شرطی سازی ترس وابسته به انگیزش و تحریک درونی و اختلال پانیک : نقش پیش بینی انگیزشی شرطی محرک غیرشرطی

Interoceptive fear conditioning is at the core of contemporary behavioral accounts of panic disorder. Yet, to date only one study has attempted to evaluate interoceptive fear conditioning in humans (see Acheson, Forsyth, Prenoveau, & Bouton, 2007). That study used brief (physiologically inert) and longer-duration (panicogenic) inhalations of 20% CO2-enriched air as an interoceptive conditioned (CS) and unconditioned (US) stimulus and evaluated fear learning in three conditions: CS only, CS–US paired, and CS–US unpaired. Results showed fear conditioning in the paired condition, and fearful responding and resistance to extinction in an unpaired condition. The authors speculated that such effects may be due to difficulty discriminating between the CS and the US. The aims of the present study are to (a) replicate and expand this line of work using an improved methodology, and (b) clarify the role of CS–US discrimination difficulties in either potentiating or depotentiating fear learning. Healthy participants (N = 104) were randomly assigned to one of four conditions: (a) CS only, (b) contingent CS–US pairings, (c) unpaired CS and US presentations, or (d) an unpaired “discrimination” contingency, which included an exteroceptive discrimination cue concurrently with CS onset. Electrodermal and self-report ratings served as indices of conditioned responding. Consistent with expectation, the paired contingency and unpaired contingencies yielded elevated fearful responding to the CS alone. Moreover, adding a discrimination cue to the unpaired contingency effectively attenuated fearful responding. Overall, findings are consistent with modern learning theory accounts of panic and highlight the role of interoceptive conditioning and unpredictability in the etiology of panic disorder.

Fear of benign bodily sensations, including cues or contexts that might occasion them, is a characteristic of panic disorder. How such fears are learned or acquired has been of focal interest within contemporary learning models of panicogenesis (Acheson et al., 2007, Bouton et al., 2001, Forsyth and Eifert, 1996 and Mineka and Zinbarg, 2006), with most accounts casting the etiology of panic pathology in classical conditioning terms. Such learning accounts draw heavily on interoceptive conditioning. This term, first introduced to the Western world by Razran (1961), refers to “classical conditioning in which either the conditioned stimulus [CS] or the unconditioned stimulus [US] or both are delivered directly to the mucosa of a viscus” (p. 81). These notions have not only figured prominently in etiological accounts of panic disorder, but also in recent modifications of exposure-based interventions that now increasingly include both interoceptive and exteroceptive exposure exercises (Barlow, Allen, & Choate, 2004). Drug tolerance research comprises the largest body of work on interoceptive conditioning, and inferences from this line of work have been used to support the role of interoceptive learning in panicogenesis. According to this view, a stimulus repeatedly paired with drug administration becomes a CS producing a conditioned response (CR) opposite the drug's effect. This CR, in turn, overrides the pharmacologic effect of the drug to create tolerance (Siegel, 1989). For instance, in one of the first studies of its kind, Greely, Le, Poulos, and Cappell (1984) evaluated tolerance in rats using interoceptive CSs in the form of low and high doses of alcohol. In one condition, the low dose preceded the high dose, whereas in a second condition the doses were never paired. Tolerance was lost in the absence of the low dose in rats trained with the low–high combination. In contrast, in rats trained with the high dose alone, tolerance was lost when the low dose was added. Comparable results showing that associations can be formed between the early and late components of the same event have been reported with morphine tolerance (Cepeda-Benito & Short, 1997). For instance, rats given long-duration exposures to morphine exhibit a CR to a short “probe” injection designed to mimic the onset properties of the longer injection (Kim, Siegel, & Patenall, 1999; see also McDonald and Siegel, 2004 and Sokolowska et al., 2002), an effect termed “intra-administration association” (Kim et al., 1999, Siegel et al., 2000 and Sokolowska et al., 2002). Thus, interoceptive cues linked with the onset of an event can be associated with later aspects of the event. Collectively, this work is important in showing that an intero-interoceptive relation (Razran, 1961) forms with each drug administration such that animals learn to respond to an early event in anticipation of a later event. In an analogous fashion, early physiological changes during a panic attack may become signals for more intense and aversive physiological arousal (e.g., a panic attack, or intense fear; Craske, 1991) and thus elicit a panic attack (CR) on their own (Barlow, 2002). For example, a slight rise in heart rate accompanying the beginning stages of a panic attack may become a conditioned stimulus (CS) signaling a larger rise in heart rate characteristic of the later stages of a panicogenic response including other associated sensations (e.g., tachycardia, heart pounding, chest tightness, breathlessness). Such learned relations then alter the function of formerly benign bodily events such that they become significant fear-evoking events in their own right. Under the right conditions and in the context of relevant vulnerabilities (Mineka & Zinbarg, 1996) such learning may contribute to the development of hypervigilance, anxious apprehension, avoidance, and even panic disorder (Barlow, 2002, Bouton et al., 2001 and Forsyth and Eifert, 1996). Though the literature on interoceptive conditioning in a substance abuse context is vast, the same cannot be said for anxiety pathology. One line of research has demonstrated interoceptive conditioning in human subjects employing an olfactory CS and panicogenic US (Devriese et al., 2000 and van den Bergh et al., 1995). Here, though, the CS possessed different stimulus properties than the US, and thus the CS could be distinguished from the US. This is important, in part, because relevant preparations for an interoceptive learning model of panic would seem to require that the CS and US share similar stimulus properties that, in turn, make it difficult to predict when similar physiological events signal subsequent increased arousal and panic or not. As it stands, the best evidence of this sort of process comes from basic research on drug tolerance with nonhuman animals (Kim et al., 1999, McDonald and Siegel, 2004 and Sokolowska et al., 2002). In response to such issues, Acheson and colleagues (2007) sought to provide a more direct test of the interoceptive conditioning model of panic in a nonclinical human sample using 20% CO2-enriched air as an interoceptive CS (i.e., physiologically inert 5-s exposures) and US (i.e., physiologically potent 15-s exposures). The study design consisted of three separate conditions: (a) a “CS-only” condition involving only inert, 5-s CO2 exposures; (b) a “paired” condition that included pairing the CS and US in a contingency (i.e., amounting to a 20-s CS–US complex); and (c) an “unpaired” condition, wherein participants received the same amount of CS and US exposures as the paired condition, except that the CS and US occurred randomly and were never paired. The study was divided into three separate phases: a habituation phase consisting of one CS exposure, an acquisition phase consisting of six CS (all conditions) and five US exposures (for participants in the paired or unpaired conditions), and an extinction phase consisting of six CS-only exposures. Electrodermal response magnitude and individual ratings of distress and fear were employed as dependent measures. Overall, findings of this study support interoceptive fear conditioning, as evidenced by greater-magnitude electrodermal responding as well as distress and fear to the CS in the paired relative to the CS-only condition (Acheson et al., 2007). The results for the unpaired condition, however, were somewhat surprising. Participants who received the unpaired contingency showed greater-magnitude electrodermal response and reported significantly more fear and distress during acquisition relative to the CS-only condition. Moreover, electrodermal differences between the CS-only and paired conditions rapidly attenuated early in extinction. However, participants in the unpaired condition continued to respond with significantly greater fear, distress, and electrodermal activity compared with both the paired and CS-only conditions. These findings indicate that the unpaired condition yielded (a) rapidly acquired and sustained high levels of fear responding, and (b) was experienced as more aversive relative to both the paired and CS-only conditions. Moreover, fearful responding to the CS in the unpaired condition appeared resistant to extinction. Though such findings may seem anomalous, there are at least two possible explanations. First, the responding observed in the unpaired condition may have been a consequence of sensitization. This phenomenon occurs when the presentation of an intense stimulus evokes stronger responses to other similar, but less intense, stimuli (Charles & Winokur, 1963). According to this view, the effects seen in the unpaired condition could have been the result of sensitization to the US, which in turn amplified responses to the CS alone. The effect may persist because the US occurs with some frequency, thereby not allowing attenuated responding to the CS. Nonetheless, if sensitization was active in the conditioning procedure, then it should have operated equally in both the unpaired and paired conditions. That is, the same pattern of response should have been seen in both conditions. Further, sensitization does not clearly explain why the unpaired condition would yield greater-magnitude responding compared to the paired condition. The unpaired condition was exposed to a maximal 15-s dose of CO2-enriched air on some acquisition trials, whereas the paired condition consistently received a CS–US complex amounting to a 20-s dose on each trial ( Acheson et al., 2007). Thus, one would expect the paired condition to show the greatest magnitude of fearful responding overall. This was not the case. In fact, by the end of the acquisition phase, responding had reached asymptote in the paired condition, whereas responding continued to rise in the unpaired condition. Thus, the sensitization account appears untenable, and suggests that other processes were at work. Second, one could conceptualize the unpaired contingency as a “partial reinforcement” procedure in which the 5-s cue was presented alone for half of the trials and paired with a 10-s US for the other half (i.e., the first 5 s of each US exposure could be conceptualized as a CS). That is, only half of the CS presentations were reinforced by subsequent presentation of a partial US. This partially reinforced CS and US contingency leaves the individual unsure whether the CS may escalate into a full US. In short, the trial outcomes are always surprising (Pearce & Hall, 1980), and the individual can never be certain whether a given CS will be followed by a US. Thus, the individual learns that the best strategy is to respond to every CS as if it is a US. Partial reinforcement has the potential to account for both the high magnitude of fear responses during acquisition and extinction in the unpaired condition. Indeed, the elevated pattern of responding in the unpaired condition is consistent with the well-known “partial-reinforcement effect”: CRs to CSs that are intermittently paired with the US are slower to extinguish than CRs to CSs that have been paired with the US 100% of the time (e.g., Bouton, 2004 and Mackintosh, 1974). Thus, it is possible that such a “partial-reinforcement” procedure may more closely approximate the kinds of contingencies that establish “fear of fear” in persons suffering from panic disorder as it occurs in the natural environment. For example, an individual may experience many benign physiological fluctuations over the course of a given day. However, after experiencing one or more panic attacks, the individual is no longer sure whether these fluctuations will wax into a panic attack or not. Thus, the individual may find him- or herself reacting to every indication of physiological fluctuation with fear. In addition, this contingency arrangement may help explain why fear of panic attacks and their consequences does not readily extinguish despite opportunities for corrective learning under naturalistic exposure conditions. Put into different terms, the unpaired relations between topographically similar physiological events presents participants with a discrimination problem. At the onset of each CS or “US” trial, participants have no way of discriminating whether that stimulus will be just a CS or escalate into the US. They have a history of both outcomes, but the initial properties of the CS and US are identical, rendering a judgment about potential aversive consequences difficult. This discrimination problem also makes the early onset of the CS and US unpredictable in terms of their eventual consequences. In short, such issues may require a rethinking of the kinds of conditioning arrangements that may be relevant in the learning account of panic. At present, the data suggest that “unpaired” or partial-reinforcement contingencies are more aversive and difficult to extinguish than predictable paired contingencies. Discrimination failure may contribute to the genesis and maintenance of panic and “fear of fear” (Goldstein & Chambless, 1978) in the natural environment. The aims of the current study are twofold. First, the study attempted to replicate and improve upon the methodology used by Acheson and colleagues (2007). Along these lines, the three conditions of the original study were retained. However, the methodology was altered by adding more trials in each conditioning phase of the experiment, as well as more frequent self-report assessment. These modifications allowed for a more detailed and thorough evaluation of how indices of conditioned emotional responding develop across conditions and are maintained over time. The second and more focal aim of the study was to clarify the role of CS–US discrimination in creating the “unpaired” effect. This was accomplished by adding a fourth “discrimination condition.” This condition was identical to the unpaired condition, except that it included an exteroceptive stimulus presented concurrently with CS onset. This cue was intended to serve as a signal to participants that the larger US will not be presented. If the unpaired effect is due to a discrimination problem, then fearful responding to the CS ought to attenuate once participants have learned the CS–US discrimination. We anticipated that the discrimination condition would yield a pattern of attenuated responding that would be indistinguishable from the CS-only condition.