Psychophysiology, 48 (2011), 900–907. Wiley Periodicals, Inc. Printed in the USA. Copyright r 2010 Society for Psychophysiological Research DOI: 10.1111/j.1469-8986.2010.01165.x The effect of disgust on oral immune function RICHARD J. STEVENSON,a DEBORAH HODGSON,b MEGAN J. OATEN,a JAVAD BAROUEI,b and TREVOR I. CASEa a Department of Psychology, Macquarie University, Sydney, Australia Department of Psychology, University of Newcastle, Sydney, Australia b Abstract Disgust motivates avoidance of pathogen sources, but whether its role in disease avoidance extends into activating the immune system is unexplored. This was tested here by comparing oral immune markers before and after a disgust induction, relative to neutral and negative induction control groups. The disgust group, but not controls, revealed an oral inflammatory response, with increased salivary tumor necrotizing factor alpha and albumin, as well as a downregulation of immunoglobulin A (SIgA) secretion. It has been hypothesized that disgust evolved in animals to clear toxins from the oral cavity by gaping and increased salivary flow. Our data suggest down-regulated SIgA secretion may be a vestige of this response so as to conserve protein, while the inflammatory reaction may reflect an adaptive response to disease threat, selectively triggered by disgust. The broader implications of these data for a discrete neuro-gutimmune axis are examined. Descriptors: Disgust, Immune, Gut, Emotion is secreted on all mucosal surfaces, is relatively straightforward to measure in saliva, and has both antibacterial and viral effects (Lamm, 1997). Acute stressors may rapidly increase salivary IgA (SIgA) secretion rates (Lamm, 1997; Ring et al., 1999; Spangler, 1997) and, as indicated by animal data, SIgA secretion rate can be modulated via the autonomic nervous system (Carpenter, Proctor, & Garrett, 2005). While it is currently unclear whether disgust primarily invokes either sympathetic or parasympathetic activity, it is not in doubt that this emotion generates some form of autonomic response (Rohrmann & Hopp, 2008). Thus the appropriate neural connections exist for this emotion to affect SIgA secretion rate in saliva. This account is complicated by a further observation, which makes SIgA all the more interesting. As noted above, the origin of disgust seems to lie in distaste. Experiencing distaste results in excessive salivation and gaping, as the organism voids the toxin from the mouth (Rozin et al., 2000). Under these ‘ancestral’ conditions, up-regulated production of SIgA would be maladaptive, because SIgA would be lost as saliva flushed out the toxin from the gaping mouth. Thus disgust may activate a vestigial SIgA conservation system, thereby decreasing its secretion rate. Consistent with this expectation, a previous study measured SIgA before, during. and after participants were exposed to a disgust elicitorFin this case a video of graphic surgical procedures (Bosch et al., 2001)Fan approach used to generate disgust in many other studies (e.g., Demaree et al., 2006; Rohrmann & Hopp, 2008). Although this paper makes no mention of disgustFnor was the presence of this emotion establishedFthe effect of this manipulation was to reduce SIgA secretion, relative to a control condition that viewed a non-arousing video and an active stress condition using a time-paced memory test (Bosch et al., 2001). Bosch et al. (2001) attributed the difference in SIgA Disgust probably evolved from a more basic system present in most terrestrial vertebratesFdistasteFwhich functions to void the mouth of toxins (Rozin, Haidt, & McCauley, 2000). The facial expression that accompanies distaste is almost identical to the facial expression that accompanies disgust, and it has been argued that disgust represents a broadening of distaste to cues beyond toxins (Rozin et al., 2000). This broadening of distaste into disgust may have occurred via social learning, such that observing some one gaping as if to vomit in response to a particular cue may lead to the observer feeling similarly disposed, and especially so if the cue approaches the mouth (Rozin, Nemeroff, Horowitz, Gordon, & Voet, 1995). A common feature of cues that elicit disgust is that they connote disease (Oaten, Stevenson, & Case, 2009). Disgust then may function to prevent oral incorporation of pathogens. Just as the immune system may become primed by cues relating to a ‘fight or flight’ response (i.e., acute stressors), so as to prepare the body for injury or attack (Maier & Watkins, 1998), disgust may also initiate some form of preparatory response involving the immune system. Such a preparatory response could engage a number of different immune components, but all of these are likely to be linked by their relationship to the digestive system, and especially the mouth, given the role of disgust in avoiding oral incorporation and its origin in distaste. One means of bolstering oral immune defence would be to up-regulate salivary immunoglobulins, notably immunoglobulin A (IgA). IgA The authors thank the Australian Research Council for their continued support. Address correspondence to Richard J. Stevenson, Department of Psychology, Macquarie University, Sydney, NSW2109, Australia. E-mail: dick.stevenson@mq.edu.au 900 901 Disgust and immunity secretion between the two stress conditions to active vs. passive coping, in that viewing the surgical video was more akin to the formerFan uncontrollable stressor. However, we suspect that the nature of the surgical video (i.e., its disgust-inducing features) may be more relevant. If correct, then exposing participants to a disgust induction and a negative emotion induction, although both involve ‘passive coping,’ we would expect reduced SIgA secretion rates only for disgust. This formed our first hypothesis. While SIgA down-regulation may be unique to distaste and disgust, this would offer little preparation for or protection against any actual oral contact with a pathogen. Another means of counteracting a local oral disease threat would be to initiate some form of inflammatory response. The pro-inflammatory cytokines, such as interleukin-1B (IL-1B), interleukin-2 (IL-2), interleukin-8 (IL-8), and tumor necrotizing factor alpha (TNFa), are generally released in response to tissue injury or the presence of foreign antigens. In humans, acute psychological stress can affect the circulating levels of certain pro-inflammatory cytokines and related markers (i.e., unstimulated), and can prime the release of these agents in response to an antigen (i.e., stimulated; Steptoe, Willemsen, Owen, Flower, & Mohamed-Ali, 2001). Of the pro-inflammatory markers studied, meta-analysis indicates that in response to an acute stressor only C-reactive protein and the cytokine IL-1B are up-regulated, but TNF-a remains unaffected (Steptoe, Hamer, & Chida, 2007). This makes TNF-a particularly interesting as a marker here for two reasons. First, pragmatically, any change in TNF-a is unlikely to be due to acute stress from viewing disgusting images. Second, TNF-a is the most potent pro-inflammatory cytokine and is clearly produced during microbial attack (Pavlov & Tracey, 2004). If disgust does act to prepare the oral immune system for attack, then generating a pro-inflammatory response with this agent would be adaptive, but only if its production was delayed until after any voiding had taken place. If, then, a pro-inflammatory response were mounted, a further consequence would be an increase in membrane permeability, leading to the leakage of serum into the oral cavity (Kerkar et al., 2006; Ma et al., 2004; McCarthy et al., 1998). This would act to elevate levels of salivary albumin, in parallel with any increase in TNF-a. For these reasons, we tested whether disgust would result in elevated levels of TNF-a and salivary albumin. This formed our second hypothesis. To test these hypotheses, we randomly assigned healthy young males (only men were used to reduce variability) to one of three experimental conditionsFviewing disgusting images, viewing unpleasant images, and a neutral image condition. Saliva samples were obtained 1 week prior to the main experiment to familiarize participants with the procedure and to provide a reliability measure with a later collection episode immediately prior to viewing the images. Samples were also obtained straight after viewing and 10 min later. In addition, salivary cortisol was also measured to detect activation of the hypothalamic-pituitaryadrenal axis (HPA), which could underpin any observed effect on immune system markers. Although we did not entertain any particular hypothesis in regard to how disgust might influence salivary cortisol, prior work has suggested that cortisol may increase after viewing unpleasant pictures (Codispoti et al., 2003). Method Participants Male participants were recruited via advertisements, and from the psychology subject pool. Contact was made by telephone, allowing us to screen out participants taking medications, with any medical or psychiatric condition, illicit drug use in the last month, current smokers, and those aged 30 or above. One hundred and fifty-four prospective participants were then invited to attend the first experimental session. As well as repeating all of the telephone screen questions, participants were also administered the 21-question version of the Depression, Anxiety and Stress Scale (DASS-21; Lovibond & Lovibond, 1995) and the General Health Questionnaire (GHQ; Goldberg & Williams, 1978). Participants scoring higher than 25 on the DASS and 13 or more on the GHQ were excused, along with any participant who contradicted responses from the telephone screen. This left 92 participants who were then randomly assigned to one of three experimental conditions and invited to attend a second testing session. Participants received $20 Australian for taking part. Stimuli Pictorial stimuli used in the emotion manipulation were obtained solely from the International Affective Picture Series (IAPS; Lang, Bradley, & Cuthbert, 2001) so as to enable replication with the same image sets. The disgust and negative picture sets were selected to match for arousal and valence, based upon the IAPS male college age normative data (Disgust images mean valence 5 2.4/9, mean arousal 5 5.5/9; Negative images mean valence 5 2.4/9, mean arousal 5 5.9/9). Each of these sets consisted of 20 color images, which are listed here by their IAPS number along with a brief description: Disgust SetF7380, roach on pizza; 9006, HIV tattoo; 9181, dead animal; 9252, corpse; 9301, dirty toilet; 9320, vomit; 9342, pollution; 9405, sliced hand; 9432, mastectomy; 9810, KKK rally; 3030, mutilation; 3160, eye disease; 3400, severed hand; 9000, cemetery; 2710, drug addict; 2750, vagrant; 9300, dirty toilet; 9500, slaughterhouse; 9561, sick animal; 1280, rat: Negative SetF2141, widow in mourning; 6313, attack; 6300, knife attack; 6311, distressed female; 6315, battered woman; 6510, attack; 8485, car fire; 9041, scared child; 9050, plane crash; 9340, garbage; 9520, dirty kids; 9560, bird in oil; 9611, plane crash; 9800, skin heads; 9830, cigarette butts; 9910, car accident; 2800, sad child; 3180, battered woman; 3500, attack; 6230, aimed gun. The control pictures were selected for their inability to arouse emotion, as they all had near neutral valence (M 5 4.9/9) and little arousal value (M 5 2.4/9): Control setF7000, rolling pin; 7002, towel; 7009, mug; 7010, basket; 7004, spoon; 7006, bowl; 7020, fan; 7025, stool; 7030, iron; 7035, mug; 7050, hair dryer; 7090, book; 7150, umbrella; 7179, rug; 7233, plate; 7235, chair; 7217, clothes rack; 7175, lamp; 7080, fork; 7211, clock. Procedure Session 1. All testing took place between 09.30–11.45 am, and participants were instructed on the telephone screen to avoid caffeine on the test morning and to abstain from eating in the hour prior to testing. After participants had consented to participate (the study was approved by the Macquarie University Ethics Committee), the first session commenced. This served as both a face-to-face screening session, with a detailed biographical questionnaire, the GHQ, and the DASS, as well as the Disgust Sensitivity Questionnaire (DSQ; to check for any group differences on this variable; Haidt, McCauley, & Rozin, 1994) and as a training/control session as well. Once all of the questionnaires were completed, each participant was asked to sit quietly for 5 min. During this interval the screening questionnaire, GHQ, 902 and DASS were scored, and participants were excused if they did not meet the selection criteria (see above for details). Those who did meet the selection criteria were then asked to provide a saliva sample. Participants were instructed to sit quietly, not to swallow, and to passively drool their saliva (unstimulated) via a disposable 5 cm plastic straw into a pre-weighed sterile collection vessel (7.5 ml centrifuge tube). Each participant was timed from the moment the straw was placed into the collection vessel until approximately 2 ml of saliva had been obtained. Each collection vessel was then sealed and reweighed, before being frozen to 201C. Following the collection period, participants were reminded not to eat for at least 1 hour prior to their next experimental session and to avoid caffeine on the morning of the test. An appointment was then made for the same time and day on the following week. Session 2. Participants completed a brief screening questionnaire to establish when they last ate, when they last drank a caffeinated drink, whether they had been ill in the intervening week, and whether any stressful life event had occurred. They then sat quietly and provided the pre-manipulation saliva sample in the same manner as described above. Participants then received the emotion manipulation, which involved viewing an image on a computer monitor for 6 s, followed by a 3-s interstimulus interval in which they had to judge (starting from the third image presented) whether the image they had just seen was the same or different from the image presented two back. This ‘two back’ task served to ensure that all of the images were viewed. There was no significant difference in performance on this task between the three groups (univariate ANOVA  Group, Fo1). In total, 60 images were presented to each participant, these being drawn from the set of 20 images available in each particular set. Each of the 20 images appeared three times, and the order of presentation of the 60 images was randomized separately for each participant. Irrespective of how quickly a participant responded on the ‘two back’ task, the interstimulus interval was maintained at 3 s and the trial timed out if a response was not obtained by 5 s, in which case the next image was automatically presented. This task took approximately 10 min to complete. Immediately after the emotion manipulation, participants were asked to provide a second saliva sample in the same manner as described above. They then rated how they currently felt on seven 7-point category rating scales (anchors 1 ‘not at all’ to 7 ‘very’); Happy, Angry, Sad, Tense, Surprised, Disgusted, and Fearful. Ten minutes after the last saliva collection had ended, participants were asked to provide a third and final sample, collected in the same manner as above. Participants then completed the same emotion rating scales, and this concluded the experiment. Sample Analysis All sample analyses were conducted at the Laboratory of Neuroimmunology at the University of Newcastle, Australia. On the day of assay, samples were thawed completely, vortex mixed, and centrifuged at 3000 rpm for 15 min at 41C. Clear samples were then analyzed for secretory immunoglobulin A (SIgA; Salimetrics, State College, PA), secretory cortisol (Salimetrics), tumor necrotizing factor alpha (TNF-a; Invitrogen Corporation, Camarillo, CA) and albumin (Bethyl Laboratories, Inc., Montgomery, TX) concentrations using commercially available immunoassay kits. The optical densities were measured at wavelength 450 nm using a Multiskan EX microtiter plate R. J. Stevenson et al. reader (Thermo Electron Corp., Vantaa, Finland). Calculations were then performed using Ascent software version 2.6 (Thermo Electron Corp.). Statistical Analysis The principal approach used here was analysis of covariance (ANCOVA), with the pre-manipulation assay (first measure on Session 2) serving as the covariate, and the two post-manipulation measures as the dependent variables. This approach was adopted, as it is the most powerful method of detecting change in this type of design where group allocation is random (Van Breukelen, 2006; Van Breukelen & Van Dijk, 2007; Vickers, 2001). As indicated in more detail below, certain variables required transformation to meet normality assumptions, and the basic ANCOVA approach had to be modified in cases where the homogeneity of regression slopes assumption was not met. All reported within-participant effects utilize Greenhouse-Geisser adjusted values. Correlations between variables were established using Pearson’s r. Results Participant Data Of the 92 participants who completed the experiment, one’s cortisol data was excluded from the analysis as three of their four measures exceeded the sample mean by more than 15 standard deviations. A further participant’s TNF-a data was excluded from the analysis as one observation was 21 standard deviations above the sample mean. Demographic data, by group, is detailed in Table 1. There were no significant differences (univariate ANOVA  Group) for any of the demographic variables listed (all Fso2.1). Emotion Ratings Self-report ratings of emotion variables were obtained shortly after the manipulation had occurred (see Table 2). To determine whether any emotion was dominant within each Group, we conducted three separate one-way ANOVAs, one for each Group, with Emotion rating (the 7 measures, with happiness reverse coded) as the within-participant variable. For the Disgust group, there was an overall effect of Emotion, F(4.9,152.6) 5 21.84, MSE 5 1.33, po.01, partial eta-squared 5 0.41, indicating differences between ratings. Post-hoc Bonferroni adjusted contrasts revealed that only one emotion, disgust, significantly exceeded all six other emotions on these self-report ratings (all pso.0024). For the Negative group, there was also a main effect Table 1. Mean Values ( Standard Deviations) for the Demographic Variables Obtained by Experimental Group Variable Age BMI Hours since last meala GHQ DASS DSQ Disgust (n 5 32) 20.5 ( 23.9 ( 3.3 ( 7.3 ( 6.7 ( 23.9 ( 3.8) 3.4) 2.6) 2.4) 4.7) 2.9) Negative (n 5 30) 19.6 ( 23.1 ( 3.5 ( 8.3 ( 8.7 ( 24.0 ( 2.3) 2.8) 3.1) 2.8) 5.6) 2.9) Control (n 5 30) 19.0 ( 22.4 ( 3.4 ( 8.1 ( 7.9 ( 23.5 ( 1.9) 2.6) 3.0) 2.5) 4.7) 2.8) Note: BMI: Body Mass Index, GHQ: General Health Questionnaire, DASS: Depression, Anxiety and Stress Scale, DSQ: Disgust Sensitivity Score. a For Session 2. 903 Disgust and immunity Table 2. Mean Values ( Standard Deviations) for the Emotion and Arousal Ratings Obtained After the Manipulation and Then Again 101Min Later, by Experimental Group Time collected Variable Disgust (n 5 32) Immediate Happiness 4.2 ( 1.4) Anger 1.7 ( 1.0) Fear 2.3 ( 1.4) Sadness 2.8 ( 1.4) Surprise 3.4 ( 1.9) Tension 3.4 ( 1.8) Disgust 4.4 ( 1.7) Delayed (t 5 101min later) Happiness 4.7 ( 1.1) Anger 1.5 ( 0.8) Fear 1.6 ( 1.0) Sadness 1.9 ( 1.1) Surprise 2.1 ( 1.5) Tension 2.3 ( 1.5) Disgust 2.7 ( 1.4) Negative (n 5 30) Control (n 5 30) 4.0 ( 1.7 ( 1.9 ( 3.2 ( 2.7 ( 3.2 ( 2.9 ( 1.3) 0.9) 1.0) 1.6) 1.6) 1.7) 1.6) 4.8 ( 1.5 ( 1.7 ( 1.6 ( 2.2 ( 2.5 ( 1.5 ( 1.4) 1.3) 1.1) 1.1) 1.5) 1.3) 1.0) 4.5 ( 1.5 ( 1.5 ( 2.0 ( 1.8 ( 2.2 ( 1.7 ( 1.2) 0.9) 0.8) 1.3) 1.5) 1.2) 1.1) 4.9 ( 1.4 ( 1.5 ( 1.5 ( 2.0 ( 2.0 ( 1.6 ( 1.4) 1.0) 0.9) 0.9) 0.9) 1.0) 1.2) of Emotion, F(4.6,134.5) 5 8.76, MSE 5 1.61, po.01, partial eta-squared 5 0.23. In this case, Bonferroni adjusted post hoc contrasts revealed no single dominant emotion (i.e., one different from all of the others). Sadness, tension, and happiness (reverse coded) significantly exceeded anger and fear, and surprise and disgust significantly exceeded anger (all pso.0024). For the Control group, there was also a main effect of Emotion, F(4.3,124.9) 5 5.67, MSE 5 1.28, po.01, partial etasquared 5 0.16. Again, Bonferroni adjusted post hoc contrasts revealed no single dominant emotion. Here, tension significantly exceeded anger, fear, sadness, and disgust (all pso.0024). We then tested whether the negative emotions elicited in the Negative and Disgust groups exceeded that of the Control group, and whether the level of reported disgust in the Disgust group exceeded that of the other negative emotions (i.e., testing for selectivity of emotion) relative to the Negative and Control groups. Using the negative emotions identified as relevant in the first analysis above (i.e., sadness, tension, happiness [reverse coded], and surprise) combined as one dependent variable, with disgust as the other (within-factor Emotion type), we conducted a two-way ANOVA with Group (Disgust group vs. Negative group vs. Control group) as the between factor. The ANOVA revealed a main effect of Group, F(2,89) 5 18.66, MSE 5 2.98, po.01, partial eta-squared 5 0.30, with all negative emotions significantly higher (Bonferroni adjusted post hoc contrasts, all pso.017) in the Disgust group (M 5 3.7), relative to the other two groups, and with the Negative group significantly higher (M 5 3.0) than the Control group (M 5 1.9). There was also a significant interaction of Group and Emotion type, F(2,89) 5 21.52, MSE 5 0.70, po.01, partial eta-squared 5 0.33. Post hoc Bonferroni adjusted contrasts (all pso.017) for the difference between negative emotion and disgust rating scores (disgust minus negative emotion) revealed significantly higher disgust scores in the Disgust group (M 5 1.3), relative to the Negative group (M 5 0.1) and the Control group (M 5 0.6). The latter pair did not significantly differ. Three conclusions can be drawn from this analysis and the one above. First, that disgust was the dominant negative emotion reported by the Disgust group. Second, that relative to the other negative emotions aroused by the manipulations here, only the Disgust group revealed differential activation of disgust when compared to the other two experimental groups. Third, that the Negative group did not experience any particular negative emotion but rather a blend of negative emotions, which was significantly more intense relative to the Control group. Although these findings indicate the expected pattern of outcome following the manipulation, it is important to note that, as no baseline emotional measures were obtained, it is possible, although unlikely, that differences in emotional state may have existed between groups prior to the manipulation. Finally, we examined whether any negative emotion was present at the second set of emotion ratings obtained just after the collection of the last saliva sample, by repeating the two-way ANOVA design (Group and Emotion type) described in the preceding paragraph. This ANOVA revealed a main effect of Group, F(2,89) 5 3.56, MSE 5 1.93, po.05, partial etasquared 5 0.07, with negative emotions overall only differing (Bonferroni adjusted post hoc contrasts, all pso.017) between the Disgust group (M 5 2.4) and the Control group (M 5 1.8), with neither of these two groups differing from the Negative group (M 5 1.9). There was also a significant interaction of Group and Emotion type, F(2,89) 5 7.86, MSE 5 0.49, po.01, partial etasquared 5 0.15. Post hoc Bonferroni adjusted contrasts (all pso.017) on the difference between negative emotion and disgust rating scores (disgust minus negative emotion) revealed significantly higher disgust scores in the Disgust group (M 5 0.5), relative to the Negative group (M 5 0.4) and the Control group (M 5 0.3), which did not significantly differ. These data suggest that the Disgust group were still reporting more disgust, relative to other negative emotions and the other two experimental groups, following the final saliva sample collection. SIgA Data Salivary IgA secretion rates expressed as micrograms per second (note that salivary secretion rates in grams per second did not differ between groups at any collection point) are presented in Figure 1, by group. These data were initially analyzed by twoway ANCOVA, with Time of collection (immediate vs. delayed measure) as a within factor and Group (Control vs. Negative vs. Figure 1. Mean ( Standard Error) salivary IgA (mg/s), by group, across the three saliva collection points of the study. 904 Disgust) as a between factor and secretion rate prior to the manipulation as the covariate. This analysis revealed violations in the assumption of homogeneity of regression slopes, with a stronger correlation between the pre-manipulation SIgA measure (the covariate) and the immediate measure, relative to the delayed measure. To test for a main effect of Group, and for a Group  Time interaction, without violating the homogeneity of regression slopes assumption, each of these effects was tested individually, using averaged or subtracted Time variables, as described below. To test for a main effect of Group, the mean of the two Time of collection measures were analyzed in a one-way ANCOVA, with Group as the between factor and secretion rate prior to the manipulation as the covariate. The ANCOVA revealed a main effect of Group, F(2,88) 5 3.59, MSE 5 1.53, po.05, partial etasquared 5 0.08. Difference contrasts revealed no significant difference in SIgA secretion rates between the Negative (M 5 3.1) and Control (M 5 3.2) groups, but these groups combined had significantly higher secretion rates than the Disgust group (M 5 2.4; po.01)Fafter adjusting for baseline differences (see Figure 1). To test for an interaction between Group and Time, the one-way ANCOVA was repeated but this time using the difference between the immediate and delayed collection measure. There were no significant effects on this analysis. TNF-a and Albumin Data Salivary TNF-a concentration (picograms per milliliter) and albumin concentration (nanograms per milliliter) were positively correlated at all collection time-points (rs40.55) and so these two measures were examined together within the same analysis. Prior to analysis, both of these variables were square-root transformed so as to meet normality assumptions for parametric testing (see Figure 2). These data were initially analyzed by a threeway ANCOVA, with Time of collection (immediate vs. delayed measure) and Measure (TNF-a vs. albumin) as within factors and Group (Control vs. Negative vs. Disgust) as a between factor, with the TNF-a and albumin pre-manipulation measures serving as covariates. This analysis revealed violations in the assumption of homogeneity of regression slopes involving Measure (with the TNF-a covariate correlating more strongly with the TNF-a measures than with albumin, and vice versa), but no violations involving Time  Measure, or Time alone. To test for a main effect of Group, and for any Group  Time, and Group  Time  Measure effects, without violating the as- R. J. Stevenson et al. sumption of homogeneity of regression slopes, two further analyses were conducted, which avoided inclusion of the specific violation related effect. The first ANCOVA examined for a main effect of Group using the mean of the immediate and delayed collection measures, with Group as a factor and the covariates as above. There was no significant effect of Group in this analysis. The second ANCOVA, using Measure and Group, tested for interactions with Time of collection and so used the difference of the immediate and delayed collection measures. This time there was a significant effect of Group, F(2,86) 5 3.10, MSE 5 0.57, po.05, partial eta-squared 5 0.07 (i.e., a Group  Time interaction in the context of the original ANCOVA design above). Difference contrasts revealed no significant difference between the Control (M 5 0.18) and the Negative (M 5 0.12) measures, but together these significantly differed from the Disgust group (M 5 0.14, p o .02) after adjusting for baseline differences. As is evident in Figure 2, TNF-a and albumin tend to increase between the immediate and delayed collection measures, with the reverse tendency in the Control and Negative groups. Cortisol Data Salivary cortisol (micrograms per deciliter) data is illustrated in Figure 3. These data were analyzed using a two-way ANCOVA, with Time and Group as the within and between factors respectively and the pre-manipulation measure as the covariate. This analysis revealed no significant main effects or interactions. Correlations Pearson correlations within each set of markers, obtained prior to and after the manipulation, were examined first. All cortisol (rs40.75), SIgA secretion rate (rs40.54), TNF-a (rs40.75) and albumin (rs40.78) measures were significantly intercorrelated across the three collection times on Session 2. Correlations between measures were sparse, except as noted above for albumin and TNF-a, which were significantly positively associated at each time interval. There were no significant associations between cortisol and albumin, nor between cortisol and TNF-a, but there were some weak negative correlations between cortisol and SIgA secretion rate, but only between the pre-manipulation SIgA secretion rate measure and each of the three cortisol measures (rs 5 0.19, 0.23, and 0.24, across time, respectively). SIgA secretion rates were not significantly associated with TNF-a or albumin measures. Figure 2. Mean ( Standard Error) salivary TNF-a and albumin (in unitsFsee text for details), by group (left panelFDisgust; center panelF Negative; right panelFControl), across the three saliva collection points of the study. Disgust and immunity Figure 3. Mean ( Standard Error) salivary cortisol (mg/dL), by group, across the three saliva collection points of the study. We then examined for correlations between the self-report emotion measures and the immune markers, which had revealed disgust-related changes. First, we tested the relationship between secretory SIgA and self-report disgust (using the difference score, disgust minus negative emotion, reported in the emotion rating analysis above). This disgust score was then correlated with the mean SIgA flowrate (immediate and delayed measure combined), with the pre-manipulation IgA flowrate partialled out. This relationship was significant (r(88) 5 0.21, po.05), indicating that lower post-manipulation SIgA secretion rates, after adjusting for baseline differences, are selectively associated with greater feelings of disgust. Next, we examined whether the ‘feeling disgust’ score was associated with the change in TNF-a and albumin between the immediate and delayed measure, controlling for the pre-manipulation TNF-a and albumin measure. In this case, there was no significant association with the ‘feeling disgust’ score. Reliability of Immune Markers Assays were completed for the Session 1 saliva samples so as to estimate reliability with the first sample obtained on Session 2. There were significant positive Pearson correlations for SIgA (r(91) 5 0.67, po.01), cortisol (r(90) 5 0.62, po.01), TNF-a (r(90) 5 0.63, po.01), and for albumin, although this was unsurprisingly a weaker relationship (r(91) 5 0.28, po.01). Discussion Participants who were exposed to disgust stimuli demonstrated a selective reduction in secretion of salivary IgA and an increase in the quantity of TNF-a and albumin, relative to the two control groups. Unlike Codispoti et al. (2003), we found no significant group differences in salivary cortisol. This may have arisen due to the greater sensitivity of Codispoti et al.’s (2003) within-participant design and/or from their use of a blood-based measure of cortisol. The SIgA findings here provide the first evidence that experiencing the emotion of disgust can lead to a selective downregulation of oral immune function. This observation is consistent with other research noted in the introduction, in which participants who viewed a graphic surgery film demonstrated a 905 similar reduction over an equivalent time period (Bosch et al., 2001). In addition, this study is the first to suggest that viewing disgusting images can lead to an up-regulation of a pro-inflammatory cytokine, TNF-a, and that this up-regulation is associated with the increased presence of albumin, which we suggest is a consequence of increased oral vascular permeability. Maier and Watkins (1998) suggest that the neuro-immune system responds to acute psychological stress in broadly the same way as it responds to microbial attack. This then raises the important question as to whether every acute negative emotional challengeFthese all being stressfulFgenerates a broadly similar immune response, or whether certain emotional challenges generate more targeted responses. Thus in the context of the present study, why should disgust produce an apparently unique immune response relative to other acute stressors? One reason to suspect that it should would be if there were a special inter-relationship between the immune system, gut, and brain. Conditioned food aversions seem to represent one example of this type of relationship, whereby microbial activity in the gut can evoke illness, nausea, and vomiting, an immune response, and the development of an aversion to the food that made one sick (Scalera, 2002). In mammals, being confronted with a food that has made one sick can evoke a disgust-like facial expression representing the most developed form of this type of learning, but food aversion conditioning can also be observed in many lower vertebrates, as well as in invertebrates such as mollusks and insects (Kim, Lee, & Han, 2007; Sugai et al., 2007). Digestive systems across a broad range of phyla are also richly endowed with a capacity to initiate immune responses to foods that contain pathogens or toxins (Daneman & Rescigno, 2009; Eberl & Lochner, 2009), and the immune-gut axis is highly interconnected with the central nervous system, especially in vertebrates, where this has been studied most extensively (Chen et al., 2004; Goehler et al., 1995; Rubio-Godoy, Aunger, & Curtis, 2007). In respect of the central nervous system, the insular cortex is of special importance, as it has been implicated as a putative neural substrate for disgust and for visceral responses including nausea (Jones, Ward, & Critchley, 2010), and, at least in rats, in mediating conditioned immune responses where the conditioned stimulus is a flavor (Ramirez-Amaya & Bermudez-Rattoni, 1999). In addition, the insular cortex has the requisite bi-directional links to the immune system, via the HPA and the nucleus of the solitary tract, and to the gut via the autonomic nervous system; moreover, the human insular is also selectively active during response to immunological challenge (Harrison et al., 2009). These findings suggest it may function as one putative central integrator for a brain-immune-gut axis. Finally, a recent article has reported an association between bradygastric activity (which may be a precursor to vomiting) and felt disgust, suggesting a further link between this emotion and ingestive behavior (Meissner, Muth, & Herbert, 2010). The basic argument, then, being advanced here is that just as there is a close interrelationship between acute responses to stress and the immune system, in the form of a preparation for possible injury from ‘fight or flight’ (Maier & Watkins, 1998), the literature reviewed above suggests a similar interrelationship between the brain, gut, and immune system. If this were correct, then a specific preparatory immune response, of the sort observed here, would be consistent with this broader framework. A further issue, and one particular to the data observed here, concerns the mechanisms that might account for changes in SIgA, TNF-a, and albumin. For salivary IgA, there is excellent 906 R. J. Stevenson et al. animal data to suggest rapid control of release by the autonomic nervous system, which could readily accomplish the change observed here (see the introduction). For TNF-a, the situation is more complicated because this cytokine appears to be synthesized de novo. Notably, the effect of the disgust manipulation on TNF-a is most evident 10 min after the post-manipulation saliva collection, and therefore approximately 25 min after viewing the first of the disgust-evoking images. At least three mechanisms have been proposed for in vivo increases in plasma TNF-a (Steptoe, Hamer, & Chida, 2007). These are leukocytosis, reduced plasma volume, and up-regulated synthesis of TNF-a. In the context of the oral cavity, a reduction of plasma volume is not particularly relevant. For de novo synthesis of TNF-a, data from human studies suggest that TNF-a mRNA up-regulation may be initiated immediately (Louis, Raue, Yang, Jemiolo, & Trappe, 2007), with elevated levels detectable within around 30 min of the stimulating episode (Vedder et al., 1999). Leukocytosis appears to have a much longer time span and thus does not appear such a promising candidate (Vedder et al., 1999). Disease-related cuesFpictures of sick peopleFmay also induce relatively rapid effects on other cytokines, namely IL-6 following in vitro LPS stimulation (Schaller, Miller, Gervais, Yager, & Chen, 2010). The rapid nature of these cytokine changesFhere in vivo or in Schaller et al.’s (2010) case in vitroFwould seem to suggest a role for the autonomic nervous system in initiating this up-regulation. In conclusion, this experiment is the first to focus on the consequences for oral immune function of experiencing the emotion of disgust. We observed a selective reduction in SIgA, and a delayed increase in TNF-a and albumin. These results are interpreted within a broader framework, in which disgust is seen as one facet of an interrelated neural-gut-immune network, focused on preparing for and countering pathogen threats relating to ingestion. REFERENCES Bosch, J., De Geus, E., Kelder, A., Veerman, E., Hoogstraten, J., & Amerongen, A. (2001). Differential effects of active versus passive coping on secretory immunity. Psychophysiology, 38, 836–846. Carpenter, G., Proctor, G., & Garrett, J. (2005). Preganglionic parasympathectomy decreases salivary SIgA secretion rates from the rat submandibular gland. Journal of Neuroimmunology, 160, 4–11. Chen, H., Lin, W., Wang, W., Shao, J., Yang, J., Wang, B., et al. (2004). Enhancement of antibody production and expression of c-Fos in the insular cortex in response to a conditioned stimulus after a single-trial learning paradigm. Behavioral Brain Research, 154, 557–565. Codispoti, M., Gerra, G., Montebarocci, O., Zaimovic, A., Raggi, M., & Baldaro, B. (2003). Emotional perception and neuroendocrine changes. Psychophysiology, 40, 863–868. Daneman, R., & Rescigno, M. (2009). The gut immune barrier and the blood-brain barrier: Are they so different? Immunity, 31, 722–735. Demaree, H., Schmeichel, B., Robinson, J., Pu, J., Everhart, D., & Berntson, G. (2006). Up- and down-regulating facial disgust: Affective, vagal, sympathetic, and respiratory consequences. Biological Psychology, 71, 90–99. Eberl, G., & Lochner, M. (2009). The development of intestinal lymphoid tissues at the interface of self and microbiota. Mucosal Immunology, 2, 478–485. Goehler, L., Busch, C., Tartaglia, N., Relton, J., Sisk, D., Maier, S., & Watkins, L. (1995). Blockade of cytokine-induced conditioned taste aversion by subdiaphragmatic vagotomyFfurther evidence for vagal mediation of immune-brain communication. Neuroscience Letters, 185, 163–166. Goldberg, D., & Williams, P. (1978). Manual of the General Health Questionnaire. Windsor, UK: NFER Publishers. Harrison, N., Brydon, L., Walker, C., Gray, M., Steptoe, A., Dolan, R., & Critchley, H. (2009). Neural origins of human sickness in interoceptive responses to inflammation. Biological Psychiatry, 66, 415–422. Haidt, J., McCauley, C., & Rozin, P. (1994). Individual differences in sensitivity to disgust: A scale sampling seven domains of disgust elicitors. Personality and Individual Differences, 16, 701–713. Jones, C. L., Ward, J., & Critchley, H. (2010). The neuropsychological impact of insular cortex lesions. Journal of Neurology, Neurosurgery and Psychiatry, 81, 611–618. Kerkar, S., Williams, M., Blocksom, J., Wilson, R., Tyburski, J., & Steffes, C. (2006). TNF-a and IL-1b increase pericyte/endothelial cell co-culture permeability. Journal of Surgical Research, 132, 40–45. Kim, Y., Lee, H., & Han, K. (2007). D-1 dopamine receptor dDA1 is required in the mushroom body neurons for aversive and appetitive learning in Drosophila. Journal of Neuroscience, 27, 7640–7647. Lamm, M. (1997). Interactions of antigens and antibodies at mucosal surfaces. Annual Review of Microbiology, 51, 311–340. Lang, P., Bradley, M., & Cuthbert, B. (2001). International affective picture system (IAPS): Instruction manual and affective ratings. Technical Report A-5, The Center for Research in Psychophysiology, University of Florida. Louis, E., Raue, U., Yang, Y., Jemiolo, B., & Trappe, S. (2007). Time course of proteolytic, cytokine and myostatin gene expression after acute exercise in human skeletal muscle. Journal of Applied Physiology, 103, 1744–1751. Lovibond, P., & Lovibond, S. (1995). The structure of negative emotional states: Comparison of the Depression Anxiety Stress Scales (DASS) with the Beck Depression and Anxiety Inventories. Behavior Research and Therapy, 33, 335–342. Ma, T., Iwamoto, G., Hoa, N., Akotia, V., Pedram, A., Boivin, M., & Said, H. (2004). TNF-a-induced increase in intestinal epithelial tight junction permeability requires NF-kB activation. American Journal of Physiology: Gastrointestinal and Liver Physiology, 286, G367–376. Maier, S., & Watkins, L. (1998). Cytokines for psychologists: Implications of bidirectional immune-to-brain communication for understanding behavior, mood and cognition. Psychological Review, 105, 83–107. McCarthy, E., Sharma, R., Sharma, M., Li, J., Ge, X., Dileepan, K., & Savin, V. (1998). TNF-a increases albumin permeability of isolated rat glomeruli through the generation of superoxide. Journal of the American Society of Nephrology, 9, 433–438. Meissner, K., Muth, E., & Herbert, B. (2010). Bradygastric activity of the stomach predicts disgust sensitivity and perceived disgust intensity. Biological Psychology, DOI:10.1016/j.biopsycho.2010.09.014. Oaten, M., Stevenson, R., & Case, T. (2009). Disgust as a disease avoidance mechanism. Psychological Bulletin, 135, 303–321. Pavlov, V., & Tracey, K. (2004). Neural regulators of innate immune responses and inflammation. Cellular and Molecular Life Sciences, 61, 2322–2331. Ramirez-Amaya, V., & Bermudez-Rattoni, F. (1999). Conditioned enhancement of antibody production is disrupted by insular cortex and amygdala but not hippocampal lesions. Brain, Behavior and Immunity, 13, 46–60. Ring, C., Carroll, D., Willemsen, G., Cooke, J., Ferraro, A., & Drayson, M. (1999). Secretory IgA and cardiovascular activity during mental arithmetic and paced breathing. Psychophysiology, 36, 602–609. Rohrmann, S., & Hopp, H. (2008). Cardiovascular indicators of disgust. International Journal of Psychophysiology, 68, 201–208. Rozin, P., Haidt, J., & McCauley, C. (2000). Disgust. In M. Lewis & J. Haviland-Jones (Eds.), Handbook of Emotions (pp. 637–653). New York: Guilford. Rozin, P., Nemeroff, C., Horowitz, M., Gordon, B., & Voet, W. (1995). The borders of the self: Contamination sensitivity and potency of the body apertures and other body parts. Journal of Research in Personality, 29, 318–340. Rubio-Godoy, M., Aunger, R., & Curtis, V. (2007). SerotoninFA link between disgust and immunity? Medical Hypotheses, 68, 61–66. Scalera, G. (2002). Effects of conditioned food aversions on nutritional behavior in humans. Nutritional Neuroscience, 5, 159–188. Schaller, M., Miller, G. E., Gervais, W. M., Yager, S., & Chen, E. (2010). Mere visual perception of others’ disease symptoms facilitates a more aggressive immune response. Psychological Science, 21, 649–652. 907 Disgust and immunity Spangler, G. (1997). Psychological and physiological responses during an exam and their relation to personality characteristics. Psychoneuroendocrinology, 22, 423–441. Steptoe, A., Hamer, M., & Chida, Y. (2007). The effects of acute psychological stress on circulating inflammatory factors in humans: A review and meta-analysis. Brain, Behavior and Immunity, 21, 901–912. Steptoe, A., Willemsen, G., Owen, N., Flower, L., & Mohamed-Ali, V. (2001). Acute mental stress elicits delayed increases in circulating inflammatory cytokine levels. Clinical Science, 101, 185–192. Sugai, R., Azami, S., Watanabe, T., Sadamoto, H., Kobayashi, S., Hatakeyama, D., et al. (2007). One-trial conditioned taste aversion in Lymnaea: Good and poor performers in long-term memory acquisition. Journal of Experimental Biology, 210, 1225–1237. Van Breukelen, G. (2006). ANCOVA versus change from baseline had more power in randomized studies and more bias in nonrandomized studies. Journal of Clinical Epidemiology, 59, 920–925. Van Breukelen, G., & Van Dijk, K. (2007). Use of covariates in randomized controlled trials. Journal of the International Neuropsychological Society, 13, 903–904. Vedder, H., Schreiber, W., Yassouridis, A., Gudewill, S., Galanos, C., & Pollmacher, T. (1999). Dose dependence of bacterial lipopolysaccharide (LPS) effects on peak response and time course of the immune-endocrine host response in humans. Inflammation Research, 48, 67–74. Vickers, A. (2001). The use of percentage change from baseline as an outcome in a controlled trial is statistically inefficient: A simulation study. BMC Medical Research Methodology, 1, 6–10. (Received August 12, 2010; Accepted November 10, 2010)