| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychology (S.D., W.W., K.S., I.V.D., P.E., O.V.) and Faculty of Medicine (H.V., B.N., K.V.), University of Leuven, Leuven, Belgium.
Address reprint requests to: Omer Van den Bergh, PhD, Department of Psychology, University of Leuven, Tiensestraat 102, B-3000 Leuven, Belgium. Email: omer.vandenbergh{at}psy.kuleuven.ac.be
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: Conditional odor stimuli (CS) were (diluted) ammonia and niaouli. One odor was mixed with 7.4% CO2-enriched air (unconditional stimulus) during 2-minute breathing trials (CS+ trial), and the other odor was presented with air (CS- trial). Three CS+ and three CS- trials were conducted in a semirandomized order (acquisition phase). The test phase involved one CS+-only (CS+ without CO2) and one CS- test trial, followed by three trials using new odors (butyric acid, acetic acid, and citric aroma). Half of the subjects (N = 28) were tested immediately, and the other half were tested after 1 week. Ventilatory responses were measured during and somatic symptoms were measured after each trial.
RESULTS: Participants had more symptoms in response to CS+-only exposures, but only when ammonia was used as the CS+. Also, generalization occurred: More symptoms were reported in response to butyric and acetic acid than to citric aroma and only in participants who had been conditioned. Both the selective conditioning and the generalization effect were mediated by negative affectivity of the participants. The delay between the acquisition and test phases had no effect.
CONCLUSIONS: Symptoms that occur in response to odorous substances can be learned and generalize to new substances, especially in persons with high negative affectivity. The findings further support the plausibility of a pavlovian perspective of multiple chemical sensitivity.
Key Words: classic conditioning • generalization, • psychosomatic symptoms • multiple chemical sensitivity.
Abbreviations: ANOVA = analysis of variance; CR = conditioned response; CS = conditioned stimulus; f = frequency per minute; MCS = multiple chemical sensitivity; NA = negative affectivity; NEM/PEM = Negative and Positive Emotionality Measure; UR = unconditioned response; US = unconditioned stimulus; VE = minute ventilation; VT = tidal volume.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Because experimental evidence is scarce, we developed a conditioning paradigm to test these assumptions in the laboratory. Harmless odorous substances (CSs) were presented in compound with CO2-enriched air (US) in 2-minute breathing trials. After three acquisition trials, the conditioning effects were tested by presenting the odors only. In a series of experiments, we observed increased respiratory frequency, and elevated levels of somatic symptoms on presenting the CS odor alone. This conditioning effect was selective: Symptoms and respiratory responses were learned only in response to a foul-smelling odor (ammonia) and not to a fresh-smelling one (niaouli) used as the CS (14, 20, 21). When both CS odors were foul smelling (irritant ammonia and nonirritant butyric acid), conditioned symptoms emerged to both, suggesting that the affective valence of the odors, not irritancy, is critical (22). This selectivity in conditioning effects occurred despite roughly equal conscious awareness of the arranged contingencies between CS and US for both types of odors (ammonia and niaouli), suggesting that the critical processes were not identical to aware cognition. The conditioning effects were overall stronger in a group of psychosomatic patients with hyperventilation-related symptoms than in normal subjects, in particular for the set of preexisting symptoms (20). In addition, conditioning effects were overall stronger in normal subjects scoring high on neuroticism (negative affectivity) than in normal subjects scoring low (21, 23). A straightforward pavlovian extinction procedure, involving a series of unreinforced exposures to CSs, readily reduced the learned symptoms (22). Conditioned symptoms and altered respiratory behavior were also observed when mental images of situations rather than odors were used as CSs: Merely evoking an image that was previously paired with a CO2 challenge elicited an increase in symptoms and altered respiratory behavior, but only when the imagined situations were stressful (24).
This set of studies provided strong experimental evidence supporting the plausibility of a pavlovian conditioning explanation of MCS. CO2 inhalation, used as an US in our paradigm, may act as a laboratory analog for two types of USs: First, it may mimic a toxic exposure; second, it may mimic the sensations of stress-induced hyperventilation. This latter argument may explain why a toxic exposure is not always found in a patient’s history and why stress may act as the major trigger of MCS in some cases (3). Recent findings have indeed documented the presence of hyperventilation in some cases (25). However, if episodic hyperventilation would act as an US subserving learning mechanisms for symptoms, it is apparent that physiological indicators of hyperventilation need not always be found during symptom episodes.
An intriguing aspect of MCS is that symptoms tend to spread to many, often chemically unrelated, substances. Patients respond to smoke, gasoline, and cleaning products but also to perfume. This diversity in symptom-triggering substances could be explained by the concept of generalization. In the conditioning literature, this phenomenon refers to "the tendency to respond to stimuli other than the training stimulus that has been associated with reinforcement" (26). Generally, the strength of the response tends to decline as test stimuli become increasingly different from the training stimulus (26–28). The similarity between the training stimulus and the test stimulus is usually described along some physical or psychological dimension, but the mere passage of time also may affect the tendency to generalize to other stimuli (26, 29). The mechanism behind this phenomenon is probably the forgetting of stimulus attributes, which, as a consequence, allows more different stimuli to elicit a response.
In the present study, we tested the occurrence of generalization of learned symptoms and respiratory responses to other substances than the one used in the acquisition phase. As in our previous studies, we used a differential conditioning procedure. This procedure involved use of two odors (ammonia and niaouli), one of which was used in one group of subjects as a CS+ (the odor was presented in compound with CO2-enriched air) and the other as a CS- (the odor was presented in regular room air). The specific combination was reversed in another group. In the test phase, we first tested the conditioning effect (one CS+ and one CS- only), followed by three new odors differing along the dimension of affective valence (fresh vs. foul smelling) and along the dimension of irritancy (irritant vs. nonirritant). For half of the subjects, the generalization effect was tested immediately after the acquisition phase; for the other half, the test phase was run after 1 week. This delay was introduced to stimulate forgetting of the stimulus attributes, which has been shown to modulate generalization effects (26). In line with our earlier findings, we expected a selective conditioning effect: Learned symptoms would occur only when foul-smelling ammonia was used as the CS+ and not when neutral- or positive-smelling niaouli was used. Because affective valence and not irritancy of the CSs seemed critical for conditioning effects to emerge (22), we expected generalization along the affective valence dimension: More symptoms are expected to occur in response to foul-smelling odors than to positive-smelling substances regardless of their irritant properties.
We also analyzed the impact of NA of the subjects. NA, or neuroticism, is defined as a broad range of aversive mood states, such as anger, disgust, fearfulness, and depression (30). High-NA subjects have more psychosomatic symptoms in general (31), which seems to be due to a greater interoceptive tendency and an inclination to interpret somatic sensations negatively (44). This variable seemed to modulate the conditioning effect in previous studies: Stronger effects were generally found in high-NA subjects (21, 44).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
On the basis of the NEM (score range = 0–14), 30 participants were categorized as having low NA (score <7) and 26 were categorized as having high NA (score 
7). The distribution of high- and low-NA participants was roughly equal across the cells made up by crossing the between-subject variables (CS+ odor and delay; seven of each in each cell), except when niaouli was used as the CS+ in the immediate condition, although there were slightly more low-NA participants (nine compared with five high-NA participants).
Materials
Odors and measures were identical to those used in other studies conducted by our group (14, 20–22). Only the most important and new features are described here.
Odors.
Ammonia (NH3) and niaouli (a volatile oil containing 65% eucalyptus) were used in the acquisition phase as CSs. In the test phase, three new odors were introduced: butyric acid, acetic acid, and citric aroma. Butyric acid and acetic acid are both foul smelling, the first being a pure odor and the latter an irritant. Citric aroma was experienced as a positively valent odor. All irritants were administered at levels below the irritancy threshold: Ammonia and acetic acid were dissolved in water in concentrations of 0.4 and 5%, respectively.
Subjective Measures.
Before the experiment, we administered (besides the brief health questionnaire) a symptom checklist measuring 39 daily life symptoms (32), the NEM/PEM (A. Tellegen, unpublished), the Positive and Negative Affect Schedule (33), and the Anxiety Sensitivity Index (34). The last three questionnaires measure NA. NEM/PEM scores were chosen on the basis of a comparison of the three measures on internal consistency and correlations with symptom scores (45). In addition, a median split of scores on the NEM/PEM and Positive and Negative Affect Schedule produced exactly the same composition of the two NA groups.
Symptoms after each trial were measured using the 16-item checklist of our previous studies. Participants indicated the extent to which the symptoms were experienced (not at all, slight, medium, strong, and very strong, coded from 1 to 5, respectively). The sum of symptoms was used as a continuous variable. The data of Wientjes and Grossman (32) were used to form five symptom subsets (arousal, respiratory, cardiac/warmth, tingling sensations, and unclassified). These subsets were analyzed separately. A sixth symptom subset, containing five items not sensitive to CO2 inhalation, was added as a control subset.
Physiological Measures and Apparatus.
All odors were administered through vaporization using a Devilbiss 646 nebulizer. Airflow was constant at a rate of 50 liter/h. The CO2-enriched air consisted of 7.5% CO2, 21% O2, and 71.5% N2. After decompression, the CO2-enriched air was led into a meteorological balloon connected to a valve, allowing easy switching between CO2-enriched air and regular room air. This valve was connected to a tube system ending in a double one-way valve, separating inspired and expired air. This double one-way valve was connected to a pneumotachograph (Fleish no. 2, Epalinges, Switzerland), which was fitted on a Rudolph mask. A small vinyl tube from the nebulizer was attached to the side of the mask, allowing the odor to be mixed with the inspired gas mixture. On the other side of the mask, another small vinyl tube was connected to an infrared CO2 monitor (Poet II, Criticare, Waukesha, WI), which sampled inspired and expired air (because of equipment failure, CO2 data were unreliable and are not discussed further). Airflow waveforms were sampled at a rate of 20 Hz using propriety software and stored on a PC. Offline, a software program (35) was used to count and remove pauses and irregularities from the data file and the following primary parameters were extracted per breathing cycle: inspiratory time, expiratory time, inspiratory volume, and expiratory volume.
Procedure
Participants were led into a waiting room to complete the questionnaires and to receive written information explaining the purpose and possible effects of the experiment: Several gas mixtures would be administered, and some of these mixtures could temporarily cause harmless symptoms like headache and shortness of breath. The symptoms would all disappear quickly after the trial. In addition, it was stated that the participants could stop the experiment at any time. A short description of the procedure was included. Next, participants signed an informed consent form and were led into the laboratory. All trials were 5 minutes in duration. The first 2 minutes of a trial consisted of breathing through the mask; during the remaining 3 minutes, participants rested and completed the symptom checklist.
The acquisition phase ( Figure 1) consisted of seven trials. Participants always started with a context exposure trial: Only room air and no odor was breathed through the system. Next, six trials were run in a semirandomized order. Three of them were CS+ trials: One odor (ammonia or niaouli) was mixed with CO2-enriched air. The other three were CS- trials: The other odor was administered with room air. The trials were presented in a semirandomized order with the constraint that no more than two trials in a row could be the same. Half of the participants received ammonia as CS+, and the other half received niaouli as CS+. Within each CS+ odor condition, half of the participants were randomly assigned to the immediate test condition, in which the test phase began after a 20-minute pause spent in the waiting room. The other half were assigned to the delayed condition, in which the test phase was run after 1 week.
|
Before and after the acquisition and test phases, all odors were presented in small containers. The participants lifted the lid and briefly inhaled the odor to rate the affective valence on a Likert-type scale from -5 to +5. Also, after the acquisition and test phases, the contingency awareness of the relation CS+/US was measured. Participants indicated which odor had caused the most symptoms in the previous phase. The answers were coded as follows: 0 = incorrect (eg, answering ammonia when niaouli had been the CS+), 2 = correct, and 1 = do not know.
Data Analysis and Design
Analyses on subjective measures were run on the total symptom score per trial and on the various subsets of symptoms separately (arousal, respiratory, cardiac/warmth, tingling sensations, unclassified, and dummy symptoms). The two context exposure trials were analyzed using the following design: CS+ Odor (ammonia, niaouli) by Time of Test (immediate, delayed) by NA (high, low) by Trial (one, two). The context data before acquisition and testing were used as covariates for analysis of the acquisition and test data, respectively. The acquisition trials were analyzed using the following design: CS+ Odor (ammonia, niaouli) by Time of Test (immediate, delayed) by NA (high, low) by Conditioning (CS+, CS-) by Trial (one, two, three). The test trials had a similar design except that there was no Trial variable. For the analysis of the generalization effect, the variable Conditioning was replaced by Generalization Odor (butyric acid, acetic acid, or citric aroma). CS+ Odor, Time of Test, and NA were between-subjects variables, whereas Conditioning and Generalization Odor were within-subject variables.
Several parameters were calculated from the raw physiological data: frequency per minute (f), tidal volume (VT), and minute ventilation (VE = VT x f). The analyses on means per trial were essentially the same as for the symptom score analysis, except that for each analysis an extra within-subject variable, Minute, was added because each trial consisted of 2 minutes of breathing. Because breathing behavior is different for men and women, we included Gender as an additional variable. However, this extra between-subjects variable caused an incomplete design for the distribution of NA. Therefore, the latter variable was left out of the analysis of the physiological data. This option was chosen because NA, although substantially related to conditioning of symptoms, was unrelated to conditioning of breathing behavior in our previous studies (22). Greenhouse-Geisser corrections were used when appropriate to control for violations of the compound symmetry ANOVA assumption (36).
Contingency awareness was analyzed using the Mann-Whitney U test. Odor evaluations were analyzed using the following ANOVA design: Moment (before or after acquisition, before or after test) by Odor (ammonia, niaouli, acetic acid, butyric acid, citric aroma).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Physiological Responses.
Women breathed faster with a lower VT and VE than men (F(1,45) = 4.16 for f, 17.71 for VT, and 5.89 for VE; all p values < .05 or smaller).
Unconditional Odor Effects.
These effects were tested using the data from participants who started the acquisition phase with a CS- trial (see Ref. 20). As in our previous studies, no indication of different effects of the odors themselves before they were used as CSs was found. This was so for both symptoms and respiratory behavior.
Acquisition
Subjective Symptoms.
Participants reported overall more symptoms when breathing CO2-enriched air than when breathing room air (Conditioning: F(1,48) = 83.31, p < .00001). This was found for all symptom subsets (see Table 1). The participants became gradually less aroused (Trial effect for the arousal set: (F(2,96) = 4.03, p < .05).
|
Physiological Responses.
Participants had a higher VT and VE on CS+ compared with CS- trials (F(1,46) = 73.11, p < .0001 and F(1,46) = 86.38, p < .0001, respectively). However, this difference was modulated by the type of CS+ odor: The frequency was higher only when ammonia was the CS+ and not when niaouli was the CS+ (CS+ Odor by Conditioning: F(1,46) = 4.57, p < .05). The participants also showed an increased f, VT, and VE from the first to the second minute during CS+ but not during CS- trials (Conditioning by Minute: F(1,46) = 14.72 for f, 84.85 for VT, and 154.79 for VE; p values < .001 or smaller).
Test of Conditioning Effects
Subjective Symptoms.
Replicating the selective conditioning effect found earlier (14, 20, 21), we observed more symptoms in response to the CS+ than to the CS- odor, but only when ammonia had been the CS+ odor (CS+ Odor by Conditioning: F(1,48) = 5.5, p < .05). In addition, this effect was modulated by NA: The selective conditioning effect was found in high-NA participants but not low-NA participants (CS+ Odor by NA by Conditioning: F(1,48) = 6.20, p < .05; CS+ Odor by Conditioning, high NA: F(1,48) = 10.37, p < .05; CS+ by Conditioning, low NA: F(1,48) = 0.19; see Figure 2).
|
Physiological Responses.
No (selective) conditioning effects were observed for respiratory behavior. A CS+ Odor-by-Conditioning-by-Minute interaction (F(1,41) = 5.05, p < .05) showed that the volume of the CS+ increased from the first to the second minute when ammonia was the CS+ odor; this increase did not occur with niaouli, but the difference between CS+ and CS- within this interaction was not significant. Overall, participants breathed faster in the immediate condition than in the delayed condition (Time of Test: F(1,40) = 8.11, p < .01), but this was only so during the second minute of measurement (Time of Test by Minute: F(1,41) = 5.11, p < .01).
Test of the Generalization Effects
Subjective Symptoms.
Overall, butyric acid made participants report more symptoms than acetic acid or citric aroma (Generalization Odor: F(2,96) = 8.96, p < .001). However, participants who received ammonia as the CS+ (ie, the group showing conditioning effects) reported more symptoms in response to butyric acid and acetic acid, but not citric aroma, than participants for whom niaouli had been the CS+ (those who did not show conditioning effects). However, this was only so when they were high in NA (CS+ Odor by NA by Generalization Odor: F(2,96) = 4.80, p < .05; butyric acid: F(1,47) = 7.27, p < .01; acetic acid: F(1,47) = 4.61, p < .05). Low-NA participants did not show this effect ( Figure 3).
|
|
Contingency Awareness and Odor Evaluations
Contingency Awareness.
To test whether participants showed any difference in knowing which odor had been presented with CO2-enriched air after the acquisition phase, Mann-Whitney U tests for several group comparisons were performed across the immediate and delayed test conditions, because this distinction was not relevant at that moment. The participants showed no better contingency awareness of the CS-US relationship with ammonia as the CS+ than they did with niaouli as the CS+. Also, high-NA participants were overall not more aware than low-NA participants. The data for ammonia as CS+ were as follows: correct = 17 (8 low NA, 9 high NA), don’t know = 8 (4 low NA, 4 high NA), and incorrect = 3 (2 low NA, 1 high NA). The data for niaouli as CS+ were as follows: correct = 17 (9 low NA, 8 high NA), don’t know = 7 (5 low NA, 2 high NA), and incorrect = 4 (2 low NA, 2 high NA). This means that participants in the condition in which symptom learning occurred (ammonia CS+) were overall not more aware than those in the condition in which no learning occurred (niaouli CS+). Therefore, our findings cannot be reduced to the effects of contingency awareness and/or demand characteristics.1
Odor Evaluations.
A main effect of Odor only (F(4,176) = 179.71, p < .0001) appeared. Ammonia, acetic acid, and butyric acid were all rated as negative odors. Ammonia and butyric acid were scored equally negatively (-3.48 and -3.31, respectively), whereas acetic acid was somewhat less negative (-1.36). Niaouli and citric aroma were rated positively (1.49 and 2.43, respectively), with citric aroma as most positive (post hoc Tukey Honestly Significant Difference test). Unlike in a previous study (22), no evaluative conditioning appeared (ie, a shift in evaluative judgment depending on the conditioning manipulation).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Despite the learning effects on symptoms, there were no significant conditioning effects on breathing responses. These effects have been less reliable, being found in some studies (14, 20, 22, 24) but not others (Ref. 21 and the present study). This is probably due to the use, for obvious ethical reasons, of a relatively weak US and a small number of learning trials. In any case, it corroborates our earlier findings (21), showing that automatically activated memory representations of the acquisition experiences were responsible for the learned symptoms and not specific informational input from bodily responses.
Most importantly in the present context is that the learned symptoms seemed to generalize to new odors: Only the participants showing conditioning effects also reported more symptoms in response to odors that had not been involved in the acquisition phase. Two aspects of this finding deserve particular attention. First, the generalization effect is confined to the negatively valenced odors, butyric and acetic acid, regardless of their potential irritancy. Overall then, it seems that the affective quality of the CSs is a critical cue for both the conditioning and generalization of symptoms. Second, the generalization effect was not modulated by the delay between acquisition and test. According to Riccio et al. (26), generalization occurs because of the forgetting of stimulus attributes: No or little generalization occurs when testing immediately follows acquisition, because stimuli attributes are "remembered" well. Introducing a delay between acquisition and testing often produces more generalization. From this perspective, the fact that we already found a generalization effect after 20 minutes that was not affected by adding a delay of 1 week may suggest either fast forgetting or poor learning about specific odor attributes. Because 34 of 56 participants correctly identified, right after acquisition, the CS+ odor among five odors presented in a row as the one that was associated with pronounced symptoms (15 did not know, and 7 were incorrect), it must be concluded that most participants clearly recognized the specific attributes. Moreover, the trials testing the conditioning effects (which were presented before the generalization odors) may be considered as "reminders" of the CS+ and CS- and, at the same time, serve as extinction trials. However, despite these possible counteracting effects, when an odor was presented 30 minutes later in a separate breathing trial, it seemed that the participants’ responses were merely guided by an association between a (vaguely specified) negative odor and symptoms. Most of the available findings on the effect of delay on generalization pertain to stimuli other than odors. It is currently not clear what forgetting about stimulus attributes actually means for this class of stimuli. Other parameter values, if not other processes than those known for other classes of stimuli, may be involved in the present generalization effect.
The conditioning effects and, as a consequence, the generalization effects were observed mainly in participants scoring high on NA. Other experimental studies in our laboratory, focusing specifically on the mechanisms mediating high NA and elevated symptom reporting during CO2 inhalation, have shown that high-NA subjects have a stronger attentional bias toward bodily sensations and are more inclined to contemplate possible negative health effects of the induced symptoms (23, 38). However, the effect of NA on symptoms seemed to depend on the attentional demands of the ongoing events: A new and/or a strong respiratory challenge wiped out the NA effects on symptoms, because in those circumstances both high and low NA were attentionally involved with processing the events (23). In conditions of minor challenges, such as breathing air (23), or when attentionally distracted during CO2 breathing (21), the NA effects on symptoms showed up (K. Stegen and O. Van den Bergh), probably because in those circumstances habitual differences in attentional direction among high and low NA show up. This may explain why in an earlier conditioning study of our group using distraction during acquisition (21), the conditioning effects reflected NA-related differences that were already present during the acquisition phase. In the present study, NA effects appeared only in the test phase and not during acquisition, because the absence of distractive elements allowed both high- and low-NA participants to attend directly to the events in acquisition.
The finding that the broader dimension of negative affectivity modulates the probability of both conditioning and generalization of symptoms to odors is particularly relevant for patients with MCS. They typically show quite high psychiatric comorbidity and stressful personal histories, pointing to the presence of high NA in those subjects (39).
Our results further support the plausibility of a pavlovian conditioning model for MCS: Acquired symptoms can generalize to new odors, not having been involved in the acquisition context. Negative affective valence of the odors seemed to be a critical dimension along which both learning and generalization occurred. However, clinical cases of symptoms triggered by positively valent odors like perfumes have been reported. Several reasons may account for this discrepancy. For example, we used a weak US between acquisition and testing. In real life, toxic exposures or episodes of hyperventilation may provide much more intense sensations as USs, which may affect the generalization gradient. Also, the delay was chosen rather arbitrarily and, compared with real-life situations, was relatively short. It is possible that longer delays and stronger USs will eventually trigger symptoms to stimuli that are qualitatively different from the original CSs. In addition, Davey (40) argued that cognitive processes may strongly modulate learning and generalization processes. The simple a priori conviction held by a person that "we live in a chemically polluted environment, which gradually will compromise everybody’s health" may potentially turn the evaluation of perfumes into a negatively valent stimulus. Such convictions may eventually help to spread generalization to other kinds of stimuli, such as foods. This is also relevant when considering possible treatments for patients with MCS. Based on the conditioning account of MCS, for example, systematic desensitization (a well-documented behavioral treatment technique that relies on extinction and counterconditioning principles) has been applied to MCS. Positive effects with this technique have been documented in a number of cases (41–43). Our results suggest that merely applying such behavioral techniques to the original CS may not be sufficient.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication September 16, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  P. J. Bulsing, M. A.M. Smeets, and M. A. Van den Hout The Implicit Association between Odors and Illness Chem Senses, February 1, 2009; 34(2): 111 - 119. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Fiedler, K. Kelly-McNeil, P. Ohman-Strickland, J. Zhang, J. Ottenweller, and H. M. Kipen Negative Affect and Chemical Intolerance as Risk Factors for Building-Related Symptoms: A Controlled Exposure Study Psychosom Med, February 1, 2008; 70(2): 254 - 262. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K Osterberg, R Persson, B Karlson, F C. Eek, and P Orbaek Personality, mental distress, and subjective health complaints among persons with environmental annoyance Human and Experimental Toxicology, March 1, 2007; 26(3): 231 - 241. [Abstract] [PDF]
|
|
|
|
 |
|
 M. Lopez-Grancha, C. Sanchez-Amate, M. Navarro, F. Carvajal, F. Sanchez-Santed, and I. Cubero Lateral Parabrachial Lesions Disrupt Paraoxon-Induced Conditioned Flavor Avoidance Toxicol. Sci., May 1, 2006; 91(1): 210 - 217. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. De Peuter, I. Van Diest, V. Lemaigre, W. Li, G. Verleden, M. Demedts, and O. Van den Bergh Can Subjective Asthma Symptoms Be Learned? Psychosom Med, May 1, 2005; 67(3): 454 - 461. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Saito, H. Kumano, K. Yoshiuchi, N. Kokubo, K. Ohashi, Y. Yamamoto, N. Shinohara, Y. Yanagisawa, K. Sakabe, M. Miyata, et al. Symptom Profile of Multiple Chemical Sensitivity in Actual Life Psychosom Med, March 1, 2005; 67(2): 318 - 325. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. Winters, S. Devriese, I. Van Diest, B. Nemery, H. Veulemans, P. Eelen, K. Van de Woestijne, and O. Van den Bergh Media Warnings About Environmental Pollution Facilitate the Acquisition of Symptoms in Response to Chemical Substances Psychosom Med, May 1, 2003; 65(3): 332 - 338. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
