I
CONCEPTUAL FOUNDATIONS
1
Theoretical Approaches to the Fear of Anxiety
Richard J. McNally
Harvard University
Pathological fears have often been defined by their eliciting stimuli. Nosologists, for example, have traditionally defined acrophobia as fear of heights, claustrophobia as fear of enclosed spaces, and agoraphobia as fear of open, public places. This tradition is notable for its neglect of the intentionality of fear, Intentionality does not mean deliberateness as in someone doing something intentionally. It refers to the aboutness of something (Brentano, 1889/1984). Rather than specify why an agoraphobic person fears shopping malls, subways, and so forth, traditional approaches to nosology merely identify the range of external cues that evoke excessive fear. Yet specifying the intentional object of fear (i.e., what the fear is about), and not merely its eliciting stimuli, has important nosological implications. For example, people with agoraphobia and people with specific flight phobia both nominally fear the same stimuli, yet the motivation for avoidance is entirely different. The intentional object for agoraphobics is panicking while aloft, whereas the intentional object for flight phobics is crashing (McNally & Louro, 1992).
In their landmark article on the reanalysis of agoraphobia, Goldstein and Chambless (1978) departed from the traditional perspective in important ways. Instead of viewing agoraphobia as a fear of stimuli embodied in public places, they emphasized the intentional object of agoraphobic fearânamely, panic and its presumed consequences. They also revived the notion that oneâs own fear responses could constitute the focus of dread, popularizing the idea that agoraphobia is best conceptualized as a fear of fear itself.
The idea that fear can itself be the intentional object of anxiety had been previously expressed by a diverse range of authors. In his essay entitled Of Fear, Montaigne (1573/1948) said, âThe thing I fear most is fearâ (p. 53). He did not elaborate further, and this one sentence paragraph appears somewhat out of the blue. Nearly 400 years later, during the Great Depression, Roosevelt (1933/1965) famously proclaimed in his first Inaugural Address: âSo first of all let me assert my firm belief that the only thing we have to fear is fear itselfânameless, unreasoning, unjustified terror which paralyzes needed efforts to convert retreat into advanceâ (p. 274).
Commentators on the fear of anxiety have not been confined to French essayists of the 16th century or to American politicians of the 20th century. Psychoanalytic theorists have also discussed the concept. Freudâs (1895/1962) personal experience with panic attacks may have led him to remark that, âin the case of agoraphobia, etc., we often find the recollection of an anxiety attack; and what the patient actually fears is the occurrence of such an attack under the special conditions in which he believes he cannot escape itâ (p. 81, italics in original). Likewise, Otto Fenichel (cited in S. Reiss, 1987) proposed that:
In the first stages, the neurosis may be complicated by a secondary traumatic neurosis, induced by the first anxiety spell which is experienced as a trauma. Many anxiety hysterias develop out of such an experience, a fear of anxiety, and simultaneously a readiness to become frightened very easily, which may create a vicious circle. (p. 586)
Psychoanalyst Rosenberg (1949) conceptualized the capacity to tolerate anxiety as an individual difference variable rather than solely as a feature of agoraphobia. But veering off on a weird theoretical tangent, she then traced the fear of anxiety to its presumptive roots in unconscious aggressive and sexual impulses.
In his comprehensive historical review, Reiss (1987) observed that theorists of nearly every school of clinical psychology have addressed the fear of anxiety. The chief difference among them concerned whether they viewed it mainly as a facet of agoraphobia or whether they believed it relevant to a wide range of neurotic disturbances. Thus, Freud (1895/1962) held the former view, whereas Ellis (1978) expressed the latter view in his theory of discomfort anxiety. Reiss (1987) defended a hybrid perspective, noting that the fear of anxiety has implications for psychopathology in general, but acknowledging its special relevance for understanding agoraphobia.
Concluding his article with a call for more research, Reiss (1987) sketched an agenda for the study of the fear of anxiety. During the next 10 years, investigators have tackled many of the topics he mentioned. Much of this work concerned the relationship between the fear of anxiety and panic disorder (and agoraphobia; see, e.g., chaps. 6 and 10, this volume). However, recent studies have addressed topics unforeseen by Reiss (1987), such as the connection between anxiety sensitivity and essential hypertension (Pagotto, Fallo, Fava, Boscaro, & Sonino, 1992), asthma (Carr, Lehrer, Rausch, & Hochron, 1994), and chronic pain (Asmundson & Norton, 1995; see chap. 12, this volume).
The purpose of this introductory chapter is to provide a comparative analysis of the three main approaches to the fear of anxiety: (a) Pavlovian interoceptive conditioning, (b) catastrophic misinterpretation of bodily sensations, and (c) anxiety sensitivity. Because of its historical importance, panic disorder figures prominently in the discussion.
PAVLOVIAN INTEROCEPTIVE CONDITIONING
Goldstein and Chambless (1978) persuasively argued that agoraphobia is best conceptualized as a fear of fear and not as a fear of public places. They also suggested that such fear might develop through processes akin to Pavlovian interoceptive conditioning:
Having suffered one or more panic attacks, these people become hyperalert to their sensations and interpret feelings of mild to moderate anxiety as signs of oncoming panic attacks and react with such anxiety that the dreaded episode is almost invariably induced. This is analogous to the phenomenon described by Razran (1961) as interoceptive conditioning in which the conditioned stimuli are internal bodily sensations. In the case of fear of anxiety, a clientâs own physiological arousal becomes the conditioned stimuli for the powerful conditioned response of a panic attack. (p. 55)
Accordingly, just as an external cue (e.g., tone) can be established as a conditioned stimulus (CS) if it predicts an aversive unconditioned stimulus (US), so might an internal cue (e.g., dyspnea) be established as a CS if it predicts panic.
Although many theorists agree that interoceptive conditioning may figure in the genesis of panic disorder (e.g., Barlow, 1988; Seligman, 1988; Wolpe & Rowan, 1988), others have been skeptical (e.g., Clark, 1988; McNally, 1990; Reiss, 1988). Critics have noted that, although Razranâs (1961) widely cited article described many Pavlovian interoceptive conditioning preparations (see e.g., McNally, 1990), none involved aversive learning that might provide a plausible model for panic attacks. Moreover, the conceptual clarity of the laboratory preparations is lost once one attempts to map interoceptive conditioning concepts onto the phenomenon of panic. For example, Goldstein and Chambless (1978) suggested that bodily sensations function as CSs that elicit the conditioned response (CR) of panic. Yet because these sensations are partly constitutive of panic itself, it is unclear which ones count as the CS and which ones count as the CR. Does a âdizziness CSâ elicit a âpalpitation CRâ or vice versa? Distinctions become vague if one defines the CS and CR as two points on a continuum of arousal (i.e., CS = a little arousal, CR = a lot of arousal). That is, the phenomenon assigned the conceptual role of CS (e.g., an increase in heart rate) is the same as that assigned the conceptual role of CR (e.g., a further increase in heart rate).
Identifying the US and unconditioned response (UR) also presents problems. Because certain bodily perturbations are classified as CSs, they must, by definition, have been established as such via their association with a US. But what is the US and what UR does it evoke? The individualâs first panic is often viewed as the US. Unfortunately, the bodily sensations associated with panic have already been identified with both the CS and CR. Some theorists maintain that panic is the UR (e.g., Wolpe & Rowan, 1988), but this would imply that an event (i.e., panic as US) causes its own occurrence (i.e., panic as UR). Moreover, the interoceptive conditioning account of panic is based on a contiguity model whereby stimuli present during an attack become established as CSs. But the contiguity model has long been abandoned in favor of information-processing models of Pavlovian conditioning, which describe how organisms learn predictive relations among events (Reiss, 1980).
There was more to the Goldstein and Chambless (1978) theory than just the postulate about Pavlovian interoceptive conditioning (Chambless & Goldstein, 1988). They also emphasized that people with agoraphobia held beliefs about the catastrophic consequences of feared bodily sensations in addition to exhibiting Pavlovian CRs to these sensations. To measure these aspects of the syndrome, Chambless and her colleagues developed the Body Sensations Questionnaire (BSQ) and the Agoraphobic Cognitions Questionnaire (ACQ; Chambless, Caputo, Bright, & Gallagher, 1984).
CATASTROPHIC MISINTERPRETATION OF BODILY SENSATIONS
Purely biological approaches to panic disorder had become increasingly dominant until the publication of Clarkâs (1986) influential article, âA Cognitive Approach to Panic.â In this article, Clark showed how his model could account for findings adduced in support of various biological theories. The core idea was that panic attacks âresult from the catastrophic misinterpretation of certain bodily sensationsâ (p. 461). Accordingly, an individual might misinterpret dizziness as an impending faint, derealization as impending insanity, or palpitations as an impending heart attack. Initiating a positive feedback loop, these misinterpretations worsen anxiety and intensify bodily sensations until fullblown panic results. Although the misinterpreted bodily sensations may arise from anxiety, they may be associated with other emotional (e.g., anger) or nonemotional (e.g., caffeine ingestion) states as well. Regardless of their source, bodily sensations will not lead to panic unless they are misinterpreted as harbingers of imminent danger.
Reviewing Clark and Beckâs theories, Alford and Beck (1997) noted an âidentity of conceptualization, even though different words are usedâ (p. 123). Clarkâs model was foreshadowed by Beck, Emery, and Greenberg (1985), who stated that âher [the person experiencing a panic] interpretation of sudden uncontrollable symptoms as signs of impending physical or mental disaster then accelerates the process until a full-blown panic occursâ (p. 136). Theoretical resemblances notwithstanding, Clarkâs (1986) version of the cognitive approach to panic was remarkably influential. Indeed, Clarkâs article was the second most frequently cited one in the entire field of psychology among the more than 50,000 articles published during the years 1986 through 1990 (Garfield, 1992).
Clarkâs theory stimulated a series of critiques and rejoinders (for a review, see McNally, 1994). A few are discussed next. One issue concerned whether catastrophic interpretations are causally implicated in the genesis of panic or whether they are merely epiphenomenal correlates of an unfolding, autonomous biological process. For example, Wolpe and Rowan (1988) noted that their patients reported experiencing catastrophic thoughts after their panic attacks were well underway.
Another issue concerned whether catastrophic misinterpretations are necessary for panic to occur. Several studies suggest that full-blown panics may occur in the apparent absence of antecedent catastrophizing (e.g., Aronson, Whitaker-Azmitia, & Caraseti, 1989; Rachman, Lopatka, & Levitt, 1988), and as Teasdale (1988) pointed out, just because some attacks are preceded by catastrophic misinterpretation of bodily sensations, one cannot conclude that all panics are so initiated.
The occurrence of panic attacks that erupt during nondreaming sleep pose another challenge to Clarkâs central hypothesis. In addressing this anomaly, Clark (1988) suggested that panic patients might nevertheless monitor, detect, and catastro...