Physiology & Behavior 138 (2015) 107–112
Contents lists available at ScienceDirect
Physiology & Behavior journal homepage: www.elsevier.com/locate/phb
The effect of disgust on pain sensitivity Megan J. Oaten a, Richard J. Stevenson b,⁎, Trevor I. Case b a b
Department of Applied Psychology, Griffith University, Southport QLD4222, Australia Department of Psychology, Macquarie University, Sydney NSW2109, Australia
H I G H L I G H T S • • • • •
Disgust can induce effects similar to the acute phase response. Here we examine if it can also induce increased pain sensitivity. Immediately after a disgust induction pain was reduced, but later it increased. Negative and positive inductions produced the reverse outcome. We suggest that disgust may enhance pain sensitivity.
a r t i c l e
i n f o
Article history: Received 22 April 2014 Received in revised form 20 October 2014 Accepted 23 October 2014 Available online 29 October 2014 Keywords: Disgust Pain Emotion
a b s t r a c t Experiencing the emotion of disgust leads to delayed up-regulation of immune-related functions, increased corebody temperature and reduced appetite. These changes parallel those of the acute phase response, which occurs when a pathogen is detected by the immune system. Here we examined whether a further predicted aspect of the acute phase response is evident following disgust induction, namely increased pain sensitivity. Participants attended a two-session experiment. On one session they experienced an emotion induction (being randomly assigned to either disgust, negative or positive groups) and on the other they received a neutral control induction. Before and after each induction, and at 15 and 30 min post-induction, participants engaged in a cold-pressor task, rating pain intensity at 10 s intervals for 90 s on each occasion. Relative to neutral control and pre-test, average pain intensity decreased then increased across time following the disgust induction, with the reverse pattern in the negative and positive emotion inductions. These findings are the first to suggest that disgust may lead to an increase in pain sensitivity over a time course paralleling changes observed for core-body temperature and immune-related function, although the mechanisms underpinning these effects remain to be identified. © 2014 Elsevier Inc. All rights reserved.
1. Introduction It has been suggested that the emotion of disgust functions in humans as part of a larger system of defensive behaviours and physiological responses, which assist us in avoiding infectious disease [1]. This functional account of disgust arose primarily because of the close association between stimuli that elicit this emotion (e.g., corpses, faeces, wounds, rotten food) and their capacity to transmit pathogens [2]. This account has received additional support in recent years from the finding that exposure to disgust elicitors, as well as cues that remind participants of disease, can serve to activate the innate immune system [3–5]. For example, participants who viewed pictures of disgusting stimuli were found to have elevated levels of tumour-necrosis factor alpha (TNF-a) in their saliva around 30 min post-induction — an effect not observed in control conditions ⁎ Corresponding author. Tel.: +61 2 98508098; fax: +61 2 98508062. E-mail address: [email protected] (R.J. Stevenson).
http://dx.doi.org/10.1016/j.physbeh.2014.10.023 0031-9384/© 2014 Elsevier Inc. All rights reserved.
that use equally affectively negative stimuli [5,6]. Not only can disgust stimuli activate the innate immune system, this seems to extend to increasing core body temperature as well [6]. Viewing disgusting images, but not equally emotionally negative control images, resulted in progressive increases in core body temperature, which were maximal at around 30 min post-induction [6]. The temperature-related finding, and the immune system changes, led us to hypothesise that disgust might produce changes in the body that parallel the acute phase response. In the acute phase response, the immune system detects the presence of a pathogen and reacts to them both locally and systemically [7]. This systemic reaction includes fever, elevated levels of cytokines, tiredness and social withdrawal, loss of appetites (e.g., food), and increased pain sensitivity [8]. Paralleling this pattern of responding, disgust does appear to increase core body temperature [6], elevate levels of cytokines including TNF-a and IL-6 [3,5,6], and disgust is associated with feelings of nausea and reduced appetite for food [9,10]. In the study reported in this manuscript we focussed on whether a further parallel to the acute phase response is present, namely increased pain
108
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
sensitivity. To our knowledge only one study has investigated pain perception following experiencing the emotion of disgust, and this found that latency to report pain was increased immediately after induction of disgust, relative to a neutral control [11]. This study did not, however, examine whether any increased pain sensitivity occurred at later time points (i.e., around 30 min post-induction), which would parallel changes observed for temperature and immune-related effects. While we might hypothesise differences in pain sensitivity based upon disgust inducing an acute phase response, this is of course not the only way that experiencing this emotion might affect pain perception. Unlike the other basic negative emotions (i.e., fear and anger), disgust primarily activates the parasympathetic branch of the autonomic nervous system e.g., [12]. As autonomic function changes markedly during pain perception e.g., [13], it is likely that disgust-induced changes in parasympathetic activation would affect participants' experience of pain during the period that the emotion was being experienced. However, while some effect would be expected, it is not obvious what the precise nature of this impact would be nor its time course. A further emotion-related effect on pain perception concerns the degree to which they affect participants' orientation to the internal or external milieu. As the negative emotions generally deal with external threats, they may favour an outward shift of attention, thereby acting to reduce pain perception via distraction [14]. In this case, we would expect the effect to occur in close proximity to the experience of the emotion. To test whether disgust leads to increased pain sensitivity following longer delays – and to determine its effects on pain more broadly – we had participants randomly assigned to one of three experimental groups. Each group received an emotion induction on one day and a neutral control induction (viewing images of everyday household objects) on another, in counterbalanced order. This allowed us to ascertain the unique effects of each emotion induction on participants' perception of pain by measuring their baseline responding during the neutral control induction. The content of the emotion inductions differed between groups, with one set of participants receiving a disgust induction, one a negative induction (i.e., unpleasant, fear and anger provoking stimuli) and another a positive induction. This then allowed us to compare between groups, the unique effect of each emotion induction on pain perception. This between group manipulation was adopted so as to reduce the demands the study made upon participants (i.e., number of pain tests and experimental sessions). We included a negative emotion induction so that we could determine if any pain-related effect arose simply because disgust induces negative affect. A positive emotion induction was included to determine whether any form of valenced and arousing stimulation might account for any pain-related effect (e.g., via distraction). For the emotion inductions the requisite state was induced by showing participants particular sets of images. We established the success of the emotion inductions by having participants evaluate their emotional state before and after the induction. Pain sensitivity was tested using a variant of the cold pressor task, which is a well-established experimental technique for inducing pain, and from which reliable and valid selfreports of pain can be obtained e.g., [15]. On each day of testing participants' pain sensitivity was established, prior to the induction, and then three times afterwards; immediately, at 15-min and 30-min post induction. As we expected the most interesting alterations in pain sensitivity to emerge at the later time points (i.e., consistent with the immune and temperature related changes noted above), this necessitated multiple cold-pressor tests. Consequently, the temperature of the water was set at a warmer-level than normal. On each test participants were asked to keep their lower arm immersed in the cold water for 90 s, reporting their pain intensity at 10 s intervals. Finally, we asked all participants to complete individual difference measures of disgust sensitivity [16, 17] so that we could determine if any observed effects were stronger in participants who report experiencing this emotion with greater frequency and intensity.
2. Method 2.1. Participants Ninety-six undergraduate participants (31% male; M age = 20.2, SD = 3.3), all with no pre-existing medical condition that could affect pain responses, were randomly assigned to one of three experimental groups. Participants were recruited from the first-year psychology subject pool and from the university community, the latter being paid a small sum ($20) for taking part. Half of the sample self-reported as being Australians of Caucasian descent, with the remainder mainly composed of Australians of Asian descent. This proportion did not significantly differ between the experimental groups. Just over one-third of the sample was born overseas, and again this proportion did not significantly differ between the experimental groups. Each participant consented to take part in the experiment and the protocol was approved by the Macquarie University Human Research Ethics Committee. 2.2. Stimuli The pictorial stimuli used for the inductions were all obtained from the International Affective Picture Series (IAPS, [18]) and were shown twice in randomised order to participants on a 60 cm computer monitor. Disgust induction: IAPS disgust-related stimuli (e.g., vagrant, roaches, vomit), items; 1280, 2710, 2750, 3030, 3051, 3150, 3160, 3400, 7359, 7380, 9140, 9181, 9252, 9290, 9300, 9301, 9320, 9342, 9405, 9500. Negative induction: IAPS negatively-valenced stimuli (e.g., pointed guns, domestic violence, plane crash wreckage), items; 2141, 2455, 2800, 3180, 3500, 6230, 6300, 6311, 6313, 6315, 6510, 6571, 6830, 6838, 8485, 9041, 9050, 9421, 9611, 9910. Positive induction: IAPS positively-valenced stimuli (e.g., skydiving, white-water rafting, water skiing), items; 5480, 5621, 5950, 8030, 8034, 8080, 8160, 8178, 8179, 8180, 8185, 8186, 8191, 8192, 8200, 8260, 8300, 8341, 8400, 8490. Based upon the IAPS normative data, there was no significant difference in affective valence between the Disgust (M = 2.5, SD = 0.4) and Negative (M = 2.4, SD = 0.4) image sets, and both of these sets differed in affective valence from the Positive image set (M = 6.8, SD = 0.8; twosample t-tests, both ps b 0.01). For arousal, there was no significant difference on the IAPS normative data between the Disgust (M = 5.6, SD = 0.8) and the Negative (M = 5.9, SD = 0.8) image sets, and both of these sets were significantly less arousing than the Positive (M = 6.6, SD = 0.5; two-sample t-tests, both ps b 0.01) image set. The Neutral induction set (experienced by all participants) contained the following images: IAPS household objects (e.g., chair, rolling pin, book), items: 7000, 7002, 7004, 7006, 7009, 7010, 7020, 7025, 7030, 7035, 7050, 7080, 7090, 7150, 7175, 7179, 7211, 7217, 7233, 7235. Using the IAPS normative data, these images had a mean valence score of 4.9 (SD = 0.2) and a mean arousal score of 2.6 (SD = 0.5). Needless to say, both of these scores significantly differed from the mean valence and arousal scores of each of the emotion image induction sets (two-sample t-test; all ps b .01). Pain intensity was measured using a 100-point scale (anchors; 0 = No pain, 20 = Mild pain, 40 = Moderate pain, 60 = Moderately severe pain, 80 = Very severe pain, 100 = Worst possible pain). This scale was visible throughout testing and participants were asked to select a number between 0 and 100, which best reflected the degree of pain intensity that they were currently feeling. Emotion ratings were completed on seven point category scales (anchors 1 = Not at all to 7 Very). Participants were asked to rate how sad, angry, disgusted, tense, fearful and happy they were. These items were selected for rating as they encompass the principal states likely to be induced by the image sets. The cold-pressor task was conducted using a refrigerated circulating water bath (Lab Companion RW-2025G), with the temperature set at 13.5 °C. This temperature was based upon pilot work, so that
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
participants would be willing to undergo multiple repeated assessments on two different days. Prior to and just after each cold-pressor task was conducted, participants were asked to place the hand to be used (or that had just been used) inside a heated woollen glove for 90 s to ensure that skin temperature was similar before and after each task. The hand chosen was alternated across the four tests in each session, starting either with the dominant or non-dominant hand on each session, this being determined by the counterbalancing schedule. 2.3. Procedure Participants attended two one-hour experimental sessions oneweek apart. Each session was conducted at the same time of day and on the same day of the week for each participant. All testing conducted between 9 am and 4 pm, in an air-conditioned laboratory, with just the experimenter present. Prior to attendance participants were telephone screened to ensure that they did not have any current medical conditions that would preclude them from taking part (e.g., Raynaud's syndrome, cardiovascular complaints, drug abuse etc). On arrival, participants were asked to read (and if happy sign) the information and consent form, which described the experimental procedure in some detail but only outlined the study aims in a general way. On one session participants completed the emotion ratings, a baseline cold-pressor test (warm glove, cold-pressor and ratings, warm glove), a further set of emotion ratings, an induction, a third set of emotion ratings, and then a second cold-pressor test. This was followed by two-further cold-pressor tests, one at 15 min after the induction and another 30 min after the induction. On one of the sessions, all participants received the neutral induction. On the other session they either received the disgust, negative or positive induction dependent on the group to which they had been assigned. Completion of the two inductions was in counterbalanced order. At the end of the second session, participants were asked to complete two questionnaires to assess their disgust sensitivity — the Three-Domain Disgust Scale [16] and the Revised Disgust Sensitivity Scale [17]. All participants were debriefed regarding the specific aims of the study at the end of their second session. 2.4. Analysis Participants completed three sets of emotion ratings on each of their two test sessions. From each set we generated two key variables. The first was the disgust score, to determine if the disgust induction had worked. The second was the negative score, which was a composite of the negative emotions, sad, angry, tense, and fearful, with happiness reverse coded. These ratings were all collapsed together as they were all positively correlated (median r = 0.39), reducing the need for multiple significance tests. The two key variables from the first set of emotion ratings, which were obtained before any pain testing or emotion induction had taken place, were compared using ANOVA to check that participants did not initially differ by Group (i.e., disgust vs. negative vs. positive) or Session (i.e., neutral control induction vs. emotion induction). The disgust and negative scores obtained just before and just after the induction (the control neutral induction session data are not reported as there were no significant changes in ratings in this condition), were subtracted from each other to measure any induction-related changes. These data were then compared by ANOVA to check that disgust increased following the disgust induction, and that negative emotion increased following the negative induction, and fell in the positive induction. Gender and the counterbalancing order (i.e., neutral first vs. emotion first) were initially included in all of these analyses, but as neither generated any significant effects, so the analyses were repeated and reported without these variables. The participants, who did not attend both sessions, were excluded from these analyses. We used planned contrasts to determine whether each group differed following a significant omnibus F test.
109
The key measure of pain sensitivity was the average pain score. This was the average of all pain ratings obtained on a given cold-pressor test. This score was used as participants varied in how many ratings they made before reaching the agreed cut-off of 70, at which point participants were instructed to remove their arm. A key assumption with this approach is that the pain ratings changed in an approximately linear fashion across time on each cold-pressor test. This appears a reasonable assumption as we could find no significant differences in linear slope coefficients by emotion induction type (between groups) or when these were compared to the control session (within groups). Indeed, the best fitting model was linear when tested on various subsets of these data (i.e., recall that participants differed in how many pain ratings they gave on each trial). Each participant completed eight cold pressor tests and this resulted in eight average pain ratings — four for the emotion induction session (pretest baseline, immediate post-test, 15 min post induction, 30 min post induction) and four for the control session (pretest baseline, immediate post-test, 15 min post induction, 30 min post induction). For three cases, participants attended the emotion induction session but did not return for the neutral induction control session (1 case attended the neutral induction and did not return for the emotion induction). We chose to include these 3 cases, in the analyses, using their pre-test baseline average pain rating in lieu of their control session (noting that their omission does not change the study outcome). As there were highly significant violations of sphericity, these data were analysed using MANOVA with Group (between factor; Disgust, Negative, Positive emotion inductions), Session type (within factor; Emotion induction vs. Control induction), and Time (within factor; Pretest baseline vs. Immediate post-test vs. 15 min post induction vs. 30 min post induction) as factors. We also initially included Gender and Counterbalancing order, but as these two last-mentioned variables generated no significant effects, we repeated (and report) the analysis without these factors. For the second analysis we conducted the following additional modifications to the data set so as to isolate the unique effect of each emotion induction. For the emotion induction session, we subtracted its pretest baseline average pain rating, from the three post-induction scores. The same procedure was then followed for the control session. This left six scores composed of: (1) the emotion-induction scores three variables representing the change in average pain relative to baseline for the emotion induction session immediately after the induction (T1), at 15 min (T2) and 30 min (T3) post-induction; and (2) the control-induction scores - three variables representing the change in average pain relative to baseline for the control induction session immediately after the induction (T1), at 15 min (T2) and 30 min (T3) postinduction. Finally, we subtracted each respective control induction score (T1 to T3) from its corresponding emotion induction score (T1 to T3), leaving three target pain scores. These scores reflect the unique effect generated by each emotion induction, at T1, T2 and T3, relative to pre-test baseline and the neutral control condition. Negative values for the three target pain scores indicate a fall in average pain and positive values an increase in average pain, relative to pre-test baseline and the neutral control condition. These data were analysed with a mixed design ANOVA using Group (between factor; Disgust, Negative, Positive emotion inductions) and Time (within factor; T1, T2, T3), with Gender and Counterbalancing order initially included. As these two last-mentioned variables again generated no significant effects, we repeated (and report) the analysis without these factors. For significant effects involving Group (i.e., disgust vs. negative vs. positive) we adopted the following planned comparison procedure. First, we established if the two control groups (i.e., negative vs. positive) differed. If they did, then each control group was compared to the disgust group. If the control groups did not differ, they were combined and then contrasted with the disgust group. Alpha was set at .017 (i.e., .05 divided by the three potential comparisons) for these follow-up analyses.
110
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
3. Results 3.1. Participant characteristics There were no significant differences (all ps N .12) in age, disgust sensitivity or gender distribution between Groups (see Table 1). 3.2. Induction effects on emotion ratings There were no significant differences for the disgust or negative score, between Groups (i.e., disgust vs. negative vs. positive emotion inductions) or by Session (neutral induction vs. emotion induction) for the first set of emotion ratings obtained at the start of each experimental session. Thus participants in all of the groups started each session in a similar emotional state. Following the emotion inductions, significant changes were observed for the disgust and negative emotion ratings. For disgust, the one-way ANOVA by Group revealed a significant main effect, F(2,92) = 46.94, MSE = 1.10, p b .001, partial eta-squared = .51, with the largest change in disgust between the pre and post-induction emotion ratings being in the disgust group (M = 2.5), followed by the negative group (M = 1.1) and then the positive group (M = 0.0). Planned contrasts revealed that all three changes significantly differed from each other (all ps b .001). Thus the study design achieved its basic objective of inducing greater disgust in the disgust group, relative to the two control groups. For the negative emotion score, a one-way ANOVA by Group revealed a significant main effect, F(2,92) = 16.85, MSE = 0.36, p b .001, partial eta-squared = .27, with the largest increase in negative emotion in the disgust group (M = 0.8) followed by the negative group (M = 0.5) and with a small reduction in the positive group (M = −0.2). Planned contrasts revealed that the disgust and negative group did not significantly differ, but that both of these groups significantly differed from the positive group (both ps b .001). Thus the study design achieved its second objective of inducing negative emotion in the negative control group of a similar magnitude to the negative emotion induced by the disgust induction in the disgust group, with a small decrease in negative emotion score in the positive induction group. 3.3. Average pain The primary analysis (see Fig. 1a, c and e for data) used MANOVA (due to marked violations of sphericity) with Group (Disgust vs. Negative vs. Positive emotion inductions) as a between factor, and Session type (Emotion induction vs. Control [neutral] induction), and Time (Pretest baseline vs. Immediate post-test vs. 15 min post induction vs. 30 min post induction) as within-factors. The MANOVA revealed two significant effects. The first was a main effect of Time, F(3,90) = 9.41, Wilks Lambda = .76, p b .001, partial eta-squared = .24, with average pain ratings tending to decrease across the successive tests within a session (see Fig. 1a, c and e for data). The second was an interaction between Group, Session type and Time, F(6,180) = 2.17, Wilks Lambda = .87, p b .05, partal eta-squared = .07, indicating differences in average pain between the emotion inductions and the neutral control
Table 1 Participant characteristics of the three emotion induction groups. Variable
Gender (M/F) Mean age (SD) DSS-Ra TDDSb a b
Emotion induction group Disgust
Negative
Positive
11/23 20.2 (3.4) 14.4 (4.6) 66.0 (21.1)
12/20 20.0 (3.4) 12.5 (4.7) 69.2 (21.4)
7/23 20.3 (3.3) 14.8 (4.5) 69.4 (20.5)
DSS-R = disgust sensitivity scale revised. TDDS = three domain disgust scale.
conditions, over the course of the multiple cold-pressor tests. A number of possible differences are evident in Fig. 1a, c and e. First, there is a general downward trend in average pain in all of the emotion induction conditions, with this being least apparent in the disgust induction. Second, performance on the neutral control condition also varies between these three groups (e.g., see disgust control). As group assignment was random, these latter differences on the neutral control likely reflect baseline differences in response to the pain task. The secondary analysis examined the likely basis of the Group, Session type and Time interaction effect. The secondary analysis used relative average pain reports (i.e., each emotion induction with its neutral control and baseline average pain score subtracted out) for the three post-induction time points (see Fig. 1b, d and f for the data used in this analysis). These were analysed in a two-way mixed design ANOVA, with Time (T1 Immediate, T2 15min, T3 30-min) as the within factor and Group (disgust vs. negative vs. positive) as the between factor. The ANOVA revealed an interaction between Group and Time, F(4,184) = 4.32, MSE = 18.95, p b .002, partial eta-squared = .09, which had a significant linear component, F(2,92) = 6.09, MSE = 25.15, p b .005, partial eta-squared = .12. This interaction is apparent when comparing Fig. 1b, d and f, as relative to baseline and the neutral control, the disgust induction results in an initial depression of average pain reports, which then increase over time, with the reverse pattern evident in the negative and positive induction groups. We then tested whether the two control groups differed by conducting a further two-way mixed design ANOVA with Time (as above), and Group (negative vs. positive). The ANOVA revealed a main effect of Time, F(2,118) = 5.14, MSE = 17.46, p b .01, partial etasquared = .08, with significant linear (p = .02) and marginal quadratic (p = .053) components. As can be seen in Fig. 1d and f, relative to baseline and the neutral control, average pain ratings decreased across Time, with most of this effect seemingly occurring between the post-test and the 15 min post induction cold-pressor tests. There were no effects involving Group, indicating a similar pattern of response in both control groups. As there was no difference between the control groups, we then compared these together against the disgust group in another two-way ANOVA of Time, and Group (disgust vs. controls combined). This revealed a significant interaction between Group and Time, F(2,186) = 6.85, MSE = 19.10 p b .001, partial eta-squared = .07, with a significant linear component (p = .002). There were no other significant effects. Thus, average pain fell immediately following the disgust induction and then increased across time (the linear trend being significant when tested alone, p = .05, but not the quadratic trend p = .80 for the disgust group), with the reverse pattern in the two control groups combined (linear, p = .021 and quadratic, p = .054, trends both being evident). 3.4. Pain, induction strength and disgust sensitivity in the disgust group We examined, in the disgust group, the relationship between disgust sensitivity and the change in disgust following the induction (relative to the neutral condition), and average pain responding across factor Time (i.e., the slope coefficient). There was a significant negative correlation between the average pain slope coefficient across Time and the change in disgust following the induction, r(32) = −0.35, p b .05. That is people who reported feeling considerable disgust following the induction tended to report less average pain than baseline control and they did so consistently across Time. In contrast those who felt less disgust following the induction also tended to report less average pain initially but greater average pain 30 min post-induction. There were no significant associations between the two individual difference measures and the average pain slope coefficient across Time, noting that both correlations were also negative. Finally, both the TDDS and DSS-R scores were positively correlated with the change in disgust score generated by the induction (rs respectively, .39 and .46).
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
4 Negative values r epr esent less pain and positive values mor e pain
A ver age pain intensity (M ax = 100)
35
30
25
Disgust Neutr al
20
15
Pre-Test
Post-Test
15 Mins
Negative values r epr esent less pain and positive values mor e pain
A ver age pain intensity (M ax = 100)
30
25
20
Pre-Test
Post-Test
15 Mins
2 1 0 -1 -2 -3 -4 -5 Post-test
15 Mins
30 M ins
3 2 1 0 -1 -2 -3 -4
Negative relative to baseline and neutral control
-5 -6
30 M ins
Post-test
15 Mins
30 M ins
4 Negative values r epr esent less pain and positive values mor e pain
35
A ver age pain intensity (M ax = 100)
Disgust rela tive to baseline and neutral control
4 Negative Neutr al
Positive Neutr al
30
25
20
15
3
-6
30 M ins
35
15
111
Pre-Test
Post-Test
15 Mins
30 M ins
Positive relative to baseline and neutral control
3 2 1 0 -1 -2 -3 -4 -5 -6
Post-test
15 Mins
30 M ins
Fig. 1. Top left a. Mean average pain (plus SEM) intensity by time for the disgust group and their neutral control data; Top right b. Mean average pain (plus SEM) intensity for the disgust group after subtracting out baseline pain and the response from their neutral control; Middle left c. Mean average pain (plus SEM) intensity by time for the negative group and their neutral control data; Middle right d. Mean average pain (plus SEM) intensity for the negative group after subtracting out baseline pain and the response from their neutral control; Lower left e. Mean average pain (plus SEM) intensity by time for the positive group and their neutral control data; Lower right f. Mean average pain (plus SEM) intensity for the positive group after subtracting out baseline pain and the response from their neutral control.
4. Discussion In this study participants were randomly assigned to receive one of three emotion inductions on one session (either disgust, negative or positive) and a neutral control induction on another. Self-report emotion ratings confirmed that the disgust induction induced significantly greater disgust than any other induction and that the negative induction induced a negative mood state that matched the intensity of the negative mood state induced by disgust. All inductions were preceded by a cold-pressor test employing multiple selfreport measures of pain intensity, with three more cold-pressor tests of the same design then being applied after the induction was complete (immediately and at 15 min and 30 min post-induction).
Three findings emerged. First, unlike the negative and positive emotion induction conditions, disgust resulted in a relatively immediate reduction in average pain intensity, relative to the baseline pre-test and the neutral control condition. Second, the disgust induction was also associated with a significant increase in average pain intensity as time elapsed from the induction. This contrasted with the positive and negative control conditions, which were associated with progressive reductions in pain sensitivity as time elapsed from the induction. Third, we found that the magnitude of participants' response to the disgust induction was related to their self-report of the inductions effectiveness. As we discuss below, this may reflect the operation of two processes, one leading to pain suppression and another to pain enhancement.
112
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
The principal motivation for this study was to see if disgust induced greater pain sensitivity, especially at around 30 min post-induction. This delayed effect was hypothesised based upon our previous studies, which suggested that cytokine and core body temperature changes appeared to be maximal at around 30 min post-induction [5,6]. One reason for this delayed effect is that it represents the biological time-course for up-regulation of cytokine production and thus also for the putative effects of cytokine release on body temperature [7,8]. Relative to the positive and negative emotion control inductions, disgust was unique in generating a significant increasing trend in pain sensitivity across trials. This can be seen in Fig. 1a, c and e, where the general trend is for a reduction in average pain intensity across successive trials, an effect which is least evident following the disgust induction. While we suggest that these findings provide some initial support for the idea that disgust can lead to elevated pain sensitivity, it is important to note that it could equally result from alterations in parasympathetic activity unique to disgust, which then differentially affects longer-term pain perception. We would note, however, that only the acute phase response account would seem to predict the direction of this longer-term effect. A further alternative also needs to be considered, namely whether differing heterogeneity among each emotion induction picture set could account for these results. It is possible that more rapid emotional habituation occurs for image sets that are more homogenous, which could then affect the magnitude of any physiological effect (arguably independent of any self-reported difference in emotional intensity, as these have not always been found to correlate with emotion-induced physiological effects; [e.g., see [4–6]]). As we did not measure image homogeneity, the precise impact of this factor could not be assessed, but it should be considered in future emotion inductions of the type employed here. A further potential limitation concerns the choice of control emotion inductions used in this design, namely the negative and positive emotion induction conditions. Fear and happiness might have made better comparison conditions, in that both are primary emotions like disgust. However, the problem we faced was in ensuring that we could appropriately induce each emotion in the laboratory. While movie clips provide a good means of generating fear and happiness, very few movie clips persistently induce disgust over a similar time course. In contrast, static images are a well-tested means of inducing disgust (as indicated by the many studies using this technique; [e.g., [3–6]]), but are less effective at generating happiness and especially fear. We selected static images as we knew these would effectively induce disgust and we used inductions of negative and positive affect rather than fear and happiness for the same reason ([e.g., [4–6]]). While we cannot say for certain what results would ensue from using fear and happiness as emotion control conditions, we would suggest that given the key features of fear are negative affect and arousal, and of happiness are positive affect and arousal, that there might not be major differences in outcome. Disgust produced an initial reduction in average pain intensity relative to the baseline pre-test and the neutral control. In contrast, Meagher et al. [11] found that a disgust induction increased the speed with which participants reported their first pain-related rating, suggesting increased pain sensitivity. Meagher et al,'s [11] findings seem broadly typical of the literature examining the impact of negative and positive emotions on pain perception [e.g., [19]], with augmentation of pain following negative inductions and pain suppression following positive inductions, although there are many exceptions [e.g., [20]]. Importantly, one potential confound in a comparison between our data and Meagher et al.'s [11], concerns the intensity of the pain induction, as we used a far less aversive stimulus. The level of pain used in a study has not, to our knowledge, been well explored in the context of the pain-emotion relationship and may constitute another important variable, alongside the intensity and valence of the emotion. In addition, and as we noted in the Introduction, it may be that the disgust-inducing stimuli, at least in the short term, generate an external orientation, which is sufficient to distract from the lower level of pain intensity present in this study.
We suggested earlier that the correlation with self-report changes in disgust and changes in average pain over the multiple cold-pressor tests might reflect the operation of two processes. One possibility is that the disgusting images lead to an initial suppression of pain in all participants, an effect that lasts longer in those who report finding the images especially disgusting. On this basis, the initially reduced reaction to pain following the disgust induction may reflect a judgmental process based upon the context e.g., [21], namely a comparison of the negativity of the pain stimuli relative to the negativity of the disgusting images. Equally, a distraction-type effect, as noted earlier, might also account for this as well. As this effect wanes – at varying speed depending upon how disgusting the images were found – the possible longer-term effect of disgust on the immune system might then start to become apparent, with an increase in pain sensitivity emerging. In conclusion, the findings here tentatively suggest that disgust can act to initially suppress and later augment average pain intensity on a modified repeat-series of cold-pressor tests. This contrasts with negative and positive emotional control groups, who show the reverse effect, with average pain intensity decreasing as time elapses since the induction. It is important to note that the study population for this experiment contained a high proportion of female participants, and while we observed no gender-related effects, this potentially limits the generalizability of our findings. In sum, we suggest that the findings for disgust, are supportive of the idea that this emotion can produce changes that seem to parallel aspects of the acute phase response, pointing to a connection between this emotion and disease-avoidance [1].
References [1] Oaten M, Stevenson R, Case T. Disgust as a disease avoidance mechanism. Psychol Bull 2009;135:303–21. [2] Curtis V, Aunger R, Rabie T. Evidence that disgust evolved to protect from risk of disease. Proc R Soc Lond Biol Sci B 2004;271:131–3. [3] Ersche K, Hagan C, Smith D, Abbot S, Jones S, Apergis-Schoute A, et al. Aberrant disgust responses and immune reactivity in cocaine dependent men. Biol Psychiatry 2014;75:140–7. [4] Schaller M, Miller G, Gervais W, Yager C, Chen E. Mere visual perception of others' disease symptoms facilitates a more aggressive immune response. Psychol Sci 2010;21:649–52. [5] Stevenson R, Hodgson D, Oaten M, Barouei J, Case T. The effect of disgust on oral immune function. Psychophysiology 2011;48:900–7. [6] Stevenson R, Hodgson D, Oaten M, Moussavi M, Langberg R, Case T, et al. Disgust elevates core body temperature and up-regulates certain oral immune markers. Brain Behav Immun 2012;26:1160–8. [7] Cray C, Zaias J, Altman N. Acute phase response in animals: a review. Comp Med 2009;59:517–26. [8] Baumann H, Gauldie J. The acute phase response. Immunol Today 1994;15:74–80. [9] Rozin P, Haidt J, McCauley C. Disgust. In: Lewis M, Haviland-Jones J, editors. Handbook of emotion. New York: Guilford Press; 2008. p. 757–77. [10] Case T, Oaten M, Stevenson R. Disgust and moral judgment. In: Mackenzie C, Langdon R, editors. Emotions, imagination and moral reasoning. Hove: Psychology Press; 2012. p. 195–219. [11] Meagher M, Arnau R, Rhudy J. Pain and emotion: effects of affective picture modulation. Psychosom Med 2001;63:79–90. [12] Levenson R. Autonomic nervous system differences among emotions. Psychol Sci 1992;3:23–7. [13] Kyle B, McNeil D. Autonomic arousal and experimentally induced pain: a critical review of the literature. Pain Res Manag 2014;19:159–67. [14] Eccleston C, Crombez G. Pain demands attention: a cognitive affective model of the interpretative function of pain. Psychopharmacol Bull 1999;125:356–66. [15] Ferreira-Valente M, Pais-Ribeiro J, Jensen M. Validity of four pain intensity rating scales. Pain 2011;152:2399–404. [16] Tybur J, Lieberman D, Griskevicius V. Microbes, mating and morality: individual differences in three functional domains of disgust. J Pers Soc Psychol 2009;97:103–22. [17] Olatunji B, Haidt J, McKay D, David B. Core, animal reminder and contamination disgust: three kinds of disgust with distinct personality, behavioural, physiological and clinical correlates. J Res Pers 2008;42:1243–59. [18] Lang P, Bradley M, Cuthbert B. International affective picture system (IAPS): instruction manual and affective ratings. Technical report A-5. The Center for Research in Psychophysiology, University of Florida; 2001. [19] Rhudy J, Bartley E, Williams A. Habituation, sensitization, and emotional valence modulation of pain responses. Pain 2010;148:320–7. [20] Hollin G, Derbyshire S. Cold pressor pain reduces phobic fear but fear does not reduce pain. J Pain 2009;10:1058–64. [21] Ward L. Remembrance of sounds past: memory and psychophysical scaling. J Exp Psychol Hum Percept Perform 1987;13:216–27.
Contents lists available at ScienceDirect
Physiology & Behavior journal homepage: www.elsevier.com/locate/phb
The effect of disgust on pain sensitivity Megan J. Oaten a, Richard J. Stevenson b,⁎, Trevor I. Case b a b
Department of Applied Psychology, Griffith University, Southport QLD4222, Australia Department of Psychology, Macquarie University, Sydney NSW2109, Australia
H I G H L I G H T S • • • • •
Disgust can induce effects similar to the acute phase response. Here we examine if it can also induce increased pain sensitivity. Immediately after a disgust induction pain was reduced, but later it increased. Negative and positive inductions produced the reverse outcome. We suggest that disgust may enhance pain sensitivity.
a r t i c l e
i n f o
Article history: Received 22 April 2014 Received in revised form 20 October 2014 Accepted 23 October 2014 Available online 29 October 2014 Keywords: Disgust Pain Emotion
a b s t r a c t Experiencing the emotion of disgust leads to delayed up-regulation of immune-related functions, increased corebody temperature and reduced appetite. These changes parallel those of the acute phase response, which occurs when a pathogen is detected by the immune system. Here we examined whether a further predicted aspect of the acute phase response is evident following disgust induction, namely increased pain sensitivity. Participants attended a two-session experiment. On one session they experienced an emotion induction (being randomly assigned to either disgust, negative or positive groups) and on the other they received a neutral control induction. Before and after each induction, and at 15 and 30 min post-induction, participants engaged in a cold-pressor task, rating pain intensity at 10 s intervals for 90 s on each occasion. Relative to neutral control and pre-test, average pain intensity decreased then increased across time following the disgust induction, with the reverse pattern in the negative and positive emotion inductions. These findings are the first to suggest that disgust may lead to an increase in pain sensitivity over a time course paralleling changes observed for core-body temperature and immune-related function, although the mechanisms underpinning these effects remain to be identified. © 2014 Elsevier Inc. All rights reserved.
1. Introduction It has been suggested that the emotion of disgust functions in humans as part of a larger system of defensive behaviours and physiological responses, which assist us in avoiding infectious disease [1]. This functional account of disgust arose primarily because of the close association between stimuli that elicit this emotion (e.g., corpses, faeces, wounds, rotten food) and their capacity to transmit pathogens [2]. This account has received additional support in recent years from the finding that exposure to disgust elicitors, as well as cues that remind participants of disease, can serve to activate the innate immune system [3–5]. For example, participants who viewed pictures of disgusting stimuli were found to have elevated levels of tumour-necrosis factor alpha (TNF-a) in their saliva around 30 min post-induction — an effect not observed in control conditions ⁎ Corresponding author. Tel.: +61 2 98508098; fax: +61 2 98508062. E-mail address: [email protected] (R.J. Stevenson).
http://dx.doi.org/10.1016/j.physbeh.2014.10.023 0031-9384/© 2014 Elsevier Inc. All rights reserved.
that use equally affectively negative stimuli [5,6]. Not only can disgust stimuli activate the innate immune system, this seems to extend to increasing core body temperature as well [6]. Viewing disgusting images, but not equally emotionally negative control images, resulted in progressive increases in core body temperature, which were maximal at around 30 min post-induction [6]. The temperature-related finding, and the immune system changes, led us to hypothesise that disgust might produce changes in the body that parallel the acute phase response. In the acute phase response, the immune system detects the presence of a pathogen and reacts to them both locally and systemically [7]. This systemic reaction includes fever, elevated levels of cytokines, tiredness and social withdrawal, loss of appetites (e.g., food), and increased pain sensitivity [8]. Paralleling this pattern of responding, disgust does appear to increase core body temperature [6], elevate levels of cytokines including TNF-a and IL-6 [3,5,6], and disgust is associated with feelings of nausea and reduced appetite for food [9,10]. In the study reported in this manuscript we focussed on whether a further parallel to the acute phase response is present, namely increased pain
108
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
sensitivity. To our knowledge only one study has investigated pain perception following experiencing the emotion of disgust, and this found that latency to report pain was increased immediately after induction of disgust, relative to a neutral control [11]. This study did not, however, examine whether any increased pain sensitivity occurred at later time points (i.e., around 30 min post-induction), which would parallel changes observed for temperature and immune-related effects. While we might hypothesise differences in pain sensitivity based upon disgust inducing an acute phase response, this is of course not the only way that experiencing this emotion might affect pain perception. Unlike the other basic negative emotions (i.e., fear and anger), disgust primarily activates the parasympathetic branch of the autonomic nervous system e.g., [12]. As autonomic function changes markedly during pain perception e.g., [13], it is likely that disgust-induced changes in parasympathetic activation would affect participants' experience of pain during the period that the emotion was being experienced. However, while some effect would be expected, it is not obvious what the precise nature of this impact would be nor its time course. A further emotion-related effect on pain perception concerns the degree to which they affect participants' orientation to the internal or external milieu. As the negative emotions generally deal with external threats, they may favour an outward shift of attention, thereby acting to reduce pain perception via distraction [14]. In this case, we would expect the effect to occur in close proximity to the experience of the emotion. To test whether disgust leads to increased pain sensitivity following longer delays – and to determine its effects on pain more broadly – we had participants randomly assigned to one of three experimental groups. Each group received an emotion induction on one day and a neutral control induction (viewing images of everyday household objects) on another, in counterbalanced order. This allowed us to ascertain the unique effects of each emotion induction on participants' perception of pain by measuring their baseline responding during the neutral control induction. The content of the emotion inductions differed between groups, with one set of participants receiving a disgust induction, one a negative induction (i.e., unpleasant, fear and anger provoking stimuli) and another a positive induction. This then allowed us to compare between groups, the unique effect of each emotion induction on pain perception. This between group manipulation was adopted so as to reduce the demands the study made upon participants (i.e., number of pain tests and experimental sessions). We included a negative emotion induction so that we could determine if any pain-related effect arose simply because disgust induces negative affect. A positive emotion induction was included to determine whether any form of valenced and arousing stimulation might account for any pain-related effect (e.g., via distraction). For the emotion inductions the requisite state was induced by showing participants particular sets of images. We established the success of the emotion inductions by having participants evaluate their emotional state before and after the induction. Pain sensitivity was tested using a variant of the cold pressor task, which is a well-established experimental technique for inducing pain, and from which reliable and valid selfreports of pain can be obtained e.g., [15]. On each day of testing participants' pain sensitivity was established, prior to the induction, and then three times afterwards; immediately, at 15-min and 30-min post induction. As we expected the most interesting alterations in pain sensitivity to emerge at the later time points (i.e., consistent with the immune and temperature related changes noted above), this necessitated multiple cold-pressor tests. Consequently, the temperature of the water was set at a warmer-level than normal. On each test participants were asked to keep their lower arm immersed in the cold water for 90 s, reporting their pain intensity at 10 s intervals. Finally, we asked all participants to complete individual difference measures of disgust sensitivity [16, 17] so that we could determine if any observed effects were stronger in participants who report experiencing this emotion with greater frequency and intensity.
2. Method 2.1. Participants Ninety-six undergraduate participants (31% male; M age = 20.2, SD = 3.3), all with no pre-existing medical condition that could affect pain responses, were randomly assigned to one of three experimental groups. Participants were recruited from the first-year psychology subject pool and from the university community, the latter being paid a small sum ($20) for taking part. Half of the sample self-reported as being Australians of Caucasian descent, with the remainder mainly composed of Australians of Asian descent. This proportion did not significantly differ between the experimental groups. Just over one-third of the sample was born overseas, and again this proportion did not significantly differ between the experimental groups. Each participant consented to take part in the experiment and the protocol was approved by the Macquarie University Human Research Ethics Committee. 2.2. Stimuli The pictorial stimuli used for the inductions were all obtained from the International Affective Picture Series (IAPS, [18]) and were shown twice in randomised order to participants on a 60 cm computer monitor. Disgust induction: IAPS disgust-related stimuli (e.g., vagrant, roaches, vomit), items; 1280, 2710, 2750, 3030, 3051, 3150, 3160, 3400, 7359, 7380, 9140, 9181, 9252, 9290, 9300, 9301, 9320, 9342, 9405, 9500. Negative induction: IAPS negatively-valenced stimuli (e.g., pointed guns, domestic violence, plane crash wreckage), items; 2141, 2455, 2800, 3180, 3500, 6230, 6300, 6311, 6313, 6315, 6510, 6571, 6830, 6838, 8485, 9041, 9050, 9421, 9611, 9910. Positive induction: IAPS positively-valenced stimuli (e.g., skydiving, white-water rafting, water skiing), items; 5480, 5621, 5950, 8030, 8034, 8080, 8160, 8178, 8179, 8180, 8185, 8186, 8191, 8192, 8200, 8260, 8300, 8341, 8400, 8490. Based upon the IAPS normative data, there was no significant difference in affective valence between the Disgust (M = 2.5, SD = 0.4) and Negative (M = 2.4, SD = 0.4) image sets, and both of these sets differed in affective valence from the Positive image set (M = 6.8, SD = 0.8; twosample t-tests, both ps b 0.01). For arousal, there was no significant difference on the IAPS normative data between the Disgust (M = 5.6, SD = 0.8) and the Negative (M = 5.9, SD = 0.8) image sets, and both of these sets were significantly less arousing than the Positive (M = 6.6, SD = 0.5; two-sample t-tests, both ps b 0.01) image set. The Neutral induction set (experienced by all participants) contained the following images: IAPS household objects (e.g., chair, rolling pin, book), items: 7000, 7002, 7004, 7006, 7009, 7010, 7020, 7025, 7030, 7035, 7050, 7080, 7090, 7150, 7175, 7179, 7211, 7217, 7233, 7235. Using the IAPS normative data, these images had a mean valence score of 4.9 (SD = 0.2) and a mean arousal score of 2.6 (SD = 0.5). Needless to say, both of these scores significantly differed from the mean valence and arousal scores of each of the emotion image induction sets (two-sample t-test; all ps b .01). Pain intensity was measured using a 100-point scale (anchors; 0 = No pain, 20 = Mild pain, 40 = Moderate pain, 60 = Moderately severe pain, 80 = Very severe pain, 100 = Worst possible pain). This scale was visible throughout testing and participants were asked to select a number between 0 and 100, which best reflected the degree of pain intensity that they were currently feeling. Emotion ratings were completed on seven point category scales (anchors 1 = Not at all to 7 Very). Participants were asked to rate how sad, angry, disgusted, tense, fearful and happy they were. These items were selected for rating as they encompass the principal states likely to be induced by the image sets. The cold-pressor task was conducted using a refrigerated circulating water bath (Lab Companion RW-2025G), with the temperature set at 13.5 °C. This temperature was based upon pilot work, so that
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
participants would be willing to undergo multiple repeated assessments on two different days. Prior to and just after each cold-pressor task was conducted, participants were asked to place the hand to be used (or that had just been used) inside a heated woollen glove for 90 s to ensure that skin temperature was similar before and after each task. The hand chosen was alternated across the four tests in each session, starting either with the dominant or non-dominant hand on each session, this being determined by the counterbalancing schedule. 2.3. Procedure Participants attended two one-hour experimental sessions oneweek apart. Each session was conducted at the same time of day and on the same day of the week for each participant. All testing conducted between 9 am and 4 pm, in an air-conditioned laboratory, with just the experimenter present. Prior to attendance participants were telephone screened to ensure that they did not have any current medical conditions that would preclude them from taking part (e.g., Raynaud's syndrome, cardiovascular complaints, drug abuse etc). On arrival, participants were asked to read (and if happy sign) the information and consent form, which described the experimental procedure in some detail but only outlined the study aims in a general way. On one session participants completed the emotion ratings, a baseline cold-pressor test (warm glove, cold-pressor and ratings, warm glove), a further set of emotion ratings, an induction, a third set of emotion ratings, and then a second cold-pressor test. This was followed by two-further cold-pressor tests, one at 15 min after the induction and another 30 min after the induction. On one of the sessions, all participants received the neutral induction. On the other session they either received the disgust, negative or positive induction dependent on the group to which they had been assigned. Completion of the two inductions was in counterbalanced order. At the end of the second session, participants were asked to complete two questionnaires to assess their disgust sensitivity — the Three-Domain Disgust Scale [16] and the Revised Disgust Sensitivity Scale [17]. All participants were debriefed regarding the specific aims of the study at the end of their second session. 2.4. Analysis Participants completed three sets of emotion ratings on each of their two test sessions. From each set we generated two key variables. The first was the disgust score, to determine if the disgust induction had worked. The second was the negative score, which was a composite of the negative emotions, sad, angry, tense, and fearful, with happiness reverse coded. These ratings were all collapsed together as they were all positively correlated (median r = 0.39), reducing the need for multiple significance tests. The two key variables from the first set of emotion ratings, which were obtained before any pain testing or emotion induction had taken place, were compared using ANOVA to check that participants did not initially differ by Group (i.e., disgust vs. negative vs. positive) or Session (i.e., neutral control induction vs. emotion induction). The disgust and negative scores obtained just before and just after the induction (the control neutral induction session data are not reported as there were no significant changes in ratings in this condition), were subtracted from each other to measure any induction-related changes. These data were then compared by ANOVA to check that disgust increased following the disgust induction, and that negative emotion increased following the negative induction, and fell in the positive induction. Gender and the counterbalancing order (i.e., neutral first vs. emotion first) were initially included in all of these analyses, but as neither generated any significant effects, so the analyses were repeated and reported without these variables. The participants, who did not attend both sessions, were excluded from these analyses. We used planned contrasts to determine whether each group differed following a significant omnibus F test.
109
The key measure of pain sensitivity was the average pain score. This was the average of all pain ratings obtained on a given cold-pressor test. This score was used as participants varied in how many ratings they made before reaching the agreed cut-off of 70, at which point participants were instructed to remove their arm. A key assumption with this approach is that the pain ratings changed in an approximately linear fashion across time on each cold-pressor test. This appears a reasonable assumption as we could find no significant differences in linear slope coefficients by emotion induction type (between groups) or when these were compared to the control session (within groups). Indeed, the best fitting model was linear when tested on various subsets of these data (i.e., recall that participants differed in how many pain ratings they gave on each trial). Each participant completed eight cold pressor tests and this resulted in eight average pain ratings — four for the emotion induction session (pretest baseline, immediate post-test, 15 min post induction, 30 min post induction) and four for the control session (pretest baseline, immediate post-test, 15 min post induction, 30 min post induction). For three cases, participants attended the emotion induction session but did not return for the neutral induction control session (1 case attended the neutral induction and did not return for the emotion induction). We chose to include these 3 cases, in the analyses, using their pre-test baseline average pain rating in lieu of their control session (noting that their omission does not change the study outcome). As there were highly significant violations of sphericity, these data were analysed using MANOVA with Group (between factor; Disgust, Negative, Positive emotion inductions), Session type (within factor; Emotion induction vs. Control induction), and Time (within factor; Pretest baseline vs. Immediate post-test vs. 15 min post induction vs. 30 min post induction) as factors. We also initially included Gender and Counterbalancing order, but as these two last-mentioned variables generated no significant effects, we repeated (and report) the analysis without these factors. For the second analysis we conducted the following additional modifications to the data set so as to isolate the unique effect of each emotion induction. For the emotion induction session, we subtracted its pretest baseline average pain rating, from the three post-induction scores. The same procedure was then followed for the control session. This left six scores composed of: (1) the emotion-induction scores three variables representing the change in average pain relative to baseline for the emotion induction session immediately after the induction (T1), at 15 min (T2) and 30 min (T3) post-induction; and (2) the control-induction scores - three variables representing the change in average pain relative to baseline for the control induction session immediately after the induction (T1), at 15 min (T2) and 30 min (T3) postinduction. Finally, we subtracted each respective control induction score (T1 to T3) from its corresponding emotion induction score (T1 to T3), leaving three target pain scores. These scores reflect the unique effect generated by each emotion induction, at T1, T2 and T3, relative to pre-test baseline and the neutral control condition. Negative values for the three target pain scores indicate a fall in average pain and positive values an increase in average pain, relative to pre-test baseline and the neutral control condition. These data were analysed with a mixed design ANOVA using Group (between factor; Disgust, Negative, Positive emotion inductions) and Time (within factor; T1, T2, T3), with Gender and Counterbalancing order initially included. As these two last-mentioned variables again generated no significant effects, we repeated (and report) the analysis without these factors. For significant effects involving Group (i.e., disgust vs. negative vs. positive) we adopted the following planned comparison procedure. First, we established if the two control groups (i.e., negative vs. positive) differed. If they did, then each control group was compared to the disgust group. If the control groups did not differ, they were combined and then contrasted with the disgust group. Alpha was set at .017 (i.e., .05 divided by the three potential comparisons) for these follow-up analyses.
110
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
3. Results 3.1. Participant characteristics There were no significant differences (all ps N .12) in age, disgust sensitivity or gender distribution between Groups (see Table 1). 3.2. Induction effects on emotion ratings There were no significant differences for the disgust or negative score, between Groups (i.e., disgust vs. negative vs. positive emotion inductions) or by Session (neutral induction vs. emotion induction) for the first set of emotion ratings obtained at the start of each experimental session. Thus participants in all of the groups started each session in a similar emotional state. Following the emotion inductions, significant changes were observed for the disgust and negative emotion ratings. For disgust, the one-way ANOVA by Group revealed a significant main effect, F(2,92) = 46.94, MSE = 1.10, p b .001, partial eta-squared = .51, with the largest change in disgust between the pre and post-induction emotion ratings being in the disgust group (M = 2.5), followed by the negative group (M = 1.1) and then the positive group (M = 0.0). Planned contrasts revealed that all three changes significantly differed from each other (all ps b .001). Thus the study design achieved its basic objective of inducing greater disgust in the disgust group, relative to the two control groups. For the negative emotion score, a one-way ANOVA by Group revealed a significant main effect, F(2,92) = 16.85, MSE = 0.36, p b .001, partial eta-squared = .27, with the largest increase in negative emotion in the disgust group (M = 0.8) followed by the negative group (M = 0.5) and with a small reduction in the positive group (M = −0.2). Planned contrasts revealed that the disgust and negative group did not significantly differ, but that both of these groups significantly differed from the positive group (both ps b .001). Thus the study design achieved its second objective of inducing negative emotion in the negative control group of a similar magnitude to the negative emotion induced by the disgust induction in the disgust group, with a small decrease in negative emotion score in the positive induction group. 3.3. Average pain The primary analysis (see Fig. 1a, c and e for data) used MANOVA (due to marked violations of sphericity) with Group (Disgust vs. Negative vs. Positive emotion inductions) as a between factor, and Session type (Emotion induction vs. Control [neutral] induction), and Time (Pretest baseline vs. Immediate post-test vs. 15 min post induction vs. 30 min post induction) as within-factors. The MANOVA revealed two significant effects. The first was a main effect of Time, F(3,90) = 9.41, Wilks Lambda = .76, p b .001, partial eta-squared = .24, with average pain ratings tending to decrease across the successive tests within a session (see Fig. 1a, c and e for data). The second was an interaction between Group, Session type and Time, F(6,180) = 2.17, Wilks Lambda = .87, p b .05, partal eta-squared = .07, indicating differences in average pain between the emotion inductions and the neutral control
Table 1 Participant characteristics of the three emotion induction groups. Variable
Gender (M/F) Mean age (SD) DSS-Ra TDDSb a b
Emotion induction group Disgust
Negative
Positive
11/23 20.2 (3.4) 14.4 (4.6) 66.0 (21.1)
12/20 20.0 (3.4) 12.5 (4.7) 69.2 (21.4)
7/23 20.3 (3.3) 14.8 (4.5) 69.4 (20.5)
DSS-R = disgust sensitivity scale revised. TDDS = three domain disgust scale.
conditions, over the course of the multiple cold-pressor tests. A number of possible differences are evident in Fig. 1a, c and e. First, there is a general downward trend in average pain in all of the emotion induction conditions, with this being least apparent in the disgust induction. Second, performance on the neutral control condition also varies between these three groups (e.g., see disgust control). As group assignment was random, these latter differences on the neutral control likely reflect baseline differences in response to the pain task. The secondary analysis examined the likely basis of the Group, Session type and Time interaction effect. The secondary analysis used relative average pain reports (i.e., each emotion induction with its neutral control and baseline average pain score subtracted out) for the three post-induction time points (see Fig. 1b, d and f for the data used in this analysis). These were analysed in a two-way mixed design ANOVA, with Time (T1 Immediate, T2 15min, T3 30-min) as the within factor and Group (disgust vs. negative vs. positive) as the between factor. The ANOVA revealed an interaction between Group and Time, F(4,184) = 4.32, MSE = 18.95, p b .002, partial eta-squared = .09, which had a significant linear component, F(2,92) = 6.09, MSE = 25.15, p b .005, partial eta-squared = .12. This interaction is apparent when comparing Fig. 1b, d and f, as relative to baseline and the neutral control, the disgust induction results in an initial depression of average pain reports, which then increase over time, with the reverse pattern evident in the negative and positive induction groups. We then tested whether the two control groups differed by conducting a further two-way mixed design ANOVA with Time (as above), and Group (negative vs. positive). The ANOVA revealed a main effect of Time, F(2,118) = 5.14, MSE = 17.46, p b .01, partial etasquared = .08, with significant linear (p = .02) and marginal quadratic (p = .053) components. As can be seen in Fig. 1d and f, relative to baseline and the neutral control, average pain ratings decreased across Time, with most of this effect seemingly occurring between the post-test and the 15 min post induction cold-pressor tests. There were no effects involving Group, indicating a similar pattern of response in both control groups. As there was no difference between the control groups, we then compared these together against the disgust group in another two-way ANOVA of Time, and Group (disgust vs. controls combined). This revealed a significant interaction between Group and Time, F(2,186) = 6.85, MSE = 19.10 p b .001, partial eta-squared = .07, with a significant linear component (p = .002). There were no other significant effects. Thus, average pain fell immediately following the disgust induction and then increased across time (the linear trend being significant when tested alone, p = .05, but not the quadratic trend p = .80 for the disgust group), with the reverse pattern in the two control groups combined (linear, p = .021 and quadratic, p = .054, trends both being evident). 3.4. Pain, induction strength and disgust sensitivity in the disgust group We examined, in the disgust group, the relationship between disgust sensitivity and the change in disgust following the induction (relative to the neutral condition), and average pain responding across factor Time (i.e., the slope coefficient). There was a significant negative correlation between the average pain slope coefficient across Time and the change in disgust following the induction, r(32) = −0.35, p b .05. That is people who reported feeling considerable disgust following the induction tended to report less average pain than baseline control and they did so consistently across Time. In contrast those who felt less disgust following the induction also tended to report less average pain initially but greater average pain 30 min post-induction. There were no significant associations between the two individual difference measures and the average pain slope coefficient across Time, noting that both correlations were also negative. Finally, both the TDDS and DSS-R scores were positively correlated with the change in disgust score generated by the induction (rs respectively, .39 and .46).
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
4 Negative values r epr esent less pain and positive values mor e pain
A ver age pain intensity (M ax = 100)
35
30
25
Disgust Neutr al
20
15
Pre-Test
Post-Test
15 Mins
Negative values r epr esent less pain and positive values mor e pain
A ver age pain intensity (M ax = 100)
30
25
20
Pre-Test
Post-Test
15 Mins
2 1 0 -1 -2 -3 -4 -5 Post-test
15 Mins
30 M ins
3 2 1 0 -1 -2 -3 -4
Negative relative to baseline and neutral control
-5 -6
30 M ins
Post-test
15 Mins
30 M ins
4 Negative values r epr esent less pain and positive values mor e pain
35
A ver age pain intensity (M ax = 100)
Disgust rela tive to baseline and neutral control
4 Negative Neutr al
Positive Neutr al
30
25
20
15
3
-6
30 M ins
35
15
111
Pre-Test
Post-Test
15 Mins
30 M ins
Positive relative to baseline and neutral control
3 2 1 0 -1 -2 -3 -4 -5 -6
Post-test
15 Mins
30 M ins
Fig. 1. Top left a. Mean average pain (plus SEM) intensity by time for the disgust group and their neutral control data; Top right b. Mean average pain (plus SEM) intensity for the disgust group after subtracting out baseline pain and the response from their neutral control; Middle left c. Mean average pain (plus SEM) intensity by time for the negative group and their neutral control data; Middle right d. Mean average pain (plus SEM) intensity for the negative group after subtracting out baseline pain and the response from their neutral control; Lower left e. Mean average pain (plus SEM) intensity by time for the positive group and their neutral control data; Lower right f. Mean average pain (plus SEM) intensity for the positive group after subtracting out baseline pain and the response from their neutral control.
4. Discussion In this study participants were randomly assigned to receive one of three emotion inductions on one session (either disgust, negative or positive) and a neutral control induction on another. Self-report emotion ratings confirmed that the disgust induction induced significantly greater disgust than any other induction and that the negative induction induced a negative mood state that matched the intensity of the negative mood state induced by disgust. All inductions were preceded by a cold-pressor test employing multiple selfreport measures of pain intensity, with three more cold-pressor tests of the same design then being applied after the induction was complete (immediately and at 15 min and 30 min post-induction).
Three findings emerged. First, unlike the negative and positive emotion induction conditions, disgust resulted in a relatively immediate reduction in average pain intensity, relative to the baseline pre-test and the neutral control condition. Second, the disgust induction was also associated with a significant increase in average pain intensity as time elapsed from the induction. This contrasted with the positive and negative control conditions, which were associated with progressive reductions in pain sensitivity as time elapsed from the induction. Third, we found that the magnitude of participants' response to the disgust induction was related to their self-report of the inductions effectiveness. As we discuss below, this may reflect the operation of two processes, one leading to pain suppression and another to pain enhancement.
112
M.J. Oaten et al. / Physiology & Behavior 138 (2015) 107–112
The principal motivation for this study was to see if disgust induced greater pain sensitivity, especially at around 30 min post-induction. This delayed effect was hypothesised based upon our previous studies, which suggested that cytokine and core body temperature changes appeared to be maximal at around 30 min post-induction [5,6]. One reason for this delayed effect is that it represents the biological time-course for up-regulation of cytokine production and thus also for the putative effects of cytokine release on body temperature [7,8]. Relative to the positive and negative emotion control inductions, disgust was unique in generating a significant increasing trend in pain sensitivity across trials. This can be seen in Fig. 1a, c and e, where the general trend is for a reduction in average pain intensity across successive trials, an effect which is least evident following the disgust induction. While we suggest that these findings provide some initial support for the idea that disgust can lead to elevated pain sensitivity, it is important to note that it could equally result from alterations in parasympathetic activity unique to disgust, which then differentially affects longer-term pain perception. We would note, however, that only the acute phase response account would seem to predict the direction of this longer-term effect. A further alternative also needs to be considered, namely whether differing heterogeneity among each emotion induction picture set could account for these results. It is possible that more rapid emotional habituation occurs for image sets that are more homogenous, which could then affect the magnitude of any physiological effect (arguably independent of any self-reported difference in emotional intensity, as these have not always been found to correlate with emotion-induced physiological effects; [e.g., see [4–6]]). As we did not measure image homogeneity, the precise impact of this factor could not be assessed, but it should be considered in future emotion inductions of the type employed here. A further potential limitation concerns the choice of control emotion inductions used in this design, namely the negative and positive emotion induction conditions. Fear and happiness might have made better comparison conditions, in that both are primary emotions like disgust. However, the problem we faced was in ensuring that we could appropriately induce each emotion in the laboratory. While movie clips provide a good means of generating fear and happiness, very few movie clips persistently induce disgust over a similar time course. In contrast, static images are a well-tested means of inducing disgust (as indicated by the many studies using this technique; [e.g., [3–6]]), but are less effective at generating happiness and especially fear. We selected static images as we knew these would effectively induce disgust and we used inductions of negative and positive affect rather than fear and happiness for the same reason ([e.g., [4–6]]). While we cannot say for certain what results would ensue from using fear and happiness as emotion control conditions, we would suggest that given the key features of fear are negative affect and arousal, and of happiness are positive affect and arousal, that there might not be major differences in outcome. Disgust produced an initial reduction in average pain intensity relative to the baseline pre-test and the neutral control. In contrast, Meagher et al. [11] found that a disgust induction increased the speed with which participants reported their first pain-related rating, suggesting increased pain sensitivity. Meagher et al,'s [11] findings seem broadly typical of the literature examining the impact of negative and positive emotions on pain perception [e.g., [19]], with augmentation of pain following negative inductions and pain suppression following positive inductions, although there are many exceptions [e.g., [20]]. Importantly, one potential confound in a comparison between our data and Meagher et al.'s [11], concerns the intensity of the pain induction, as we used a far less aversive stimulus. The level of pain used in a study has not, to our knowledge, been well explored in the context of the pain-emotion relationship and may constitute another important variable, alongside the intensity and valence of the emotion. In addition, and as we noted in the Introduction, it may be that the disgust-inducing stimuli, at least in the short term, generate an external orientation, which is sufficient to distract from the lower level of pain intensity present in this study.
We suggested earlier that the correlation with self-report changes in disgust and changes in average pain over the multiple cold-pressor tests might reflect the operation of two processes. One possibility is that the disgusting images lead to an initial suppression of pain in all participants, an effect that lasts longer in those who report finding the images especially disgusting. On this basis, the initially reduced reaction to pain following the disgust induction may reflect a judgmental process based upon the context e.g., [21], namely a comparison of the negativity of the pain stimuli relative to the negativity of the disgusting images. Equally, a distraction-type effect, as noted earlier, might also account for this as well. As this effect wanes – at varying speed depending upon how disgusting the images were found – the possible longer-term effect of disgust on the immune system might then start to become apparent, with an increase in pain sensitivity emerging. In conclusion, the findings here tentatively suggest that disgust can act to initially suppress and later augment average pain intensity on a modified repeat-series of cold-pressor tests. This contrasts with negative and positive emotional control groups, who show the reverse effect, with average pain intensity decreasing as time elapses since the induction. It is important to note that the study population for this experiment contained a high proportion of female participants, and while we observed no gender-related effects, this potentially limits the generalizability of our findings. In sum, we suggest that the findings for disgust, are supportive of the idea that this emotion can produce changes that seem to parallel aspects of the acute phase response, pointing to a connection between this emotion and disease-avoidance [1].
References [1] Oaten M, Stevenson R, Case T. Disgust as a disease avoidance mechanism. Psychol Bull 2009;135:303–21. [2] Curtis V, Aunger R, Rabie T. Evidence that disgust evolved to protect from risk of disease. Proc R Soc Lond Biol Sci B 2004;271:131–3. [3] Ersche K, Hagan C, Smith D, Abbot S, Jones S, Apergis-Schoute A, et al. Aberrant disgust responses and immune reactivity in cocaine dependent men. Biol Psychiatry 2014;75:140–7. [4] Schaller M, Miller G, Gervais W, Yager C, Chen E. Mere visual perception of others' disease symptoms facilitates a more aggressive immune response. Psychol Sci 2010;21:649–52. [5] Stevenson R, Hodgson D, Oaten M, Barouei J, Case T. The effect of disgust on oral immune function. Psychophysiology 2011;48:900–7. [6] Stevenson R, Hodgson D, Oaten M, Moussavi M, Langberg R, Case T, et al. Disgust elevates core body temperature and up-regulates certain oral immune markers. Brain Behav Immun 2012;26:1160–8. [7] Cray C, Zaias J, Altman N. Acute phase response in animals: a review. Comp Med 2009;59:517–26. [8] Baumann H, Gauldie J. The acute phase response. Immunol Today 1994;15:74–80. [9] Rozin P, Haidt J, McCauley C. Disgust. In: Lewis M, Haviland-Jones J, editors. Handbook of emotion. New York: Guilford Press; 2008. p. 757–77. [10] Case T, Oaten M, Stevenson R. Disgust and moral judgment. In: Mackenzie C, Langdon R, editors. Emotions, imagination and moral reasoning. Hove: Psychology Press; 2012. p. 195–219. [11] Meagher M, Arnau R, Rhudy J. Pain and emotion: effects of affective picture modulation. Psychosom Med 2001;63:79–90. [12] Levenson R. Autonomic nervous system differences among emotions. Psychol Sci 1992;3:23–7. [13] Kyle B, McNeil D. Autonomic arousal and experimentally induced pain: a critical review of the literature. Pain Res Manag 2014;19:159–67. [14] Eccleston C, Crombez G. Pain demands attention: a cognitive affective model of the interpretative function of pain. Psychopharmacol Bull 1999;125:356–66. [15] Ferreira-Valente M, Pais-Ribeiro J, Jensen M. Validity of four pain intensity rating scales. Pain 2011;152:2399–404. [16] Tybur J, Lieberman D, Griskevicius V. Microbes, mating and morality: individual differences in three functional domains of disgust. J Pers Soc Psychol 2009;97:103–22. [17] Olatunji B, Haidt J, McKay D, David B. Core, animal reminder and contamination disgust: three kinds of disgust with distinct personality, behavioural, physiological and clinical correlates. J Res Pers 2008;42:1243–59. [18] Lang P, Bradley M, Cuthbert B. International affective picture system (IAPS): instruction manual and affective ratings. Technical report A-5. The Center for Research in Psychophysiology, University of Florida; 2001. [19] Rhudy J, Bartley E, Williams A. Habituation, sensitization, and emotional valence modulation of pain responses. Pain 2010;148:320–7. [20] Hollin G, Derbyshire S. Cold pressor pain reduces phobic fear but fear does not reduce pain. J Pain 2009;10:1058–64. [21] Ward L. Remembrance of sounds past: memory and psychophysical scaling. J Exp Psychol Hum Percept Perform 1987;13:216–27.
