The idea that fear and anxiety are acquired by a process of learning, most particularly by conditioning, has a long and fruitful history. The idea can be traced back to Pavlov's original discovery of conditioning processes and their relevance to the acquisition of emotional responses. This idea was revised and developed by Watson and Rayner (1920) and Jones (1924), and subsequently elevated to a formal theory by Mowrer (1939) in his classic paper, A stimulus±response theory of anxiety. Some of the key ideas were subjected to experimental analysis and later applied to clinical circumstances by Wolpe (1958) and were incorporated (in part) by Eysenck 1957; Eysenck &
Rachman, 1965; into his general theory of personality and its appli-cation to abnormal psychology. Eysenck's successor at the Institute of Psychiatry, Gray (1971, 1982, 1987) later developed an essentially psychophysiological extension of these ideas and introduced many novel ones. In keeping with the increasing in¯uence of cognitive analyses in psychology in general, the learning theory analysis of anxiety has now been expanded to include important cognitive components. In particular, the writings of Barlow (1988, 2002), Beck and Emery (1985), Beck and D A Clark (1997), D M Clark (1986, 1999), and Salkovskis (1985, 1996a), have strongly in¯uenced the way that anxiety is construed. Historical accounts of these developments are given by Kazdin (1978) and Rachman (1996).
The essence of the learning theory approach is that fears are considered to be acquired by conditioning or other learning processes
and that these acquired fear responses in turn generate escape and or avoidance behaviour. The fear-generated behaviour persists because it is at least partly successfulÐescape or avoidance is typically followed by a signi®cant reduction in fear or anxiety, thereby reinforcing the successful behaviour itself. In one of the original statements of the learning theory of fear, it was believed that any neutral stimulus was potentially capable of being converted through a process of conditioning into a stimulus:
Phobias are regarded as conditioned anxiety (fear) reactions. Any ``neutral'' stimulus, simple or complex, that happens to impact on an individual at about the time that a fear reaction is evoked acquires the ability to evoke fear subsequently. (Wolpe & Rachman, 1960, p. 145)
As will be seen, this original statement of the theory has undergone signi®cant revision.
Theories by Eysenck and Wolpe
Eysenck's theory (Eysenck & Rachman, 1965; Eysenck, 1967), and that of Wolpe (1958), had important implications for the psychology of anxiety. Both drew on the early works of Pavlov and Watson and set their theories of neuroses in a learning theory framework, empha-sizing the role of conditioning processes. Eysenck was most inter-ested in neuroses and the personality factors that predispose people to develop these disorders. He argued that emotionally unstable introverts are at high risk for acquiring conditioned anxiety responses whereas unstable extroverts are at risk for developing conduct dis-orders, personality problems, or hysteria. Wolpe's aim was to explain the genesis of neuroses with a view to developing effective methods of treatment. Wolpe was a clinician±researcher, Eysenck a personality theorist who drew clinical implications from his main theory.
Eysenck's (1957) original work on the anxiety neuroses was based on his two-dimensional model of personality. The two dimensions, which can be plotted at right angles to each other (see Figure 3.1) are emotional instability (neuroticism) and introversion/extroversion:
In our general system introverts are postulated to condi-tion more easily and, therefore, to acquire the condicondi-tioned anxieties and fears characteristic of the dysthymic more
easily than other people, whereas psychopaths and prisoners generally are people who condition poorly and who, therefore, fail to acquire the conditioned responses characterizing the socialization process. (Eysenck &
Rachman, 1965, p. 24)
Conditioned anxiety responses are the result of single traumatic event, or a series of subtraumatic events, involving strong nervous-system reactions. It was assumed that a previously neutral stimulus becomes connected through association with an unconditioned stimulus giving rise to the traumatic emotional reactions.
From now on it will be found that the conditioned stimulus, as well as the unconditioned stimulus, produces the original maladaptive emotional behaviour. This, it seems to us, is the essential learning process which takes place in the development of a neurosis. (Eysenck &
Rachman, 1965, p. 4)
Those conditioned responses that are not reinforced begin to extinguish; this development occurs in conditioned anxiety responses
5 Normal 5
Introversion Extraversion
Psychosomatic disorders Anxiety
disorders
10
5 Neuroticism
Psychopathic patients
Male prisoners
Figure 3.1.
Eysenck's two-dimensional model of neurosis. Adapted from Eysenck and Rachman (1965).
as in all others, giving rise to the clinical phenomenon of ``spontaneous remission''.
Eysenck relied on Mowrer's (1939, 1960) theory to explain the persistence of anxiety and the associated avoidance behaviour. In brief, anxiety reactions, once established, also take on motivating properties. In an attempt to reduce their anxiety, people engage in escape or avoidance behaviour. To the extent that the escape or avoidance is indeed followed by a reduction in anxiety, this beha-viour becomes strengthened and the anxiety reactions are preserved from extinction. The personality theory was supported by a great deal of psychometric and experimental data and the clinical implications were drawn out (Eysenck & Rachman, 1965).
Wolpe's theory (1958) of the neuroses was similarly based on the assumption that the anxiety, which is so characteristic of the neur-oses, arises through a process of conditioning. Wolpe's explanation for the maintenance of the conditioned anxiety and other neurotic reactions similarly relied on Mowrer's ideas of reinforcement through the reduction of anxiety. He then went on to develop laboratory techniques for inhibiting the conditioned anxiety reactions, ulti-mately transferring these to clinical practice (Wolpe, 1958). The best-established of the fear-reducing techniques that he introducedÐ systematic desensitizationÐwas assumed to operate through a process of reciprocal inhibition. The deliberate and repeated super-imposition of an incompatible response on the anxiety reaction steadily gives rise to the development of inhibitory processes that will eventually lead to the elimination of the anxiety pattern. The tech-nique was investigated intensively in the laboratory and widely adopted in clinical practice, laying the foundation for one of the two major streams of what became known as behaviour therapy (Rachman, 1996).
Gray's theory
Gray (1971, 1982, 1987) agreed with several components of Eysenck's theory, including the dimensional approach to the analysis of per-sonality predispositions to anxiety, recognition of the biological contributions to the main dimensions of personality, the role of con-ditioning, and so forth. Gray's elaboration of the theory is primarily psychophysiological, being based on a formidable assembly of psy-chopharmacological data and countless experiments on learning in animals. He suggested a rotation of Eysenck's dimensions of introversion and emotional stability to emphasize the importance
of impulsivity (Gray, 1987, p. 350). Gray claimed that, contrary to Eysenck's suggestion, extroverts have dif®culty in acquiring fear reactions and are not slow at developing conditioned responses in general. According to Gray, neuroticism indicates heightened sensi-tivity to reinforcing events and introversion represents increasing sensitivity to signals of punishment rather than to signals of reward.
Gray collected convincing evidence that the behavioural effects of punishment and of non-reward are similar and probably mediated by common processes. There is also evidence suggesting a similarity between reward and non-punishment. Signals of punishment or of non-reward trigger the behavioural inhibition system, a psychophy-siological system that Gray describes in considerable detail and which has the effect of producing increases in arousal, increased attention, and an inhibition of ongoing behaviour. This behavioural inhibition system can also be triggered by novel stimuli or by stimuli that are innately fear producing. Gray (1986) regards anxiety ``as a central state that mediates behavioural responses to stimuli that signal either punishment or non-reward'' (p. 220).
This analysis led Gray to reconsider the nature of avoidance conditioning and, in particular, to suggest some improvements to the Mowrer theory. This, it will be recalled, states that stimuli that are followed by punishment come to elicit conditioned fear reactions.
Behaviour that produces a reduction in this fear is reinforced. As noted by Seligman and Johnston (1973), Mowrer's theory has dif®-culty in accounting for the persistence of the avoidance behaviour, among other problems. Gray makes extensive use of the concept of
``safety signals'' in his reinterpretation of this persistence. These are signals that arise from the omission of an anticipated punishment.
Safety signals act as rewards and elicit approach behaviour. They also signal a period and a place of safety, a period and place in which the person or animal can expect freedom from punishment: ``Safety signals reduce fear and provide secondary reward for the avoidance response; but at the same time they preserve fear from complete extinction and so ensure their own continued potency'' (Gray, 1987, p. 227). As will be seen presently, this elegant analysis of safety signals, the details of which are not necessary for present purposes, is relevant for understanding various aspects of anxiety, including so-called generalized anxiety. It is worth mentioning that safety signals do not necessarily preserve fear; for example, they can facilitate prolonged and therapeutic exposures to fear cues.
Gray's inclusion of ``innate fear stimuli'' as one of the four triggers of the behavioural inhibition system is consistent with the views of
Seligman, Ohman, and others. According to Seligman, ``. . . the great majority of phobias are about objects of natural importance to the survival of the species . . . The theory does not deny that other phobias are possible, it only claims that they should be less frequent, since they are less prepared'' (Seligman & Hager, 1972, p. 450). He argued that human phobias are largely restricted to objects that have
``threatened survival, potential predators, unfamiliar places, and the dark'' (Seligman & Hager, 1972, p. 465). Like Gray, Seligman argues that certain kinds of fears are readily acquired because of an inherent biological preparedness. According to Seligman, prepared fears are easily acquired, selective, resistant to extinction, and probably non-cognitive. The prepared fears can even be acquired by watered-down representations of the actual threat. This idea of selectivity runs contrary to the earlier view of conditioning in which it was assumed that any neutral stimulus could be turned into a conditioned signal of fear and that no differences should be anticipated between stimuli.
Seligman's introduction of the concept of preparedness, con-structed within the framework of modern learning theory, provoked a good deal of interest, and promising results from the pioneering experiments by Ohman and colleagues (1975) were followed closely.
Working in the laboratory, these researchers demonstrated that people respond differently to presentations of prepared stimuli than to unprepared neutral stimuli. For example, fear responses to photo-graphs of snakes differed from those evoked by photophoto-graphs of
¯owers or mushrooms. Thus they were able to promote Seligman's ideas and provide a test-bed for further investigations. Regrettably, over the next few years there was a gradual accumulation of dis-appointing results and interest in the concept of prepared phobias waned (McNally, 1987).
Evidence that the prepared fears generated in the laboratory are easily wiped out by verbal instructions undermined one of the most appealing characteristics of the concept, because these laboratory fears fall far short of the main features of phobias. The laboratory fears are not easily acquired, or stable, or non-cognitive; they bear little resemblance to the concept of prepared phobias as set out by Seligman.
It has been suggested that these disappointing results are attri-butable to a weak methodology and that more powerful stimuli and more appropriate measures of fear are needed before the theory can be subjected to rigorous testing. The prepared phobias that are the subject of the theory are intense, vivid, resistant, and persisting fears, and it is on such fears that the theory must be assessed, and not on
the fragile and ¯eeting fears that were reproduced in the original laboratory experiments. An excellent research model was provided by Mineka's outstanding experiments on intense and vivid fears in monkeys (Mineka, 1985). She proved that monkeys are indeed pre-disposed to fear snakes (but not ``unprepared'' cues) and that their fears can be triggered by observational learning. The broader
signi-®cance of this research is considered in Chapter 4.
Longitudinal studies, exempli®ed by the Dunedin cohort study (see Chapter 4), are another source of information about learned and prepared fears. The additional information about the genesis of fears provided by experimental analyses and longitudinal research is an essential supplement to the most direct and common method of interviewing and/or questionnaires because a person's recollections might be incorrect, incomplete, or absent. A total reliance on this direct method is unsatisfactory.