Copyright 1991 by the American Psychological Association, Inc. 0022-3514/91/S3.00

Journal of Personality and Social Psychology 1991, Vol. 61, No. 3, 504-510

Cognitive Mediators of Pain Perception and Tolerance Sharon L. Baker and Irving Kirsch

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

University of Connecticut

A social learning model of pain perception and tolerance was evaluated. Responses to cold-pressor pain were examined in 4 groups of 20 Ss each. One group was given instruction in cognitive coping strategies, the 2nd was given a monetary incentive contingent on tolerance time, the 3rd was given both cognitive training and monetary reinforcement, and a no-treatment control group was given neither. Incentive and coping instructions significantly increased tolerance but did not alter pain perception. A path analysis indicated that pain tolerance was affected by self-efficacy, which in turn was predicted by pain expectancy and the provision of incentives. Incentive also had an impact on tolerance that was independent of self-efficacy. Pain perception was strongly affected by pain expectancy, but was unrelated to self-efficacy.

1985b, 1986). Outcome expectancy generally refers to people's expectations about the outcomes of their own behavior (e.g., Ajzen & Fishbein, 1980; Kirsch, 1985a; Rotter, 1954) and is hypothesized to be partially determined by self-efficacy (Bandura, 1983; Kirsch, 1985b). In the context of self-efficacy theory, however, the term outcome expectancy is sometimes used in a different manner. Instead of being used to denote the outcome that the person actually expects, it is used to refer to a belief about environmental contingencies, specifically, to the belief that a behavior will lead to a specified outcome if it is performed at a particular level of competence (Bandura, 1977; Kirsch, 1985b; Maddux, Norton, & Stoltenberg, 1986). When defined in this way, outcome expectancies are conceptually orthogonal to and may be empirically uncorrelated with self-efficacy (Maddux et al., 1986). However, this manner of conceptualizing and measuring outcome expectancies is the exception rather than the rule. Another recent addition to social learning theory is the distinction between stimulus expectancies and response expectancies (Kirsch, 1985a, 1990). Stimulus expectancies and response expectancies are two different types of outcome expectancy. Stimulus expectancies are beliefs about the occurrence of external events (e.g., money, grades, and social approval), and response expectancies are beliefs about one's own nonvolitional reactions to events (e.g., pain, fear, and joy). Because it is a subtype of outcome expectancy, response expectancy has motivational properties. People behave so as to maximize the occurrence of positively valued nonvolitional responses and to minimize the occurrence of negatively valued responses. In addition, response expectancies tend to be self-confirming, as is amply documented by the vast literature on placebo effects (reviewed in Kirsch, 1990). In relation to this study, pain expectancy is a particularly salient response expectancy. Presumably, people given training in cognitive pain control strategies would expect to experience less pain as a function of that training. As a result, they should anticipate tolerating the painful stimulus for a longer time. In addition, because response expectancies are self-confirming, the expectation of reduced pain ought to produce its occur-

Psychological manipulations have been shown to produce significant changes in both self-reported pain intensity and in pain tolerance (Avia & Kanfer, 1980; Bandura, O'Leary, Taylor, Gauthier, & Gossard, 1987; Berntzen, 1987; Chaves & Barber, 1974; Devine & Spanos, 1990; Dolce, Doleys, Raczynski, Lossie, Poole, & Smith, 1986; Grimm & Kanfer, 1976; McGlashan, Evans, & Orne, 1969; Rosenbaum, 1980; Scott & Barber, 1977; Stam & Spanos, 1987; Stevens & Heide, 1977; Thorn & Williams, 1989; Worthington & Shumate, 1981). The purpose of this study was to examine hypothesized cognitive mediators of the effects of pain reduction and tolerance enhancement strategies. The theoretical framework from which this study was generated was that of social learning theory (Rotter, 1954), in which behavior is predicted as a function of (a) the expectancy that it will lead to particular outcomes and (b) the value of those outcomes to the individual. Bandura's (1977) distinction between self-efficacy expectancies and outcome expectancies is a relatively recent addition to social learning theory. Outcome expectancies are beliefs that a behavior will produce particular outcomes, and self-efficacy expectancies are beliefs that one is capable of successfully performing the behavior on which the outcomes are thought to be dependent. A relation between selfefficacy and pain tolerance has been demonstrated in a number of studies (Bandura et al, 1987; Litt, 1988; Vallis & Bucher, 1986). In these studies, individual differences in pain tolerance were found to be positively correlated with people's predictions about the length of time that they will endure a painful stimulus. The conceptual distinctions and empirical relations between self-efficacy and outcome expectancies have been a source of some confusion (see Eastman & Marzillier, 1984; Kirsch,

This article is based on a doctoral dissertation conducted by Sharon L. Baker under the supervision of Irving Kirsch. Correspondence concerning this article should be addressed to Irving Kirsch, Department of Psychology, University of Connecticut, U-20, Room 107, 406 Babbidge Road, Storrs, Connecticut 062691020. 504

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.


rence. To the extent that the stimulus is perceived as less painful, it ought to be tolerated longer. Note, however, that response expectancy is not hypothesized to be the only mediator of treatment-induced pain reduction, as has mistakenly been claimed (Devine & Spanos, 1990). According to response expectancy theory (Kirsch, 1985a), response expectancy is only one determinant of nonvolitional responses. Pain reduction treatments can have effects that are independent of expectancy, as well as effects that are mediated by expectancy. The hypothesized relation between pain and the expectancy of its occurrence has been supported in a number of studies. Placebos reduce self-reported pain intensity and enhance pain tolerance (Baker & Kirsch, 1987; Camatte, Gerolami, & Sarles, 1969; Gelfand, Ullmann, & Krasner, 1963), pain can be increased or decreased by verbal instructions about the effect that a procedure is expected to have (Daniels & Sallie, 1981; Dworkin, Chen, LeResche, & Clark, 1983; Dworkin, Chen, Schubert, & Clark, 1984; Subotnik & Shapiro, 1984), and significant correlations have been reported between expected and experienced pain (Baker & Kirsch, 1987; Devine & Spanos, 1990; Erfanian & Council, 1987).

The Relation Between Self-Efficacy and Response Expectancy In situations that are painful or frightening, self-efficacy and response expectancy are very highly correlated. The more pain people expect, the lower their self-efficacy for pain-related movement (Council, Ahern, Folick, & Kline, 1988), and the more fear people expect, the less their self-efficacy for approaching a feared stimulus (Kirsch, Tennen, Wickless, Saccone, & Cody, 1983; Williams, Dooseman, & Kleifield, 1984; Williams, Turner, & Peer, 1985). Correlations between self-efficacy and response expectancy reported in these studies indicate that the shared variance may be as high as 80%. What is the reason for this substantial association between response expectancy and self-efficacy? The key to this relation involves differences in the types of situations in which self-efficacy is measured. Self-efficacy for nonaversive achievement tasks is not greatly affected by incentives. In those situations, self-efficacy perceptions are determinants of outcome expectancies (Bandura, 1983). For example, the expectancy of winning a table tennis game (an outcome expectancy) is largely dependent on the expectancy that one is able to win the game (a self-efficacy expectancy). When outcomes are largely dependent on skill, the conceptual and operational distinctions between outcome and efficacy expectancies are often trivial, as they are in the example just cited. A large number of studies have been identified in which outcome expectancies were operationalized in such a manner as to be virtually indistinguishable from self-efficacy (Kirsch, 1986). In aversive situations in which skill is not perceived as an important or salient component of the task, the causal relation between self-efficacy and outcome expectancy is precisely the reverse. Rather than being a cause of outcome expectancies, self-efficacy in these situations is a product of outcome expectancies. Thus, incentives have a substantial impact on self-efficacy evaluations for such tasks as refraining from smoking (Cor-


coran & Rutledge, 1989) and approaching a feared stimulus (Kirsch, 1982; Schoenberger, Kirsch, & Rosengard, in press). Why are efficacy expectations for these tasks so easily influenced by real or hypothetical incentives? Being good judges of their capabilities, people know that they are able to perform tasks that do not exceed their abilities. However, they are also aware of their strong disinclination to perform tasks that entail very aversive outcomes. In these situations, it is a common linguistic habit to say that one cannot perform the task, even though when pressed, people readily acknowledge that they can perform it if necessary (Kirsch, 1982). When asked whether they are able to perform a highly aversive task of which they know they are capable, people respond by indicating whether they will perform it. Phobic subjects report their willingness to approach a feared stimulus (Kirsch, 1982; Schoenberger et al., in press), and smokers report their willingness to tolerate withdrawal symptoms (Corcoran & Rutledge, 1989). Similarly, we hypothesize that subjects completing a self-efficacy questionnaire in an experimental pain study are indicating how long they are willing to tolerate the pain stimulus. In reasoned action theory, a person's subjective probability of performing a behavior that is under volitional control is termed an intention (Ajzen & Fishbein, 1980). Therefore, when a behavior is deemed to be within one's capacity, the prediction of its occurrence is a statement of intention. Self-efficacy is measured as a person's prediction about the performance of a behavior. Therefore, when the behavior is perceived as controllable, selfefficacy measures intention rather than perceived ability (Kirsch, 1985b). But how can one determine whether a behavior is perceived as controllable? Incentives do not greatly affect behavioral predictions for responses that are believed to exceed one's capacities (Kirsch, 1982), and they could not affect behavior that in fact is beyond the person's control. Conversely, incentives and other expected outcomes are the primary determinants of intentions and of intentional behavior (Ajzen & Fishbein, 1980; Rotter, 1954). Therefore, to the degree that self-efficacy is altered by incentives, it can be interpreted as a measure of intention rather than of perceived ability (Kirsch, 1985b), and to the degree that behavior is altered by incentives, it can be interpreted as being under volitional control. Thus, the provision of incentives to subjects in this study constitutes a test of the hypothesis that self-efficacy ratings of pain tolerance are measures of intention. The model for predicting pain perception and tolerance tested in this study was based on Kirsch's (1990) model relating response expectancy and self-efficacy to the prediction of approach behavior and fear. According to that model, self-efficacy for approaching a feared stimulus is determined by the degree of fear people expect to experience (their anxiety expectancy) and by the other outcomes that they anticipate (e.g., the threat of harm and the incentives for approach). As a behavioral intention, self-efficacy directly predicts approach behavior but has no effect on experienced fear. Conversely, experienced fear is directly affected by expected fear but not by incentives. Strong empirical support for this model has been reported by Schoenberger et al. (in press). An analogous model, in which pain perception and tolerance are related to response expectancy and self-efficacy, is presented in Figure 1. According to this model, self-efficacy medi-



ates the effects of expected pain and other outcome expectancies on tolerance, and expected pain has a direct effect on experienced pain. The experience of pain, however, is not affected by incentives or by self-efficacy.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

To test this model, we compared the effects of cognitive pain reduction strategies with those of a motivational manipulation on both pain perception and pain tolerance. Responses to a cold-pressor task were examined in four groups of subjects. One group was given training in multiple cognitive coping strategies, another was given monetary reinforcement contingent on tolerance, a third was given both cognitive training and monetary reinforcement, and a no-treatment control group was given neither. If self-efficacy is a measure of perceived ability rather than intention, then it should not be greatly affected by monetary incentives. More importantly, if the associated behavior is limited by ability, then incentives should have little effect on its performance. Conversely, if incentives have parallel effects on both self-efficacy and behavior, then we can reasonably infer that the behavior was within the person's capabilities and that the person knew it. In other words, the subject was responding to the efficacy questionnaire by indicating his or her intention rather than his or her perceived ability. Our hypothesis was that responses to a questionnaire assessing self-efficacy for coldpressor pain tolerance are indications of intention rather than perceived ability. For that reason, we expected those responses to be affected by a monetary incentive. Similarly, because coldpressor tolerance is an intentional act that is within most subjects' capabilities, we expected the incentive to affect their behavior as well as their stated intention (i.e., their self-efficacy). On the basis of the social learning model depicted in Figure 1, we predicted that cognitive training would affect subjects' expectancies about the degree of pain they would experience during the cold-pressor task and that those pain response expectancies would, in turn, affect both pain perception and selfefficacy. In contrast, we expected the monetary incentive to affect self-efficacy directly, without affecting either expected or experienced pain. Finally, consistent with the data on the relations among response expectancy, self-efficacy, fear, and avoidance (Schoenberger et al., in press), we predicted that self-efficacy would be a good predictor of behavior, even with response expectancy partialed out, and that response expectancy would

->• Pain


Method Subjects

Testing the Model

Pain Expectancy

be a good predictor of reported pain, even with self-efficacy partialed out.


Other Outcome Expectancies Figure 1. Social learning theory model of the relations between pain response expectancy, self-efficacy, pain perception, and pain tolerance.

Eighty female students enrolled in introductory psychology classes at the University of Connecticut participated voluntarily to satisfy course requirements. It was decided to limit the study to women because of an anticipated small number of men available to enroll in the study. All subjectsfirstcompleted a medical history screening to eliminate any persons who might be considered at risk for injury when undergoing the cold-pressor procedure. Subjects participated in a single experimental session and were randomly assigned to experimental conditions.

Procedure We asked subjects in all conditions to complete the MarloweCrowne Social Desirability Scale (Crowne & Marlowe, 1964). We then provided them with a brief exposure to the cold-pressor pain stimulus. While the subjects were seated next to a tank of ice water, we instructed them in the use of a visual analogue scale for rating pain intensity (Huskisson, 1974). We then instructed them to immerse their nondominant arms into the ice water, keeping theirfingersopen. After 5 s, subjects were instructed to remove their arms and then asked to rate the amount of pain they had experienced. After the baseline immersion, subjects were taken to a separate room in which the experimental treatments were administered. They were then brought back to the room where the cold-pressor test was performed, where they completed pain expectancy and self-efficacy questionnaires. After they completed these ratings, we asked subjects to immerse their arms in the ice water and instructed them to "keep your hand in the water as long as you possibly can." When they withdrew their arms from the water, we asked them to rate the amount of pain they had experienced. Subjects who kept their arms in the water for a full 8 min were instructed to remove their arms at that time.

Experimental Conditions Coping instructions. After the baseline cold-pressor immersion, subjects in this condition received 20 min of instruction in procedures taken from Meichenbaum (1977) and from Scott and Barber (1977) that were designed to enhance their ability to cope with pain. The instructions presented a number of different strategies that subjects could use to "reduce the discomfort of the pain." These included relaxation training, imagining that the hand and arm had become numb, relabeling the sensation as nonpainful, and reinforcing themselves with positive self-statements. Subjects were given practice in using these strategies and were instructed to use whichever of them seemed to work best for them. They were encouraged to change strategies if they found that one was not working. Incentive. In this condition, subjects spent 20 min with instructions to read, relax, or do homework according to their preference. After the waiting period, subjects returned to the cold-pressor room, where they received motivational instructions from the experimenter, who stressed the importance of tolerating the pain for as long as possible. They were then shown a stack of six $1 bills and told that they would receive two of the dollars for keeping their hand immersed for 4 min and an additional dollar for each additional minute up to a maximum of 8 min. Coping and incentive. We gave subjects in this group the same coping instructions that we gave to subjects in the cognitive coping condi-



This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tion. When they returned to the treatment room, we also gave them the motivational instructions and monetary inducement offered to subjects in the incentive condition. No treatment control. After the baseline immersion, subjects in the control group spent 20 min reading, relaxing, or doing homework. They then returned to the cold-pressor room for the second immersion and were asked to immerse their arms for as long as they possibly could.

Table 2 Analyses ofCovariance of Expected Pain, Reported Pain, Self-Efficacy, and Tolerance Variable

F(l, 74)


Cognitive mediators of pain perception and tolerance.

A social learning model of pain perception and tolerance was evaluated. Responses to cold-pressor pain were examined in 4 groups of 20 Ss each. One gr...
789KB Sizes 0 Downloads 0 Views