H.J. Eysenck and behaviour
therapy at the Maudsley hospital

H.J. Eysenck is a prolific writer of immense scholarship who has a penchant for espousing some extremely dubious causes. Although there are about a thousand British academic psychologists, his publications alone account for an amazing one-tenth of all the citations they receive in learned journals. No doubt many of these arise from attempted rebuttals of Eysenck’s unorthodox views on such subjects as the medical harmlessness of cigarette smoking, the dependence of one’s vocation on one’s astrological birth sign, and the reliability of statistical data and associated theories produced by the now notorious Sir Cyril Burt (whom Eysenck defended almost, but not quite, to the last). But many other citations must refer to his much more soundly based, if still iconoclastic, work on the practical and theoretical deficiencies of Freudian psychotherapy, and the virtues of ‘modern methods of treatment derived from learning theory’ (the sub-title of his influential book of readings Behaviour Therapy and the Neuroses, 1960).


Eysenck has made numerous empirical investigations and theoretical contributions of his own, but is equally important as the director until recently of the psychological laboratories of London University’s Institute of Psychiatry, associated with the Maudsley and Bethlem Royal Hospitals, and as the founding editor of the journal Behaviour Research and Therapy. A large proportion of British work on behaviour therapy has emanated from this source, and many prominent academics and clinicians, now on both sides of the Atlantic, were under the Maudsley aegis at one time or another (for instance, S. Rachman, at first a prominent student of Wolpe’s from South Africa, V. Meyer, C.M. Francks, H.R. Beech, I.M. Marks, R.J. Hodgson and many others).

Eysenck dedicated Behaviour Therapy and the Neuroses to the memory of J.B. Watson, and more than anyone else has founded his ideas on Pavlovian methods and theories. But it is worth noting that some of the credit for both the iconoclasm and the self-consciously scientific tone which characterize Eysenck’s Maudsley school of thought should go to Henry Maudsley himself, who came to London from Ribblesdale via Cheadle in the 1860s, and by 1913 had established sufficient influence as a consultant psychiatrist and author to persuade the London County Council (with the help of a bequest of £30,000) to build the hospital, with the stipulation that it should always have close associations with London University. Maudsley was a proudly sceptical positivist who violently attacked introspection, emphasizing the limitation of consciousness; vehemently asserted that psychology was essentially a matter of cerebral physiology; and claimed that the study of animal psychology would be essential to mental science (see Sir Aubrey Lewis’s introduction to The Pathology of Mind, 1895/1979). In terms of the causes and cures of insanity his views were interesting and widely read, but are not always relevant here. However, he discussed ‘hereditary disposition’ as a causal factor (something H.J. Eysenck tends to stress, rather more narrowly than Maudsley did). He also believed that the ‘action of circumstances’ was important, and thus spoke of the need for taking patients’ life histories and the need for adapting treatments to particular patients — these also are features of behaviour therapy as practised for instance by Wolpe in the examples given in the last chapter. The slightly behaviourist tinge to Maudsley’s discussion


of upbringing and early environment may be judged from the following quotations:

    “In the mental organism we have really a plastic machinery which if taken in hand sufficiently early, may be manufactured to almost any desired pattern of feeling and belief.” (1979, p. 541.

    “Let anyone from his earliest years live in conditions in which he must, as a daily matter of course, without thinking of it, practise self-denial and self-control, subdue self-regarding impulses, feel, think and act for others, he will unless the original structure of his nature be hopelessly bad, be shaped into a good social unit and rule himself wisely.” (1979, p. 542)

The ‘being shaped’ sounds Skinnerian; and how is the following for a precursor of Pavlov? — ‘So much then concerning the origin and development of language; it is the highest display of reflex function’ (1876, p. 504).

There is therefore a case that Maudsley behaviour therapy is in part a development of English Victorian belief in self- discipline and orderly habits, and not just a matter of importing Watson and Pavlov (and Eysenck) from abroad. (Interested readers should also consult Principles of Mental Physiology with Their Application to the Training and Discipline of the Mind, 1874/1896, by W.B. Carpenter, Registrar of the University of London, and of course Herbert Spencer’s two-volume Principles of Psychology, 1855/1899.)

Nevertheless, nothing could be clearer than the fact that Eysenck himself, and most of his associates, see themselves as supplying ‘the help which the application of learning theory in the hands of a competent psychologist may be able to bring’ (1960, p. ii). Because of this, ‘consistent, properly formulated theory’, and ‘experimental studies specifically designed to test basic theory’ (Eysenck, 1960, p. 11), are given a high priority, along with the development of applications of the theory to applied problems. The method of counter-conditioning introduced by Wolpe has remained a source of inspiration, but has been developed in various ways, notably by in vivo or real-life acting-out of coping with a graded hierarchy of specific problems (e4g. Meyer, 1957). The most fundamental contribution of Eysenck himself to the solution of practical problems has perhaps been his support for the direct treatment of symptoms (shyness, sexual difficulties,


phobias) rather than fanciful underlying causes. Of course this in itself is hardly new. Maudsley recommended late-night warm baths, or a good round of golf, to relieve insomnia, ‘travel with a suitable companion’ or a seaside holiday to lift depression, and the prevention of constipation by thrice-weekly enemas for practically everything. What makes Eysenck’s emphasis on the treatment of symptoms valuable is its obvious quality as an antidote to the Freudian view of symptoms as mere surface indicators of far deeper problems and therefore, in themselves, of little importance.

But Eysenck has often been more concerned with strengthening the theoretical foundations of behaviour therapy than with the development of new applications, and the continuing theoretical self-criticism of the Maudsley group, as well as the objective evaluation of the successes and failures of their practical methods, is worthy of attention. The treatment of psychological disorders by the methods of behaviour therapy is discussed in other chapters; I propose to discuss here the learning- theory analysis of the causes of neuroses rather than their cures.

In a paper written not too long ago, Eysenck (1976) presents a lucid and succinct summary of the vicissitudes of conditioning theories of neurosis and summarizes some new approaches. A central problem for the learning-theory analysis of neurosis is that principles of learning are always assumed to produce adaptive and useful behaviour, but the essence of neurosis is that it is, superficially at least, unwanted and maladaptive. Neurotic problems are at the same time self-perpetuating and self-defeating — this is what Mowrer called ‘the neurotic paradox’. The Freudian or psychoanalytic resolution of this paradox is to assume that there are unseen and disguised purposes being served by neurotic symptoms — these constitute a cry for help, or a defence against irresolvable personal conflicts, or an escape from inner threats more desperate to the individual than problems presented by the symptoms themselves. Watson attempted to sidestep the more imponderable aspects of Freudian theory by appealing to the classical conditioning of emotions produced initially by real-life traumas. The burned child fears the flame, a child who burns his hand on a black stove may come to fear neurotically all things black, and Little Albert, conditioned to fear a white rat, might in later life still feel uneasy about white fur coats.


But Watson’s theory ignores the neurotic paradox even at the nuts and bolts level. Why should a child continue to fear flames if he is only burned once, and then sees flames for years without being burned? Most textbooks say that if the conditioned stimulus is presented often enough without its unconditioned motivator, the conditioned effects will dissipate and ‘extinction’ will take place. This is one part of the paradox — classical conditioning should be adaptive in the sense that conditioned effects should correspond to real experiences, but almost by definition neurotic fears are unrealistic. There are always some cases where having been bitten once by a dog seems to produce fear of all dogs, no matter that dozens of utterly friendly dogs are thereafter encountered, and it is the absence of the dissipation of fear in these cases that is one of Eysenck’s concerns.

Conditioning by worrying

Eysenck’s solution to the neurotic paradox to some extent follows Mowrer’s, and could be said in one sense to have been anticipated by F.D. Roosevelt (1933) — the neurotic has nothing to fear except fear itself. Eysenck does not put it quite like this, of course; he refers instead to the ‘incubation’ or ‘enhancement’ of conditioned responses. This occurs when certain kinds of conditioned stimuli are presented by themselves, without other external sources of arousal or distress, after having first been paired with such an external source, perhaps in a single trauma, but possibly in a series of less notable associations. The theoretical distinction is that only some conditioned stimuli acquire ‘drive properties’. This is not very well worked out, although the single clearest theoretical advance in conditioning theory over the past decade is the realization that some conditioned stimuli are different from others. There are two aspects to this. The first one, emphasized by Seligman (1970), in his ‘preparedness’ hypothesis, is that for any species, there are both natural and unnatural kinds of association between stimuli. The second aspect, which we may call the Franklin Roosevelt effect, is that almost any stimulus, or idea, but especially naturally unpleasant or exciting ones, may acquire their own, self-generating motivational effects. This is a radical departure in learning theory, since it means that internal and possibly subjective processes become relatively independent of the exter-


nal environment, but it is a departure that in fact makes conditioning theories of neurosis considerably more realistic.

There is supporting evidence for this point of view in various animal experiments, in which cats (Wolpe, 1958) or dogs (Solomon et al., 1953; Seligman et al., 1968) receive strong electric shocks in a certain location. When they are put in the same place subsequently, they remain extremely afraid of it even though they receive no more shocks. In some cases they become more and more afraid each time they are returned to it, even though they receive no externally induced unpleasant experience whatever while they are there. You will. see that I have stressed that it is external unpleasantness that is absent in these cases — the most straightforward explanation of the persistence or enhancement of anxiety is that it is the internal fear or arousal, which the animals experience as a consequence of the initial conditioning, that becomes strong enough to serve as an additional conditioning experience each time. Being afraid of shock, if you are afraid enough, may be almost as unpleasant as the shock itself.

Even Eysenck, who has an old-fashioned respect for animal data, notes that physical pain produced by electric shock is not necessarily identical to the unpleasant experiences believed to be involved in human neurosis, and that the theory needs expanding to incorporate various forms of frustration, conflict, disappointment and guilt, along the lines pioneered by Mowrer. There is also little animal data on the question of attractive as opposed to aversive drives, which may be strong enough sometimes to get locked into the same kind of positive feedback effect. Nevertheless, the experiments at the Maudsley by Rachman on human volunteers suggest that it is not entirely implausible to suggest that the intense sexual appeal to some people of inanimate objects such as rubber mackintoshes, handbags or items of female underwear may be partly due to this effect (the fetishist has nothing to be excited about except sexual excitement itself). However, as Gray (1982) (another of Eysenck’s students) has suggested, many aspects of anxiety are basically physiological. Hence the same reactions of the autonomic nervous system, with the same ameliorating effects of appropriate sedatives and tranquillizers such as barbiturates and benzodiazepenes, may indeed be observed in many varieties of human distress as well as in animals’ reaction to fearful stimulation.


Anxious thoughts, anxious words, anxious deeds and anxious feelings

This brings us on to the question of the relations between the purely physiological symptoms that may be associated with anxiety, the subjective mental states of threat and dread that may or may not accompany them, and the effects that either or both of these may have on what we actually do or say. Clearly we are now a long way from simple analyses of conditioning, but a serious and significant attempt has been made by behaviour therapists to come to grips with this theoretically difficult but practically important set of problems in the ‘three-systems model’, due originally to Lang and taken up by Rachman (1978; for a comprehensive review see Hugdahl, 1981).

Lang’s constructive criticism of previous conditioning theories of fear and anxiety is that they consider anxiety to be a single ‘lump’ of an emotion, whereas there is a great deal of evidence to suggest that at least three measurable aspects of anxiety can vary quite independently. His alternative is to divide anxiety, as Caesar divided Gaul, into three parts, which can be dealt with separately by the therapist. The three systems are behavioural, ‘cognitive/ verbal’ and physiological, or action, awareness and autonomic nervous system reactions. A possible criticism of this is that the cognitive/verbal system is still treated as a ‘lump’ whereas it too ought to be divided into various kinds of relatively rational self-labelling, attribution and verbal formulations of a person’s problems, together with a variety of subjective feelings of panic and distress and awarenesses by the individual of his or her own actions and bodily symptoms. However, for practical and applied purposes, the three-way division works well, since separate measurements can be made of a patient’s activities (either directly or by report), verbal feelings (by interviews, self-report, and systematic questionnaires, sometimes referred to as ‘fear thermometers’) and autonomic reactivity (by various physiological techniques, including electrical recordings for heart rate and sweatiness).

At the extremes, three kinds of patient can be identified.

1.    The purely physiological responder this patient does not think or say that he or she is anxious, and reports no problems of living such as being afraid to fly in aeroplanes, feeling nervous about


going into work, interacting with family members, or such like. In that sense the patient is not necessarily a psychiatric patient at all, but is frequently encountered by GPs as a patient complaining of faintness, dizzy spells, bowel problems or insomnia, or any number of other symptoms, identified by the person as merely physical and not psychological, but classifiable by others as having psychological causes. Almost everyone probably qualifies here at some time or another — one can have a pounding heart, or butterflies in the stomach, without necessarily feeling or being neurotic, but it is not a negligible problem, since vast quantities of various sorts of tranquillizing drugs are dispensed to deal with physical symptoms, and chronic cardiovascular problems such as high blood pressure or chronic digestive disorders may result from unidentified anxiousness of this kind.

2.    The mainly behavioural responder there may not be many people like this, but it is theoretically possible, at least, for avoiding behaviours to become fixed with little evidence of arousal in physiological systems and relatively little self-knowledge. It is also possible in theory that the behavioural responder is such as a result of neurotic (or real) fears which are simply avoided. A person might sincerely believe that they are not afraid of flying; they merely strongly prefer trains, and if enough trains are available any problem they have with aeroplanes is solved. Many people have quite irrational aversions to certain foods or to certain species of animal, but as long as they can eat other things and stay out of the way of cats (or pigeons, or spiders), then they may not show any sign of a tendency towards cold sweats or numbness of the limbs, and may describe themselves in an answer to a questionnaire or in conversation as fear-free. Arguably, the other components of anxiety here are merely disguised. In a serious case someone might claim to be perfectly relaxed and happy, while never leaving their house. This is agoraphobia — fear of open places or crowds —which is not uncommon and has been treated in various ways by behaviour therapists. It may be only when a new factor intrudes — a spouse dies, or children leave home, and going out to work becomes essential — that the necessity of confronting feared circumstances leads to panic attacks and the seeking of help.

3.    The exclusively verbal/cognitive responder: the extreme case of this might be termed the Woody Allen syndrome, where someone is able to lead an entirely agreeable and productive life but wishes


to spend vast sums of money in order to talk for hours to a therapist about how neurotic they are. This extreme may be fictitious, but Lang first proposed the three-systems model after finding that subjects who had become accustomed by desensitization procedures to handle snakes, still described themselves as very afraid of snakes. Of course, we all have to do things that we don’t like, or even are afraid of doing, and it doesn’t mean because we do things, that we like doing them. But in many cases people’s beliefs about themselves constitute their problem, and the verbally expressed beliefs are an appropriate target for modification. ‘I’m shy’, ‘I’m unattractive’, ‘I’m hopeless at figures’, ‘I know I’ll fail the exam’, ‘I’m sexually incompetent’, ‘I panic whenever I see a spider’, ‘I’m always anxious’, ‘I’m hopelessly neurotic’: if held strongly enough, all these beliefs can be damaging, and of course in practical terms it is not the exclusively cognitive responders who are at risk (those students who say every year ‘I know I’ll fail, I know I’ll fail’, and then breeze in on the day and get Firsts). Rather, those people are at risk in whom the cognitive component is by no means independent of the other two but is actively perpetuating, or even initiating, unpleasant worry-induced insomnia and physiological over-arousal, and neurotic avoiding activities (Mahoney, 1974).

Variations in the causation of neurosis

It is reasonable to suppose that each of the three systems might have some special effect in the causation of neurosis, even though they are unlikely ever to be completely independent. Rachman (1977, 1981) uses the three-systems model to support his very sensible suggestion that there is more than one way of acquiring a neurosis, and that traumatic conditioning experiences are neither necessary nor sufficient as a cause.

Clearly, autonomic reactions can be sufficiently unpleasant in themselves to be bound up in neurosis, whether or not they started from external precipitating factors. Many people with phobic or even more general anxieties can report life events which they believe have an obvious relation to their fears. If a salesman has the misfortune to knock over and kill his own son when driving home, as a consequence of his car’s brake failure, the trauma itself would surely be sufficient to produce feelings of guilt and anxiety associated with driving for some time afterwards. Anyone who has


been in a car accident, or had their flat burgled, or been brutally raped, is likely to suffer after-effects connected to reminders of the horrible event. But not everyone who endures such nightmarish experiences suffers the same sort of after-effects and this is one reason for being sceptical about the pure conditioning theory. Rachman is particularly interested in the fact that air raids during the Second World War failed to produce widespread outbreaks of neurosis. There was no increase in the number of patients attending psychiatric clinics following air raids in this country. In Liverpool, Sir Aubrey Lewis noted that although eighteen volunteers were trained as auxiliary mental-health workers to comfort those made anxious both during and after raids, there was, it turned out, no comforting necessary. This is probably something of a special case, since a shared threat from a known enemy has some very peculiar consequences for both individual and group emotions (not all of them desirable, I think, as the Falklands episode shows). But it directed Rachman’s attention to social factors in neurosis, and rather belatedly behaviour therapists are adding these factors into their theories.

That the disasters of war are not without their after-effects is amply demonstrated by the studies of shell shock in the First World War and of combat fatigue in air crew in the Second World War. People in Northern Ireland are not untouched by the troubles, and children especially show symptoms of anxiety related to physical threats and social tensions (children, but only a few, also developed anxiety symptoms attributable to air raids in the Second World War). Nevertheless, Rachman is right to point out that the incidence of neurosis cannot be predicted from the incidence of externally produced traumas, and this is incompatible with a crude Watsonian conditioning theory.

Equally, vast numbers of people are afraid of snakes and spiders, even though they have had no obviously traumatic experience associated with these creatures, and many patients with crippling neurosis can report no precipitating event. The line now taken by Rachman (19fl, 1978) and Eysenck (1976) is strongly influenced by Seligman’s rather vague idea of preparedness. Children are commonly afraid of the dark, but not of their pyjamas, and commonly afraid of live animals, especially slimy or wriggly ones, but not afraid of electric plugs, sharp knives and scissors, or bottles of pills, which are in fact much more dangerous


and have often been deliberately associated with active parental disapproval (this may be over-rated as an aversive stimulus for the human young)4 Sociobiologists might claim therefore that we have an inherited tendency to fear the dark and to fear snake-like and spider-like animals, perhaps in common with other primates, although for obvious reasons we have no genetically determined propensity to become afraid of scissors or plastic bottles of pills. Behaviour therapists are not usually so explicit, but a vaguer version of preparedness almost certainly contains at least a grain of truth, and is extremely convenient for explaining variability in conditioning effects. For instance, there are two reports of attempts to replicate Watson’s experiment with Little Albert (see pp. 86—8), both of which failed4 Eysenck (1976) is able to point to the fact that in these attempts iron bars were banged while infants were confronted with inanimate objects such as curtains or a wooden duck4 These stimuli would not have the preparedness value of a live rat, which would be sufficient to account for the failure of the infants to become afraid of curtains or of wooden animals4 Simply in attention-getting value alone, a live animal which moves, wriggles its nose and occasionally stares one in the eyes is certainly likely to be a more effective item in any sort of learning or information-gathering process than the fact that curtains happen to be present4 Within learning theory, this can all be encompassed by appealing to selective attention4

Vicarious learning and the social transmission of information

Rachman (1977) wishes to augment the idea of conditioning by experienced external events with learning via ‘vicarious and informational transmission of fears’. This seems extremely sensible, although there is relatively little experimental evidence that can be quoted in support of it4 In relation to the three- systems idea, Rachman speculates that external conditioning will act most directly on behaviour and on physiological response, while in fears learned by imitation or ‘informationally’ (e.g4 by a child watching unsuitably violent films on television) the subjective/verbal system will predominate. Since he quotes no evidence at all in favour of this speculation we should not feel bound by it. The important


thing is that children especially, as Mary Cover Jones suggested, but possibly adults just as much, and adolescents even more so, are sensitive to knowledge derived from their peers, and this is only one source of cognitively based information. If anything, children have a tendency to seek out excitingly gruesome stories of monsters, and ghosts, and bogeymen, but excessive exposure to such things might conceivably lay the seeds in young minds for later neuroses, to say nothing of causing nightmares immediately afterwards.

The boundary between the rational and the neurotic in responding to the numerous sources of more mundane knowledge that we have available to us as adults is often blurred. Is it neurotic to lose sleep over our own job if a colleague is sacked (or if we receive dire warnings in the post from the Association of University Teachers), or to worry about having breast cancer or genital herpes after repeatedly coming across magazine articles on these topics? Surely not. And how much insomnia, or distraction from normal daytime activities, is appropriately caused by contemplation of the horrors of nuclear war? It is certain that such questions are unanswerable in terms of Rachman’s theories, but almost as certain that they are unanswerable in any terms. The virtue of Rachman’s critical examination of conditioning theory is that it acknowledges overheard old wives’ tales, second-hand traumas, social norms and the cultural transmission of information as possible modulators, or even as determinants, of individual emotional experience.


Eysenck’s own learning-theory model of neurosis (1976, 1979) is considerably less liberated from the crude Watsonian assertion of conditioning than is Rachman’s work, and differences of opinion concerning other developments, especially cognitive behaviour modification (Meichenbaum, 1977: see p. 153), abound. However, the emphasis on the special factors involved when anxiety-provoking ideas acquire their motivational force (see p. 114), and the willingness to consider inborn and natural patterns of fear as well as individual differences in personality, together with the separate treatment of subjective, physiological and behavioural


factors in the three-systems model (see p. 116), all show that the Eysenckian school of thought is capable of change and development.


End of Chaper 9 | Contents | Start of Chaper 10