Although the range of scores was restricted in our samples of healthy participants, we found that subsyndromal characteristics associated with anxiety disorders were related to fear-conditioning outcomes and reflected moderate effect sizes (Cohen, 1988
). Variables implicated as potentially important included fears of anxiety symptoms as assessed by the ASI, subsyndromal worry as assessed by the PSWQ, and subsyndromal behavioral avoidance as assessed by the FQ. In the univariate regression analyses, only the ASI and PSWQ reached significance as predictors of the general measure of conditionability, and only the FQ reached significance as a predictor of differential conditioning. Of these measures, only the ASI was implicated as significantly predictive of the orienting response as well, raising the possibility that the predictive significance of the ASI for the measure of general conditioning likely reflects its relationship with the orienting response; i.e., a generalized sensitivity and reactivity to any novel stimulus within the context of a conditioning study.
Further analysis with stepwise, multiple regression analyses showed that the FQ and PSWQ offered non-redundant prediction of the magnitude of conditioned responses. Specifically, the FQ and the PSWQ emerged as powerful predictors, together accounting for 23% and 19% of the variance in general and differential conditioning outcomes, respectively. We found that, within the subsyndromal range, subtle avoidance behaviors were associated with lower conditioning to fear cues. In the clinical context, subtle avoidance behaviors also appear to interfere with new learning, but this effect has been demonstrated for interference with safety
learning from exposure (e.g., Powers, Smits, & Telch, 2004
; Telch et al., 2004
) rather than interference with fear acquisition. Accordingly, the potential protective effect of subtle avoidance on the learning of new fears, as observed in our study of non-patients, needs to be balanced against the obvious detrimental effects of avoidance on maintaining anxiety disorders once developed (Barlow, 2002
). Nonetheless, our observation of potential protective effects from subtle avoidance for fear acquisition in this conditioning study is consistent with reports of a trend toward lower levels of PTSD pathology in response to indirect exposure (via media coverage) to the terrorist attacks of September 11th
in behaviorally inhibited (avoidant) children (Otto et al., in press pending
As for the detrimental effects of subsyndromal worry, studies indicate that worry has a number of effects that may help incubate or maintain fears. For example, worry appears to increase the number of intrusive images after the viewing of a stressful film (Butler, Wells, & Dewick, 1995
; Wells & Papagerorgiou, 1995
), helping ensure that negative images, when confronted, are repeatedly rehearsed. Moreover, Jones and Davey (1990)
showed that such cognitive rehearsal of aversive stimuli can enhance fear responding in a conditioning paradigm. More precisely, they found that the cognitive rehearsal of a UCS and the UCR (i.e., to think about the loud tone UCS and reactions to the tone) in the absence of further presentation of the CS and UCS, aided the persistence of the fear CR (see also Davey and Matchett, 1994
). Accordingly, worry may serve as a risk factor for greater fear acquisition after exposure to negative stimuli.
In our laboratory, the next step will be to examine the role of these variables in predicting the higher levels of fear conditioning observed in some studies of patients with anxiety disorders. Of particular focus will be investigations of the inconsistent findings for comparisons of anxiety and healthy-control samples by considering individual differences in levels of worry, avoidance, and anxiety sensitivity in these samples. Based on the present findings, individuals who are higher in worry, but who have lower levels of avoidance, would be expected to show the greatest differences in conditionability relative to healthy control samples.
In addition to identifying potential predictors of fear conditionability among healthy participants, our findings failed to support neuroticism and extraversion as predictors of the degree of fear acquisition; effect sizes for these variables were in the small range. Nonetheless, we found that neuroticism was a significant predictor of the magnitude of the SC orienting response as defined by responsivity to the “to be” CS+ during the habituation phase. Accordingly, neuroticism may discriminate individuals who are reactive to their environment, but perhaps not identify individuals who quickly learn contingencies between cues and aversive events.
Anxiety sensitivity may serve as a similar predictor of reactivity in conditioning studies. In our study, the ASI offered significant prediction for the general conditioning measure only, accounting for an additional 5% of the variance. However, this predictive significance was redundant with the prediction afforded by consideration of individual differences in the orienting response to the CS during habituation, suggesting that the ASI primarily predicted the elements of the general conditioning measure that were contaminated by the orienting response. One interpretation of this finding is that the ASI is predictive of general sensitivity and reactivity to novel stimuli rather than fear learning per se
. Such a confound would not be expected for differential conditioning scores because this measure controls for general reactivity to CSs by examining the difference in reactivity between CS+ and CS-. Indeed, consideration of differences in ASI scores among anxiety samples may have methodological value in helping explain inconsistent findings in previous studies of simple and differential conditioning. If ASI scores in the clinical range are indeed markers of the component of simple fear conditioning that is linked to general reactivity (i.e., orienting response), then samples characterized by patients with high ASI scores—such as patients with panic disorder, posttraumatic stress disorder, or social phobia comorbid with depression (Ball, Otto, Pollack, Uccello, & Rosenbaum, 1995
; Taylor et al., 1992
)--would be expected to show inflated simple-conditioning but not differential-conditioning SCRs. This speculative hypothesis awaits testing in both simple and differential conditioning comparisons between anxiety patients high and low in anxiety sensitivity, as well as comparisons of these patients to healthy controls who are relatively high and low in anxiety sensitivity.
In summary, we identified a number of subsyndromal anxiety-related characteristics that were predictive of fear acquisition in a de novo fear-conditioning paradigm. Although we have speculated that these differences in fear-conditioning may be of importance to the etiology of anxiety disorders, we can not rule out the possibility that our findings simply reflect state correlates of subtle levels of worry, avoidance, or anxiety sensitivity, but have no implications for the development of anxiety disorders. Nonetheless, these state effects of worry may be important for the maintenance of anxiety conditions, changing the ease by which additional fears may be learned once initial levels of worry and anxiety sensitivity are established. Longitudinal studies can help clarify these competing accounts.