No pain, no change: Reductions in prior negative affect following physical pain
- First Online:
- Cite this article as:
- Bresin, K., Gordon, K.H., Bender, T.W. et al. Motiv Emot (2010) 34: 280. doi:10.1007/s11031-010-9168-7
- 366 Downloads
In general, organisms are motivated to avoid stimuli that induce pain. However, some individuals intentionally inflict pain on themselves (e.g., nonsuicidal self-injury) and report doing so for the perceived emotional benefits following the experience of pain. Two controlled laboratory studies sought to expand upon the relatively limited literature on the effects of pain on emotion. In Study 1, participants provided momentary affect ratings immediately before and after experiencing physical pain. Results demonstrated that both positive affect and negative affect (NA) decreased following the experience of pain. In the Study 2, we examined the effect that individual differences in emotional reactivity had on affective reactions to pain. Individuals high in emotional reactivity experienced larger decreases in NA following the experience of pain than individuals who were low in emotional reactivity. Our findings may potentially explain why some individuals intentionally seek out the experience of pain.
KeywordsEmotional reactivity Negative affect Positive affect Pain Nonsuicidal self-injury
It has long been proposed that living organisms are motivated to approach pleasure and to avoid pain (Freud 1952). Contrary to this general principle, there are times when people seek out behaviors that involve pain (e.g., nonsuicidal self-injury [NSSI]; self-inflicted damage to one’s body tissue without fatal intent; Klonsky 2009). It may be that certain people acquire the motivation to engage in these painful behaviors because of perceived benefits, such as reductions in intensely negative emotions following the experience of pain (Solomon 1980). Though previous research has found that emotion can affect the experience of pain (e.g., Rainville et al. 2005), little research has examined the effect that pain has on the experience of emotion. In an effort to gain a better understanding of the effects of physical pain on emotional states, we conducted two studies that examine momentary changes in affect following laboratory-induced pain.
Research on NSSI may help to describe the effects that pain has on emotion in clinical populations. Studies using imagery scripts of NSSI episodes have shown that people experience a drop in heart rate and decrease in negative emotions after they have imagined engaging in NSSI (Haines et al. 1995; Brain et al. 2002; Welch et al. 2008). Meanwhile, an ecological momentary assessment study by Muehlenkamp et al. (2009) found that, following a NSSI incident, individuals with bulimia nervosa reported a significant increase in positive affect (PA). Both studies suggest that acute physical pain may lead to emotional regulation benefits. These findings are consistent with reasons frequently given for NSSI (e.g., “to stop bad feelings”; Nock and Prinstein 2004). Yet, it is unclear how these effects operate in the general population. Moreover, in these studies, the types of pain experienced varied greatly between individual participants.
Two recent studies examined the effect of physical pain on specific emotions in controlled, laboratory settings in nonclinical samples. Hollin and Derbyshire (2009) found that female participants rated fear lower in the presence of both a spider and pain delivered from a cold pressor task than in the presence of the spider alone. The presence of a spider did not affect the experience of pain, as participants reported the same pain intensity and unpleasantness during the cold pressor both with and without the spider present. This suggests that physical pain may have inhibited the fear response to the spider, but that the presence of the spider did not influence the experience of pain. Leknes et al. (2008) found that, following the termination of a heat stimulus, participants reported an increase in relief. Limitations of the studies include the reliance on only female participants (i.e., Hollin and Derbyshire 2009) and an assessment limited to one specific emotion rather than broad affective constructs such as PA and negative affect (NA; Watson et al. 1988).
Based on the current research it could be inferred that all humans, not just individuals with a history of NSSI, may have a decrease in negative emotions following the experience of pain. It is unclear then why only some individuals use self-inflicted physical pain as a way to regulate emotion. It has been proposed that individuals who are high in emotional reactivity may engage in dysfunctional behaviors, such as NSSI, to down regulate their negative emotions (Linehan 1993; Chapman et al. 2006; Selby et al. 2009). Emotional reactivity is defined as increased sensitivity, intensity, and persistence of emotions (Nock et al. 2008). Individuals who are high in emotional reactivity are more sensitive to emotional stimuli and this sensitivity elicits more intense emotions that persist for prolonged period of time. It is possible that individuals high in emotional reactivity may have greater emotional benefits from the experience of pain compared to those low in emotional reactivity.
In summary, though the effects of emotion on the perception of pain are relatively well understood, less is known about the effect that pain has on the experience of emotion. In order to expand upon previous research in the area, we conducted two studies examining the effects of laboratory-induced pain has on emotional states. In the first study, participants provided self-report measurements of their affective state immediately prior to and following exposure to a painful pressure stimulus. In the second study, we sought to examine how individual differences in emotional reactivity might moderate the effects from Study 1. Based on previous research (e.g., Hollin and Derbyshire 2009), we predicted that NA would be lower at Time 2 (i.e., following pain) than Time 1 (i.e., prior to experiencing pain). In terms of PA, we predicted that, similar to Muehlenkamp et al. (2009), PA would be higher following the experience of pain than prior to the experience of pain. We also examined if sex moderated these effects.
Materials and methods
One hundred sixty-seven undergraduate students (99 female) participated in the study for class credit. The mean age was 19 years (range 17–34 years). Smokers and left-handed participants were excluded from the study because of evidence that these factors affect pain tolerance (Murray and Hagan 1973; Pomerleau et al. 1984). Participants were also instructed to refrain from ingesting any analgesics (e.g., aspirin), sugared food, or alcohol for 8 h prior to their appointment (Mercer and Holder 1997). All procedures were approved by the institutional review board, and all participants provided informed consent prior to participation.
Pain threshold and tolerance
Pain threshold and pain tolerance were assessed with a pressure algometer (Type II; Somedic, Solletuna, Sweden). The algometer was applied perpendicularly to the skin and lowered at a rate of 5 kilopascals (kPa) per second until pain tolerance was reached, as indicated by participants’ verbal report. For pain threshold trials, participants were instructed to say “now” when they first felt pain due to the increased pressure. For pain tolerance trials, participants were instructed to say, “stop” when the pain became too uncomfortable to continue. At this point, the experimenter immediately retracted the algometer. The digital display showed the value of the pressure (in kPa) applied at the moment the algometer was retracted. All pain measurements were taken at the first dorsal interosseous muscle (i.e., behind the first knuckle of the index finger) of the participant’s right hand. To prevent habituation, there was a 90 s interval between measurements (Orbach et al. 1997). Participants completed five trials each for pain threshold and pain tolerance, and they were combined to form one average score (Threshold: M = 170.52, SD = 93.58; Tolerance: M = 388.03, SD = 216.52).
Positive and negative affect schedule (Watson et al. 1988)
The PANAS has been used by researchers in a variety of populations and has adequate psychometric properties (Watson et al. 1988). It contains 10 items that comprise the NA composite score and 10 items that comprise the PA composite score. In the state version of the PANAS, participants are asked to rate the degree to which they feel each emotion “right now” on a scale from 1 to 5 (1 = very slightly or not at all, 2 = a little, 3 = moderately, 4 = quite a bit, 5 = extremely). Sample items from the PANAS are “enthusiastic” (PA) and “ashamed” (NA). Participants rated their mood on the PANAS immediately before (Time 1) and after (Time 2) the painful algometer task. The PANAS had adequate reliability in this sample (NA: Time 1 α = .78, Time 2 α = .76; PA: Time 1 α = .87, Time 2 α = .89).
Participants arrived at the laboratory individually or two at a time, and completed questionnaires about demographic information in separate rooms. Next, they completed a PANAS with instructions to indicate the extent to which they felt each emotion “AT THIS MOMENT” (Time 1). Following this, the procedures of the algometer task were explained to the participant. Odd numbered participants completed pain threshold trials first, while even numbered participants completed pain tolerance trials first. Following all ten trials, participants completed a second PANAS (Time 2) with the same instructions as Time 1.
Means, standard deviations of affect for Study 1 and Study 2
Time 1–Time 2
Time 1–Time 2
Results from the ANOVA predicting PA revealed a trend towards main effect for sex, F(1, 162) = 3.48, p = .06, as well as a main effect for time F(1, 162) = 38.54, p < .001, η2 = .19. There was also a significant time × sex interaction F(1, 162) = 13.01, p < .05, η2 = .02. Contrary to our predictions, PA decreased from Time 1 to Time 2. The nature of the interaction was such that males (Md = .34, SD = .56) had a larger decrease in PA compared to females (Md = .17, SD = .47), and this difference was significant, t(162) = 2.09, p < .05, d = .33. Males also had significantly higher PA at Time 1 compared to females. There were no group differences at Time 2.
In order to address limitations of previous research, we examined the effect that acute physical pain has on positive and negative emotions in males and females. Consistent with predictions and previous research (Hollin and Derbyshire 2009), NA decreased following the experience of pain. Contrary to our predictions and previous research (Muehlenkamp et al. 2009), PA decreased following the experience of pain. Both effects where moderated by sex, such that females evidenced a larger drop in NA following the experience of pain, while males evidenced a larger drop in PA following the experience of pain.
The results of Study 1 suggest that the experience of laboratory pain tends to dampen the experience of emotions. These findings are consistent with reasons given for NSSI (Klonsky 2009) and previous research on fear (Hollin and Derbyshire 2009). However, these findings do not tell us which individuals feel the most motivated to engage behaviors such as NSSI. We sought to extend the results from our first study by examining the influence of a clinically relevant individual difference variable, emotional reactivity, on emotional states following the experience of pain. Emotional reactivity is defined as increased sensitivity, intensity, and persistence of emotions (Nock et al. 2008). Individuals who are high in emotional reactivity are more sensitive to emotional stimuli and this sensitivity elicits more intense emotions that persist for prolonged period of time. High levels of emotional reactivity have been associated with multiple mental health conditions, including those with elevated rates of NSSI, such as borderline personality disorder (Linehan 1993) and eating disorders (Anestis et al. 2009).
It has been proposed that individuals who are high in emotional reactivity engage in dysfunctional behaviors such as NSSI to down regulate their negative emotions (Linehan 1993; Chapman et al. 2006; Selby et al. 2009). This suggests that, in order to cope with intense, long duration emotions, individuals may turn to drastic strategies to escape feeling their emotions. Consistent with this prediction, individuals who have a history of NSSI score higher on measures of emotional reactivity than individuals with no history of NSSI (Nock et al. 2008). However, these models are primarily based on retrospective self-report measures, and lack laboratory evidence to support them.
Therefore, the goals of Study 2 were twofold. First, we sought to replicate our findings and extend them to another pain modality (thermal pain). Secondly, we tested whether individual differences in emotional reactivity moderated the effect that pain has on emotion. As in Study 1, we predicted that NA would decrease following the experience of pain, and this difference would be larger for females. Similarly, we predicted that PA would decrease following the experience of pain, and this difference would be larger for males. We also predicted that individuals high in emotional reactivity would have a larger drop in NA following pain as compared to those low in emotional reactivity. Previous research has on emotional reactivity has focused on negative emotions (Nock et al. 2008). Thus, the examination of PA was exploratory, and not based upon specific predictions.
Materials and methods
One hundred eighty-four undergraduate students (89 female) participated in the study for class credit. The mean age was 19.5 years (range 18–36 years). Similar to Study 1, smokers and left-handed participants were excluded. Participants were also instructed to refrain from ingesting any analgesics, sugared food, or alcohol for 8 h prior to their appointment. All procedures were approved by the institutional review board, and all participants provided informed consent prior to participation.
All pain stimuli were produced by the Thermal Sensory Analyzer (TSA; Medoc TSA 2001; Ramat-Yishai, Israel), which is a computerized device for the quantitative assessment of sensory and pain tolerance of heat- and cold-induced pain by means of a thermode. The thermode consists of a semiconductor function that produces a temperature gradient by the passage of an electric current. All pain measurements were taken at the first dorsal interosseous muscle (i.e., behind the first knuckle of the index finger) of the participant’s right hand. The adaptation temperature was 25°C, at which the participant feels neither warmth nor cold. On each trial, temperature levels and gradient increased at a rate of 1°C per second until it reached a temperature of 50.5°C, or the participant indicates that they felt pain by clicking a button on a computer mouse. After pressing the button, the temperature decreased at a rate of 10°C per second to the adaptation temperature. Participants completed five trials, with 30 s in between each trial. As with Study 1, all trials were averaged to compute a mean score (M = 47.54, SD = 3.51).
Positive and negative affect schedule (Watson et al. 1988)
Affect was assessed the same way as in Study 1. Internal consistency was high in this sample (NA: Time 1 α = .85, Time 2 α = .75; PA: Time 1 α = .90, Time 2 α = .91). Similar to Study 1, participants completed the PANAS immediately before and immediately after the pain assessment.
Emotional reactivity scale (ERS; Nock et al. 2008)
The ERS is a 21-item self-report measure designed to assess the individual differences in emotional reactivity. Participants rate items on a five-point Likert scale, where 1 is “not at all like me” and five is “completely like me”. Items are intended to tap sensitivity (e.g., “I tend to get very emotional very easily”), intensity (e.g., “When I experience emotions, I feel them very strongly/intensely”), and duration (e.g., “When I am angry/upset, it take me much longer than most people to calm down”). In spite of this a priori hypothesis, Nock et al. found a one factor solution adequately fit the data. Therefore, all items were averaged to create one ERS score (M = 2.14, SD = .74). Nock et al. (2008) provide initial construct and criterion validity for the ERS. The internal consistency in this sample was high (α = .93).
Participants arrived at the laboratory in groups of three or less. Even numbered participants completed the pain assessment prior to completing the ERS on an individual computer, while odd numbered participants completed the measures in the reverse order. Participants completed the pain assessment individually in a room that was separate from the other participants. Similar to Study 1, participants were asked to fill out the PANAS with instructions to indicate the extent to which they felt each emotion “AT THIS MOMENT” (Time 1). Next, the experimenter showed the participant the Thermal Sensory Analyzer and explained the instructions for the task. Following the five trials, the participants again completed the PANAS with the same instructions as before (Time 2).
In order to replicate analyses from Study 1, two mixed-model ANOVAs were conducted. Consistent with Study 1, there was a main effect of time, F(1, 179) = 77.46, p < .01, η2 = .30, no effect of sex, F(1, 179) = 1.29, p = .25 and a time × sex interaction, F(1, 179) = 9.39, p < .01, η2 = .05, for NA. Consistent with Study 1, the nature of the interaction was that both males and females had a significant drop in NA from Time 1 to Time 2, and this effect was more pronounced for females (see Table 1). For PA, there was a main effect of sex, F(1, 179) = 8.14, p < .001, η = .04 and time, F(1, 179) = 3.99, p < .05, η = .02, but no time × sex interaction, F(1, 179) = .85, p = .35. Regardless of time, males had higher PA than females. Also regardless of sex, PA was lower at Time 2 than Time 1.
In an effort to understand the effect of emotional reactivity on the experience of affect in relation to pain, we added ERS scores to the previous ANOVAs. We conducted two 2 (sex: male and female) × 2 (Time: PANAS1 and PANAS 2) × ERS ANOVAs, with sex and ERS as between-subject factors and time as a within-subject factor predicting NA and PA. There was a significant main effect of ERS on NA, F(1, 178) = 42.16, p < .001, η2 = .24, and there were no main effects of sex or time (F’s < 1). The interaction between time and sex was still significant, F(1, 178) = 4.32, p < .05, η2 = . 02. More crucial to our hypotheses, there was a time × ERS interaction, F(1, 178) = 13.15, p < .001, η2 = .07. The three-way interaction did not reach significance, and there were no significant main effects or interactions when PA was the outcome variable.
By definition, individuals high in emotional reactivity are more sensitive to emotional stimuli, and this sensitivity leads to more intense and longer duration emotions. Therefore, it stands to reason that following pain, individuals high in emotional reactivity would show an increase in NA. In spite of this, the results from Study 2 suggest that individuals high in emotional reactivity actually experience a decrease in NA following the experience of pain, which appears to be greater than the decrease in NA experienced by individuals low in emotional reactivity. This finding is consistent with theoretical models of dysregulated behavior (Linehan 1993; Chapman et al. 2006; Selby et al. 2009). The results of Study 2 replicate and extend those of Study 1.
The current studies examined the effects of acute laboratory-induced pain on emotional states. Results from two studies suggest that, following the experience of pain, both negative and positive emotions decrease. These effects were similar in direction for males and females, though females tended to have a larger decrease in NA, and this effect was found in both studies. Males had a larger drop in PA following the experience of pain in Study 1. This finding was not replicated in Study 2. Finally, emotional reactivity was a significant predictor of NA following pain induction, such that individuals high in emotional reactivity experienced a greater decrease in NA following pain than individuals with low levels of emotional reactivity.
Our results build upon previous findings (e.g., Hollin and Derbyshire 2009) by demonstrating that the induction of pain results in a general decrease in emotions, and that this effect is not restricted to fear. Secondly, in terms of the emotional circumplex, NA can be thought of as a bipolar dimension where NA ranges from intense NA to low NA (cf. relaxation; Watson 2000). Therefore, the observed drop in NA could be thought of as a rise in relief, which is consistent with the findings of Leknes et al. (2008) and Haines et al. (1995). In terms of PA, our findings are contradictory to the limited previous research that has been conducted. Muehlenkamp et al. (2009) found that, following an NSSI incident, individuals experienced a rise in PA over time. However, the time frame in our study was different (Muehlenkamp et al.’s assessment of affect was often hours after the NSSI incident), the pain induction was not standardized across individuals, and the sample was clinical (i.e., women with bulimia nervosa). It is possible that there is an immediate drop in PA for our participants, and then a gradual rise over the next few hours, but our follow-up period was not long enough to observe this change. Finally, our findings that higher levels of emotional reactivity predicted greater reductions in NA following pain are consistent with models of NSSI which suggest that individuals who are emotionally reactive utilize NSSI to escape their intense emotions (Chapman et al. 2006). Study 2 expanded upon previous tests of these models, which were primarily based on retrospective self-report data.
Our findings may have theoretical implications and point to areas for future research. For example, our findings are consistent with Eccleston and Crombez’s (1999) cognitive-affective model of pain, which suggests that the threat of a stimulus is an important factor in the experience of pain. The more threat value a stimulus has, the more attention is drawn to the sensations, and thus more pain is experienced. It is possible that, compared to the current affective state, the pain was more threatening and thus required more attention than affect, leading to the decrease of NA or PA. The items on the PANAS that tap into PA and NA are items that reflect high engagement and extremes of positive and negative valence, respectively (Watson et al. 1988). Therefore, since valence is essentially held constant, decreases in both PA and NA at the same time may reflect changes in engagement. To further test this hypothesis, a more thorough assessment of affect would be necessary (e.g., PANAS-X; Watson and Clark 1994). Future research may also benefit by experimentally manipulating the intensity of the pain and the emotional state to see if differing threat levels moderate the relationships found in these studies.
Our findings are also in line with Solomon’s opponent process theory (Solomon 1980). The opponent process theory suggests that any stimulus that causes a change from homeostasis causes a response (i.e., the A process). The A process is followed by an opponent process (i.e., the B process) opposite in valence when the original process is terminated. The opponent process theory also suggests that over repeated stimulations, the A process (e.g., pain) remains relatively unchanged, while the B process (e.g., relief) is strengthened. Thus, a negative event, such as physical pain, could be associated with more rewarding properties over time. Though our findings are consistent with the opponent process theory, future studies that directly test the model may be needed to explain how the use of physical pain to change emotions could become increasingly more powerful over time.
Our findings should be viewed within the context of the studies’ limitations. First, we did not measure PA and NA prior to the knowledge of the pain task. Our measurements were immediately before and after the experience of pain. It is possible that participants felt increased NA at Time 1 compared to Time 2 because they were experiencing anticipatory anxiety in relation to the pain task, and our results are simply NA returning to baseline. We find this explanation unlikely. First, our Time 1 measures of NA are consistent with published norms of state measures of the PANAS (Watson and Clark 1994). If the anticipation of pain caused increased NA, participants should have had NA scores higher than the baseline norms. Therefore, it is unlikely that our results are due to the anticipation of pain. Second, even if this explanation were true it does not explain why PA also decreased from Time 1 to Time 2. Future replications should further rule this explanation out by including a measure of affect prior to the knowledge of the pain task.
Second, though our effect sizes were moderate to large, it may be unclear whether these differences are clinically significant. Watson (2000) found that participants spent about 72% of their time feeling “very slightly or none at all” NA (a score of 1), and 49% of their time feeling “A little” PA (a score of 2) in a large database of momentary data. Watson (2000) has argued that high intensity emotions are relatively rare. Therefore, our effects may be considered important as they are in the range that most affect is experienced. Also, the difference in NA between Time 1 and Time 2 in our studies (.22) is similar to the difference between undergraduate students and psychiatric inpatients in the PANAS momentary norms (.53; Watson and Clark 1994), suggesting that a difference of this size may be meaningful. In spite of this, future research may benefit from using experimental mood inductions rather than naturally-occurring emotions to examine effects of pain at higher levels of affect.
Finally, our findings may not generalize to clinical populations of people who engage in NSSI. It is possible that individuals who use NSSI as a way of regulating emotions are qualitatively different from those that do not. However, to our knowledge, this is the first study to test the immediate effects of physical pain on PA and NA in any sample. In order to understand the differences between NSSI groups and non-NSSI groups, a general pattern in a normative sample may be a useful comparison. Though our sample consisted of nonclinical undergraduate students, this sample is appropriate in that this age group is among those at highest risk for NSSI (Heath et al. 2008). In conclusion, we have presented evidence that physical pain leads to a reduction in both positive and negative emotions, particularly among individuals who are highly emotionally reactive. It is hoped that future research can build upon our findings by testing for differential effects of pain on emotions in clinical samples.
This research was supported, in part, by a grant from ND EPSCoR to Kathryn H. Gordon (NSF Grant EPS-081442).