Do BSE and/or Salmonella evoke qualitatively unique rejections?


The actual dangers from BSE and Salmonella are not as great as apparently thought by most people. [While this was a valid statement at the time of data collection and when this presentation was made, our knowledge about what the actual risks of BSE may be has since greatly changed.] There is, as yet, no evidence for the transmission of BSE to humans, though circumstantial evidence is growing. However, if BSE can lead to human spongiform encephalitis, it will be fatal. Although Salmonella is easily caught by eating infected foodstuffs, Salmonella infection produces no more than mild illness for up to a week (except in immunosuppressed individuals). The risk of consuming infected foods of either type was always low. Yet both BSE and Salmonella produced dramatic scares.

In comparison, government warnings and media coverage about the effects of high fat and high salt diets go largely ignored. Clearly, there are many differences between the two cases which could be the cause of the different behaviour changes produced. Little systematic research on people's reactions to BSE and Salmonella, however, has been done.

Different patterns of food rejection:

Rozin & Fallon (1980; Fallon & Rozin, 1983) proposed a taxonomy of food rejections of 'distaste', 'disgust' and 'danger' reactions. 'Distaste' means that the food is unpalatable, but otherwise acceptable in that a very small amount of that food in an otherwise liked dish will not spoil that dish. Quinine produces a prototypical 'distaste' reaction. 'Disgust' is a more ubiquitous rejection: e.g., it can "contaminate", a small amount spoiling an otherwise liked dish. Novel meats (e.g., dog or beetle) evoke 'disgust' reactions. 'Danger' items are rejected because of their anticipated negative consequences, but they are not otherwise disliked and can be considered highly palatable.

In order to reveal such differences in the nature of a food rejection, it is necessary to have more than one measure of rejection. Rozin & Fallon (1980; Fallon & Rozin, 1983) utilised a battery - a Food Rejection Index (FRI) - of different questions, each probing a different element of the food rejection. I have used a modified version of this (see sample).

As a result of the dramatic food scares (or perhaps as a cause of them), it seemed possible that people's reactions to BSE and Salmonella might be different to the prototypical 'danger' pattern. Specifically, I thought that a 'disgust' rejection might be involved.


The data presented here were part of a larger study investigating learned food aversions. Experimental group with respect to this had no significant effects on how subjects performed on those parts of the experiment considered here.

28 foods were used, ordered pseudo-randomly. This poster will only concern some of these foods, listed below along with the descriptions read by each subject:

"BSE": steak, from a farm where a number of animals have been found to have bovine spongiform encephalitis (BSE or "mad cow disease").

"Salmonella": a soft-boiled egg, taken from a farm where a number of the chickens have been found to be infected with Salmonella.

"steak": steak.

"egg": soft-boiled egg. "food poisoning": a [fruit] that you know has been infected with bacteria that can cause food poisoning.

"carcinogen": a [fruit] that you know has been sprayed with a known, potent, tasteless carcinogen.

"cramps": a [fruit] that you know contains a tasteless and odourless toxin that will produce stomach cramps for about half an hour, but no other effects.

"fruit": a [fruit].

"disgust": a food or substance that you dislike because it is disgusting.

"mild distaste": a food that you mildly dislike because of the taste.

"[Fruit]" was one of apple, orange, banana or pear, selected to be moderately liked by each subject.

In all, 43 subjects completed the study. One was subsequently excluded as being possibly eating disordered. Only the last 23 completed FRIs for "food poisoning". Subjects were recruited through posters advertising the study (posted in a number of London colleges and in a GP surgery) and through a subject panel. Most of the subjects were university students, though there was a range of jobs. The subjects were not very representative of the general population. Subjects were also asked their sex, age, whether they had any children and their smoking status (non-smoker, occasional, regular).

Psychometric measures employed: State-Trait Anxiety Inventory or STAI (Spielberger, Gorsuch, Lushene, Vagg & Jacobs, 1983); Social Desirabiliyy (13-item form) or MCSD (Reynolds, 1982; Crowne & Marlowe, 1964); Food Neophobia Scale or FNS (Pliner & Hobden, 1992); Extraversion subscale of the Eysenck Personality Questionnaire or EPQ "E" (Eysenck, H. & Eysenck, S., 1975).

"Controlled" data: Subjects were asked both about "steak, from a farm where a number of animals have been found to have bovine spongiform encephalitis (BSE or "mad cow disease")" and about "steak". This allowed the scores for "steak" to be subtracted from those for "BSE" in order to control for subjects' feelings about steak per se, as opposed to their reaction to the threat of BSE. And likewise for "Salmonella" and "egg" and for "carcinogen", "food poisoning" and "cramps" and "fruit".

Generally, subjects scored at floor for the FRIs for "steak" and for "egg". Thus, controlled "BSE" and "Salmonella" scores differed little from the raw data. The exception was in vegetarians' reactions to "steak", which were near ceiling. Therefore, for vegetarians, the controlled scores for "BSE" were meaningless.

Analyses repeated with these controlled data were broadly similar to those using raw scores.


To analyze these results, repeated measures MANOVAs were performed. There are two within-subject factors: food for the different food items and FRI for the 13 different questions asked. A significant food × FRI interaction indicates a difference in the pattern of response to the different food items, i.e. a qualitative difference in the nature of the rejection. A significant food main effect, in the absence of a significant interaction, indicates a quantitative difference only. Significant FRI main effects are ubiquitous, merely indicating that subjects answered the different questions differently.

Given that the assumption of sphericity is typically violated in these analyses, multivariate statistics are quoted in preference to univariate ones. If higher-order significant interactions were found, lower-order significant effects are not reported.

Vegetarians were excluded as their reactions to "BSE" were confounded by their reactions to steak per se. With the two foods as "BSE" and "Salmonella", there is a significant food × FRI interaction (F(15, 12) = 2.62, p = 0.040). Visual analysis (see Figure 2) suggests that this difference is in the answers to questions 9 and 10. Subjects perceived the threat from Salmonella to be more immediate than that of BSE. The answers to these two questions were combined, giving an overall risk estimate, and the analysis repeated: this gave no significant interaction (p > 0.1), nor a significant main effect of food (p > 0.1). Subsequent analyses were performed with questions 9 and 10 combined.

Visual analysis (see Figure 2 & 3) shows that the reactions to "BSE" and "Salmonella" match the 'danger' pattern, not the 'disgust' or 'distaste' patterns. For example, if "BSE", "Salmonella" and "disgust" are compared, there is a significant food × FRI interaction: F(22, 16) = 5.93, p < 0.001.)

The pattern of responses to "BSE" and "Salmonella" were also compared to the different 'danger' foods used. Against "food poisoning", there was no significant interaction or main effect of food (ps > 0.1; given the reduced n for "food poisoning", it was not possible to calculate a multivariate F, but the uncorrected, univariate result was not significant.) Against "carcinogen", again there was no significant interaction or main effect of food (ps > 0.1). The small n for "food poisoning" also precluded a combined analysis for "BSE", "Salmonella", "food poisoning" and "carcinogen".

An analysis with "BSE", "Salmonella", "carcinogen" and "cramps" did produce a significant food × FRI interaction. n was too small for a multivariate value to be calculated: univariate F(792, 33) = 3.10; using the Huynh-Feldt epsilon correction ( = 0.47), p < 0.01.

A MANOVA was performed with a foodtype within-subjects factor (i.e., steak versus soft-boiled egg) and an infected within-subjects factor (i.e. "BSE" and "Salmonella" versus "steak" and "egg") on the responses to question 2 only. Question 2 serves as a measure of palatability. Neither the interaction nor the food type main effect was significant (ps > 0.1). There was a significant infected main effect (F(25, 1) = 12.31, p = 0.002) with "BSE" and "Salmonella" liked less.

For all of the 'danger' foods, the threat affects palatability ratings (as measured by question 2). "BSE" scored significantly higher than "steak", as did "Salmonella" than "egg". This was also true for "carcinogen", "food poisoning" or "cramps" versus "fruit" (analyses not shown). In reality, the toxic and non-toxic versions of these foods should not taste any different.


It would seem that BSE and Salmonella are not perceived in any different way to other types of food-related health threats. The hypothesis that BSE and/or Salmonella produce different reactions to 'danger' is not supported. The dramatic nature of these food scares cannot be explained in terms of the nature of the rejection evoked. There are clearly many other factors implicated, such as the vicious upward circle of media coverage or specific health fears relating to mental health or the immediacy of the threat.

The fact that palatability ratings were influenced by threat-related information suggests that palatability is not as stable as sometimes thought.

To Part 2

Return to abstract/contents