role of heredity in determining trait anxiety and neuroticism comes from twin studies, and only this research will be considered in detail. However, it should be mentioned that the findings from adoption studies are broadly in line with those from twin studies, although they suggest a somewhat lower contribution of heredity than do twin studies (Brody, 1988).

The first relevant twin study seemed to provide remarkably strong support for the importance of heredity. H.J. Eysenck and Prell (1951) studied neuroticism, and found that the correlation between monozygotic twins was +0.85, compared to only +0.22 for dizygotic twins. These findings suggested that 80% of individual differences in neuroticism are due to heredity.

All subsequent studies have indicated that the findings of H.J. Eysenck and Prell (1951) were very anomalous, especially the extremely high correlation for monozygotic twins. Zuckerman (1987) reviewed the evidence from eight other twin studies. The highest correlation for monozygotic twins in any of those studies was +0.67, and the mean correlation was only +0.52. The largest twin study in which neuroticism was assessed was that of Floderus-Myrhed, Pedersen, and Rasmusson (1980). They studied over 12,000 twin pairs, obtaining a correlation of +0.50 for monozygotic twins and of +0.23 for dizygotic twins.

In principle, especially important evidence about the influence of heredity on personality can be obtained from monozygotic twins brought up apart. The reason for this is that they have essentially the same heredity but are brought up in dissimilar environments. However, most studies (e.g., Shields, 1962) are flawed because many of the monozygotic twins were brought up together during the first few years of life. In addition, other monozygotic twin pairs who were brought up apart were placed in different branches of the same family, and sometimes even attended the same school. We will focus on the study by Pedersen, Plomin, McClearn, and Friberg (1988), which is the nearest approach yet to a definitive study. They assessed neuroticism in 95 monozygotic pairs brought up apart, 150 monozygotic pairs brought up together, 220 pairs of dizygotic twins brought up apart, and 204 dizygotic pairs brought up together. Of the twin pairs brought up apart, 48% were separated before their first birthday, and 82% were separated by the age of five.

Pedersen et al. (1988) found that the correlations for neuroticism were as follows: +0.25 for monozygotic twins brought up apart; +0.41 for monozygotic twins brought up together; +0.28 for dizygotic twins brought up apart; and +0.24 for dizygotic twins brought up together. According to Pedersen et al. (1988), these correlations indicate that 31% of individual differences in neuroticism are due to heredity.

In sum, the available evidence indicates that genetic factors contribute approximately 30% of the variance in trait anxiety or neuroticism, but it is difficult to provide a precise figure. However, it is now certain that the contribution to the variance made by genetic factors is considerably less than the 67% claimed by H.J. Eysenck (1982). At the other extreme, it would appear that Torgersen (1990) under-stated the contribution of heredity: “The development of the relatively normally distributed neuroticism or anxiousness may be modestly influenced by genetic factors ... However, by far the most important source of variance seems to be individual environmental factors’’ (p.285).

Physiological differences

The fact that the contribution of genetic factors to trait anxiety and to neuroticism is less than claimed by H.J. Eysenck (1967) and by Gray (1982) has implications for the relationship between these personality dimensions and physiological functioning. However, as we will see, the actual relationship is much less than would be expected even on the basis that only 30% of the variance in trait anxiety and neuroticism is contributed by genetic factors.

It has not been possible to obtain direct measures of individual differences in the level of activity in the behavioural inhibition system or visceral brain. However, researchers have used a wide range of indirect physiological measures such as the EEG, heart rate, skin conductance, and skin resistance. There have been several large-scale studies, with a very wide range of different physiological measures having been taken in various stressful and non-stressful conditions. All of the available evidence was reviewed by Fahrenberg (1992), who came to the following conclusion: “Over many decades research has failed to substantiate the physiological correlates that are assumed for emotionality and trait anxiety. There is virtually no distinct finding that has been reliably replicated across studies and laboratories” (pp.212–213).

The almost complete failure of psychophysiological studies to find any consistent differences between those high and low in trait anxiety is rather puzzling. Gray (1982) reviewed evidence which indicates strongly that the septo-hippocampal system is centrally involved in anxiety. This evidence comes from two strands of research. First, various anti-anxiety drugs such as the benzodiazepines, barbiturates, and alcohol have been found to exert broadly comparable effects on behaviour. Second, lesions to the septo-hippocampal system in rats and other species produce several behavioural effects. As Gray (1982) pointed out, there are 19 different kinds of behavioural measure for which drug and lesion data are available. On 18 of these measures, drugs and lesions have essentially the same effects. The obvious implication of these findings is that the septo-hippocampal system is of major importance in anxiety.

The reasons why there are so few differences in physiological activity between those high and low in self-report measures of trait anxiety are discussed more fully in Chapter 3. The main reason is that those who score low on self-reported trait anxiety form a heterogeneous group. Weinberger, Schwartz, and Davidson (1979) obtained measures of trait anxiety and of social desirability from their subjects. They argued that high scores on social desirability reflect a defensive or repressive coping style. Those who obtained low scores on social desirability as well as on trait anxiety were categorised as truly low-anxious, whereas those who scored high on social desirability but low on trait anxiety were classified as repressors. When the subjects were exposed to a moderately stressful situation, the repressors responded physiologically much more than the truly low-anxious. In fact, the repressors were more physiologically responsive than the high-anxious subjects on most of the measures.

Similar findings to those of Weinberger et al. (1979) have been obtained in a number of subsequent studies. For example, Brown, Tomarken, Orth, Loosen, Kalin, and Davidson (1996) found that high scorers on trait anxiety did not differ from low scorers in their basal salivary cortisol levels. However, when the low scorers were divided into low-anxious and repressor groups, it was found that high-anxious subjects had significantly higher salivary cortisol levels than the low-anxious subjects. One of the implications of these findings is that there are two rather different types of individuals who obtain low scores on questionnaire measures of trait anxiety. However, the major implication for theories such as those of H.J. Eysenck (1967) and Gray (1982) is as follows: the failure to discover any consistent relationship between trait anxiety and physiological reponsiveness is due mainly to the high level of responsiveness of repressors.

Evaluation

The theoretical approach to trait anxiety and neuroticism advocated by H.J. Eysenck (1967) and by Gray (1982) has only been partially successful. Genetic factors determine individual differences in trait anxiety and neuroticism to some extent, but they are less important than was assumed by H.J. Eysenck (1967) and by Gray (1982). As we have seen, it has proved difficult to demonstrate the theoretically predicted differences in physiological functioning between those high and low in trait anxiety.

In addition to the above problems, the physiological approach is severely limited in many ways. As Eysenck (1992) pointed out, “The various inadequacies of the physiological approach to trait anxiety and neuroticism arise to a large extent because environmental influences and the role of learning are de-emphasised. As soon as one considers learning, then the importance of the cognitive system becomes obvious” (p.42). Evidence suggesting the importance of environmental factors was reviewed by Conley (1984). He averaged the data from several studies, and found that the consistency of trait anxiety or neuroticism from one year to the next (with the intrinsic unreliability of the questionnaire eliminated statistically) was +0.98. This may seem high, but it still implies that fairly large changes in trait anxiety or neuroticism occur over a period of a few years. The level of year-by-year consistency is lower than that of intelligence, which is +0.99 (Conley, 1984).

It is assumed within the physiological approach that individuals with highly responsive physiological systems will be anxious across virtually all stressful situations, whereas low trait-anxious individuals will generally experience rather little anxiety. Eysenck (1992) termed this the “unidimensional view” of trait anxiety, and pointed out that there is evidence against it. Endler (1983) proposed a multi-dimensional approach, according to which the increase in state anxiety produced by a threatening environment will be greater among those high in trait anxiety only when there is congruence between the nature of the threat and the dimension or facet of trait anxiety possessed by the individual. This prediction has been confirmed several times when the dimensions of social evaluation and physical danger have been investigated (e.g., Donat, 1983; Kendall, 1978).

In sum, the physiological approach is over-simplified. This is especially apparent in its relative neglect of environmental factors, changes in personality over time, the multi-dimensional nature of trait anxiety, and individual differences in cognitive functioning. However, there is no doubt that the genetic factors emp...