Bugs are blech, butterflies are beautiful, but both are bad to bite: Admired animals are disgusting to eat but are themselves neither disgusting nor contaminating.

Acknowledgement: Both authors participated in all stages of this research, from conception to write-up.
The authors thank the Positive Psychology Center at the University of Pennsylvania for financial support of this research.

At least a billion humans enjoy consuming some types of insects (van Huis et al., 2013). In recent years, with the growing global problem of food insecurity, the possibility of widespread adoption of insects as a human food has been discussed within the entomology community (DeFoliart, 1999; Halloran, Flore, Vantomme, & Roos, 2018; Looy, Dunkel, & Wood, 2014; Paoletti & Dreon, 2005; van Huis et al., 2013; Yen, 2009). Insects are a source of high-quality protein, are very efficient at converting plant to animal calories, require relatively little water, do not emit significant levels of air pollutants, are easy to farm, occupy little space, are possible to raise and harvest under “natural” conditions, and are less likely than more commonly consumed animals to elicit concerns about animal cruelty.

The major reason that individuals reject insects as food appears to be that they elicit the emotion of disgust (Ruby, Rozin, & Chan, 2015). Indeed, insects are viewed so negatively that the “insect” category is the standard negative valence item in the IAT (Greenwald, McGhee, & Schwartz, 1998). In the course of our studies of resistance to consuming insects, we realized that there is one group of insects—butterflies—that themselves do not seem to elicit disgust, though we thought that eating butterflies would probably be disgusting. As we considered this matter, we thought that this possibility, if supported by data, might inform the understanding of disgust. One of the main challenges in understanding disgust is making sense of the broad range of elicitors for this emotion (Rozin, Haidt, & McCauley, 2017). According to our view, the elicitors could be organized into certain foods and animals (linked, at least in part, to protection from pathogens), reminders of our animal nature (most notably, death), contact with strangers, and contact with individuals or objects that represent a certain type of moral (divinity) offense (Rozin, Haidt, & McCauley, 1993, 2017; Rozin, Lowery, Imada, & Haidt, 1999). Another view, emerging out of evolutionary psychology, organizes disgust elicitors in terms of pathogen threat, threats to optimal mating, and moral offenses in general (Tybur, Lieberman, & Griskevicius, 2009; Tybur, Lieberman, Kurzban, & DeScioli, 2013). The evolutionary view has the virtue of alignment with fundamental threats to survival and reproduction, whereas our view calls on a broader range of functions. According to our view (and the Tybur et al., 2013, position), a powerful emotion originating in the food domain and perhaps as a protection against pathogens, may have been opportunistically used (by the process of preadaptation) in various cultures to encourage withdrawal from culturally undesirable entities. This view of disgust organizes the elicitors as one might organize what gets thrown into a waste basket—what organizes this diverse set of entities is simply a desire to throw them away (Rozin et al., 2017).

One of the basic features of disgust, a defining feature in our view, is that disgusting entities are contaminating: If they touch an otherwise favored entity, they render it unacceptable and disgusting (Rozin, Millman, & Nemeroff, 1986; Rozin & Nemeroff, 1990). This holds for a wide range of disgust elicitors, including “immoral” elicitors such as Adolf Hitler’s sweater. Of course, this contamination response is very much in keeping with the pathogen view—because pathogens can multiply, minimal contact with them can produce negative health outcomes.

The idea that butterflies may be disgusting to eat, without being disgusting or contaminating in and of themselves (unlike virtually all other disgusting entities) seemed to deserve attention. We (Rozin & Fallon, 1987) recognized that one group of disgusting-to-eat entities—admired animals like primates and dogs, close to humans in either form or relations to humans—were not disgusting in themselves. This issue had not been subsequently explored by either ourselves or other investigators and arose again in our work on insects.

We present our approach to this matter in two studies. In the first, we examine reactions to a range of insects, including butterflies, in terms of the disgust at the insect itself, disgust at consuming the insect, and reactions to eating desirable foods contacted by these insects. To add breadth to our conclusions, we surveyed both Indians and Americans. In a second study, with an American sample, we extended our inquiry to other admired (canaries, dogs, and koalas) and disgusting (vultures, hyenas, and rats) animals and used a more precise measurement of contamination potency.

The possibility that some admired animals are disgusting to eat, but not contaminating, challenges prior conceptions of disgust. Examination of these possibilities suggested to us that the disgust at eating admired entities might be a form of moral disgust, because in the act of ingestion, one is participating in the act of killing. Killing admired animals is immoral by any standards—therefore, what may look like what we call core disgust (which evolutionary psychologists call pathogen disgust) may in fact be a moral concern. On the other hand, if an admired animal comes into contact with a favorite food, this actually does no harm to the animal in question, so our inquiry about the absence of Contamination × Contact × Admired animals has moral implications. For these reasons, as a preliminary inquiry, we examined relations between perceived immorality of killing an animal and disgust at consuming it.

In this study, we first tested our presumption that insects are seen as generally negative and disgusting, with the exception of butterflies. We assessed this via free associations to the words insects and butterfly in Indian and American online samples. We then examined the ratings of six types of insects (including butterflies) on a variety of attributes, including willingness to touch, willingness to eat, immorality of killing, and ability to contaminate food.

Participants

Based on past research on failure to pass attention checks in U.S. and Indian adult samples recruited via Mechanical Turk, we elected to sample a larger number of Indians than Americans, with the aim of obtaining about 250 usable participants from each country, to have sufficient power to detect the moderate effects uncovered in similar past work (Ruby et al., 2015). We posted 300 participant slots in the United States and 375 in India on Mechanical Turk for a study on “attitudes toward food.” Participants were paid a modest sum for their time. Data collection stopped when all slots were filled. As some participants did not request reimbursement, we obtained a total of 692 participants—306 from the United States and 386 from India.

To ensure accuracy in responses, we excluded any participants who failed more than one of four catch questions, using a standard 5-point agree–disagree scale (173 from India, 31 from the United States). The catch questions were “I would rather eat a piece of fruit than a piece of paper,” “I regularly eat rocks,” “I enjoy eating plastic in my food,” and “The earth is a cube.” Disagreeing with the first question, or agreeing with any of the other questions, counted as a failure. To ensure more representative cross-cultural comparisons, we further excluded data from any participants who were not raised in their current country of residence (12 from India, 0 from the United States).

The final sample included 201 participants from India (34% women, Mage = 32.0, SDage = 9.72; 95% South Asian, 5% other; 57% omnivore, 16% partial vegetarian, 26% vegetarian/vegan; 74% Hindu, 10% Catholic, 10% Muslim, 6% other; 89% college degree; Indian language fluency: 60% Hindi, 49% Tamil, 29% Malayalam, 8% Telugu), and 275 participants from the United States (55% women; Mage = 35.9, SDage = 13.0; 85% White, 6% Black, 9% other; 88% omnivore, 6% partial vegetarian, 6% vegetarian/vegan; 39% Atheist/Agnostic, 25% Protestant, 19% Catholic, 17% other; 52% college degree).

Materials

The questionnaire included a series of sections, in fixed order. Many aspects of insect acceptance were probed, some of which are reported here, and others in Ruby and Rozin (2019). The subjects covered were (* indicates inclusion in the present study, with more details provided later) free associations*, measures of willingness to try insects, dietary intake, willingness to eat animals reared on different foods (including insects), food neophobia, core disgust sensitivity (Haidt et al., 1994), rating of six attributes of (or attitudes toward) six different insects*, attitudes to government involvement in promoting insects as food, beliefs about eating insects, and demographics*. The University of Pennsylvania Institutional Review Board approved the study protocol, which conformed to the Declaration of Helsinki.

The survey was in English for both samples. The first questions were the free associations. These were placed first to not be influenced by the other questions about insects. All questions were presented in the order indicated, each on a separate page. The questions on free associations were “What are the first 3 words that come to mind when you think about ‘insects’? What is the first word that comes to mind when you think of the word ‘chocolate’? What are the first 3 words that come to mind when you think of the word ‘butterfly’?”

The free association to chocolate was inserted as a break between “insects” and “butterfly.” Past research has indicated that the first free association gives very similar data to the combined results from three free associations (Rozin, Kurzer, & Cohen, 2002) and is less “contaminated” because the second free association is both to the target word and the first free association. Thus, the analyses we present here are based on the first free association to “insects” and “butterfly,” with the exception noted under results.

The attribute/belief ratings were presented as follows: “Using a scale of −100 to 100, where −100 = Disagree Completely, 0 = Neither Agree Nor Disagree, and 100 = Agree Completely, please indicate your agreement for the statements below as they pertain to the insects listed on the left.” The set of six “attributes” was presented one attribute at a time, each of which was rated with respect to six insects: cockroach, butterfly, mosquito, mealworm, ant, and cricket. The attribute/belief statements were as follows (our code word for each attribute is listed in parentheses before each item; these words were not presented to the respondent): (KILL IMMORAL) “Killing this insect is immoral”; (GOOD) “This insect does good for humans”; (HARM) “This insect does harm to humans”; (EAT) “I would be willing to eat this insect (whole & roasted)”; and (TOUCH) “I would touch this insect (live)”. For the sixth item (CRAWL), the rating scale was, “Using a scale of −100 = Extremely unwilling, 0 = Neutral, and 100 = Extremely willing, ‘How willing or unwilling would you be to eat your favorite food if one of the following insects walked/crawled over the food?’”

Given the large n and many dependent variables, we have adopted a significance level of p < .01, two-tailed for both studies.

Free associations

In tabulating word frequency, we only combined synonymous forms of the same word (e.g., scare/scary or beautiful/beauty). In coding, we used our judgments of positive and negative valence, discussing difficult categorizations among ourselves to arrive at a categorization—for example, we scored “colorful” as positive, but “color” as miscellaneous (neutral valence). For Americans, 111 (40%) of the first free associations to the word insects were “bug” or “bugs”, a synonym. In India, “bug” or “bugs” was mentioned only six times (3%). Because the synonym was not informative, in all of these cases we ignored the “bug” association and counted the second free association in its place.

After examining the full set of first free associations, we developed a simple coding system that encompassed almost all of the responses. The two categories of principal interest were positive and negative (with subcategories, see Table 1). The other categories were: activity/movement (6% of all free associations to “insects” and 6% to “butterfly”), name of a type of insect (25% of responses to “insects” and 11% to “butterfly”), insect parts (7% of responses to “insects” and 11% of responses to “butterfly”), and a miscellaneous category, which included ecology (where the insect is found), food of the insect, name of a noninsect animal, and attributes that were not obviously valenced (20% in the United States; 21% in India).

In the United States, the two most common free associations to “insects” were “gross” (11%) and “small” (8%), whereas in India, the two most common words were “small” (10%) and “bit/bite/biting” (6%). Because disgust is a major feature of reactions to insects, we report that there were 18% (50) “disgust-type” words (e.g., gross, disgust, yuck) in the United States, but only 4% (7) in India. If we examine the existence of disgust or a synonym in the second or third free association, the total rises to 31% (85) in the United States, and 17% (34) in India.

For “butterfly,” there was more consensus in both countries: in the United States, the top three words were pretty (20%), wings (15%), and beautiful/beauty (12%). The top three words in India were beautiful/beauty (49%), colorful (13%), and color (11%). There were no “disgust-type” free associations at all in either country. It is striking that although butterflies are clearly insects, and there was a “priming” for insects with the first set of associations, only 15% of American associations, and 3% of Indian associations, mentioned “insect” or a type of insect. Thus, membership in the category of “insects” does not appear to be a very salient feature of butterflies.

The most striking finding is a major difference in valence of the initial free associations, for both countries, between insects and butterflies. For insects, 41% of all associations in the United States, and 43% in India, were negative. Only 2% of associations were positive in both the United States and India. In stark contrast, there were no (0%) negative associations to butterfly in either country, but 41% positive in the United States, and 66% positive in India. Although the overall negativity to insects is about the same, among the Indians as opposed to Americans, there was more emphasis on harm in Indians (21% vs. 7% of associations, respectively) and less emphasis on disgust (3% vs. 18%). Overall, the free association data was similar across Indians and Americans, and reveals strong negativity to insects, and strong positivity to butterflies.

Attribute/action ratings

The mean scores for the six attribute/actions, for each of six insects, are displayed in Figure 1 (United States) and Figure 2 (India). Examination of both sets of results indicates a very robust finding: butterflies are the only insects considered good, and for whom killing is considered immoral (with one slight exception for ants in India). However, when it comes to ingestion (EAT), butterflies are about as negative as all of the other insects. This is a striking contrast—insects are bad, and disgusting (or at least very aversive) to eat, whereas butterflies are good, but disgusting (or at least very aversive), to eat. The goodness/beauty of butterflies is documented by both the free associations and the quantitatively higher ratings of the positive attributes (GOOD, TOUCH). For an overview, see Figures 1 and 2.

Repeated measures one-way analyses of variance (ANOVAs) with Bonferroni correction show that butterfly is significantly higher than all other insects at p < .001 for GOOD (d ranging from 0.66 to 1.98 in the United States, and from 0.82 to 2.74 in India), TOUCH (United States: d ranging from 0.54 to 1.74; India: d ranging from 0.76 to 2.18), and KILL IMMORAL ratings (United States.: d ranging from 0.57 to 1.52; India: d ranging from 0.38 to 1.05), and significantly lower for HARM (United States: d ranging from 0.45 to 3.10; India: d ranging from 1.28 to 3.42).

Insects as (and on) food

Reactions to eating the different species, and having them crawl across one’s food, are more nuanced. A repeated-measures one-way ANOVA with Bonferroni correction shows that for the Americans, eating ants is least aversive, and eating mosquitos and roaches is most aversive (see Table 2). Butterfly falls somewhere in the middle, significantly less aversive than mosquito and roach, but more aversive than ant, cricket, or mealworm. Overall, 14% of the Americans agree to being willing (EAT >0) to eat a butterfly, compared to 30% for ant and 4% for both mosquito and cockroach.

For the Indians, eating ants is least aversive (see Table 2), followed by everything else. EAT butterfly is significantly lower than EAT ant, although it is not significantly different from any other insect. As with the American data, EAT butterfly is above (but not significantly) only roach and mosquito. Overall, 4% of the Indians are willing to eat a butterfly, compared to 7% for ant and 4% for both mosquito and cockroach. It is possible that for the Indians, there is a floor effect that tends to equalize all the EAT scores.

The CRAWL item, which directly measures willingness to eat a favorite food after the designated insect crawled over it, is of particular relevance for the assessment of disgust. This is because the classic contamination situation that CRAWL represents is considered diagnostic of disgust in at least one major theory of disgust (e.g., Rozin et al., 1993, 2017; Rozin & Fallon, 1987). In the United States, CRAWL is higher for butterfly than all other insects (d = 0.31–1.56), and in India, higher for all other insects (d = 0.43–.90) except ant. Within the Indian sample, CRAWL is significantly higher for ant than for butterfly (p < .001, d = .21).

The CRAWL mean scores are compared to the EAT scores for both countries in Table 2. For all insects, except mealworm in the United States, willingness to eat one’s favorite food after an insect crawls over it is higher than willingness to eat the insect directly. However, these two items use different −100 to + 100 scales, so a direct comparison is not possible. Also, the food in question for the EAT item is a whole roasted insect, whereas for CRAWL, it is one’s favorite food, which may reduce or overcome an aversion induced by contact. However, the contamination literature typically uses highly desirable foods as the baseline, and brief contact with a cockroach almost invariably produces rejection (Rozin et al., 1986; Rozin & Nemeroff, 1990).

A majority (56%) of the Americans are willing (>0) to eat their favorite food after a butterfly crawls across it. In a sense, for these willing individuals, a butterfly does not qualify as disgusting; it is not contaminating. There are 108 (39%) Americans for whom eating butterflies is at least reasonably negative (<−10 score), but who are at least slightly willing (>0) to eat their favorite food after a butterfly has crawled over it.

The results for the Indians are similar, but less striking. Contact (CRAWL) with all insects leaves most respondents unwilling to consume the food, with ants (41% willing) less negative than butterflies (35% willing). There are 56 (28%) Indians for whom eating butterflies is at least reasonably negative (<−10), but who are at least somewhat willing (>0) to eat their favorite food after a butterfly has crawled over it. Unlike with Americans, where there is a much stronger rejection without contamination for butterflies than any other insect, for Indians, the response to ants is even more counter to the standard disgust-contamination link. For Indians, 62 participants show the aversion (<−10), but no contamination rejection for ants, as opposed to 56 for butterflies. No other insect approaches these values.

Disgust and the immorality of eating/killing butterflies

We next tested the idea that disgust at consuming butterflies arises, at least in part, from the fact that killing butterflies is judged to be immoral, as by definition, eating them involves killing them, directly or indirectly. Agreement (>0) that killing butterflies is immoral characterized 72% of Indians and 60% of Americans. Disgust may express this moral response. This moral-disgust hypothesis makes the following predictions, which we follow in each case by the relevant results. The data we used to test these hypotheses are the set of correlations between EAT and the other six attributes, for each of the six insects.

1. The lowest negative correlation (hence the strongest relation) between KILL IMMORAL and EAT will be for Butterfly. This is true for Americans, with this correlation at −.30 by far the lowest for any insect (next lowest value is −.06 for Cricket; median correlation −.04 for the six insects). However, this result does not hold for Indians, with the EAT/KILL IMMORAL correlation for Butterfly at .03 (close to the median of .08).
2. Because CRAWL, unlike EAT, does not harm the insect, Butterfly will show the smallest correlation between CRAWL and EAT. This is not quite the case for the Americans, with r = .25 (close to lowest value .22 for Mosquito; median correlation .36). The prediction is not clearly supported for Indians, with the correlation at .24 (tied for second smallest, with a median of .26).
3. On similar grounds, because TOUCH as well as CRAWL does not kill the insect, the TOUCH–EAT correlation will be lowest for Butterfly. This is the case for Americans, with r = .12 (next lowest .26 for Mosquito; median .35). For Indians, the TOUCH–EAT correlation is also lowest, with r = .01 (next lowest .11 for Mosquito; median .19).

Overall, the morality/killing hypothesis receives fairly good support for the set of tests with Americans, but only weak support with Indians.

The pattern of means we report, as displayed in Figures 1 and 2, shows a substantial difference between butterflies and the five other insects for Americans, with butterfly being by far most immoral to kill, doing most good for humans, and being least contaminating (smallest reduction in edibility of favored food after physical contact). In particular, butterflies are about as negative to eat as other insects, but much less contaminating. The results from Indians show some of the same relationships as the American data, but ants show some of the same properties as butterflies. In particular, although ants are reported to be less good and more harmful to humans than butterflies, and less immoral to kill than butterflies, ants are less contaminating than butterflies. We are surprised by this effect; in retrospect this could possibly result from our Indian participants having a greater familiarly with ants crawling over food. Killing butterflies is considered substantially more immoral than killing ants for both groups, but the linkage between this immorality and willingness to eat or tolerate “harmless” touching of crawling is much stronger for Americans than Indians. We cannot account for this difference.

The results of Study 1 confirmed our hypothesis that butterflies are regarded positively, unlike the other insects surveyed, but the negativity toward consuming them was about the same as that for the other insects. There seems to be something special about butterflies. The data suggested that butterflies are less contaminating than other insects for Americans, and about the same as ants and less than other insects) for Indians. To confirm and extend the finding from butterflies in Study 1, we changed some procedures in Study 2. To extend the finding from examples with a single positive animal (butterfly), we included four pairs of positive–negative animals: butterfly–cockroach, canary–vulture, dog–hyena, and koala–rat. Study 1 measured aversion instead of disgust; in Study 2, we explicitly measure disgust, as well as liking/aversion. The CRAWL manipulation, our measure of contamination in Study 1, is different from the usual measure we have used, which is very brief contact. In Study 2, we replaced the CRAWL item with “brief contact.”

A secondary, suggestive finding from Study 1 was that the killing of the butterfly by ingestion might add a moral dimension to the aversion/disgust to eating butterflies. The suggestive finding, present principally in the American respondents, was that for CRAWL and TOUCH, there was a much weaker aversion, and a much lower rating of immorality; unlike EAT, CRAWL, and TOUCH do not involve killing the butterfly.

The argument that disgust at eating butterflies is motivated by the immorality of killing, rather than issues of contamination or the related pathogen exposure, has two potential problems. One, we measured aversion to ingestion and not disgust. This was corrected in Study 2. Two, the general principle of immorality-based aversion/disgust for butterflies was only tested with one favored animal. In Study 2, we explored four favored animals as indicated above.

The predicted link between immorality and aversion emerged much more clearly for the American than the Indian participants. Examination of the Indian MTurk data from this and two other studies we have carried out with Indian MTurk samples has indicated a much higher rate of elimination because of failure of attention checks, but also reliably lower correlations between a wide range of variables than with the American samples (Ruby et al., 2015, Ruby & Rozin, 2019). This has led us to believe that there is more “noise” in the Indian results, and that the MTurk Indian samples are not as consistent and reliable as the American samples. We are searching for a better way to recruit Indian participants, but for the present, elected to not include Indian MTurk samples in Study 2.

Participants

We recruited a total of 251 people living in the United States. via Amazon.com’s Mechanical Turk portal, for a study on “attitudes toward food” (39% women; Mage = 33.6, SDage = 9.97, range 19–67; 45% bachelor’s degree or higher; 56% agnostic/atheist, 20% Protestant, 15% Catholic; 9% other). Participants were paid a modest sum for their time.

Procedure

Conceptually, the eight animals examined in the survey were organized into four pairs of taxonomically similar animals, one of presumed positive and one of presumed negative valence: butterfly–cockroach, canary–vulture, koala–rat, and dog–hyena. The survey began with a presentation of the name and a color photo of each of the eight animals under consideration. Animals were presented in a random order with the photographs and a different random order (varying across respondents) for each of seven questions.

The seven questions were presented in a fixed order, with each animal rated on every question. The questions covered three situations: eating the animal, measured separately as disgust and willingness to eat, the animal itself, measured separately in terms of disgust and liking, contamination instantiated by brief contact with the outcome of disgust, and separately, willingness to eat, and the same measure used in Study 1 for the immorality of killing.

The specific questions, in order of presentation, follow.

Disgust eat

“Using a scale of (0 = not at all to 100 = extremely), please rate how disgusting you find the thought of eating a bite of each of the following animals. Please note that the butterfly and cockroach would be roasted whole, and the muscle of the other animals would be cut into pieces and roasted.”

Willing rat

“Using a scale of (0 = not at all to 100 = extremely), please rate how willing you would be to eat a bite of each of the following animals. Please note that the butterfly and cockroach would be roasted whole, and the muscle of the other animals would be cut into pieces and roasted.”

Like animal

“Using a scale of (0 = not at all to 100 = extremely), please rate how likeable you find each of the following animals.”

Disgust animal

“Using a scale of (0 = not at all to 100 = extremely), please rate how disgusting you find each of the following animals.”

Disgust contact

“Imagine your favorite food. Using a scale of (0 = not at all, to 100 = extremely), please rate how disgusting you find the thought of eating a bite of this food after it has been briefly touched by the side of each of the following animals.”

Willing contact

“Imagine your favorite food. Using a scale of (0 = not at all to 100 = extremely), please rate how willing you would be to eat a bite of this food after it has been briefly touched by the side of each of the following animals.”

Immoral kill

“Using a scale of (0 = not at all to 100 = extremely), please rate how immoral you think it is to kill each of the following animals.”

The first results we report test the assumptions we made about the negative and positive animals. The bulk of the results section is then organized in terms of a set of predictions.

Animal liking/disgust

The data support our presumptions about the attributes of the animals themselves. As shown in the second and third columns of Table 3 (first column in each triad in Figure 3), mean liking for the four positive animals ranged from 78.8 to 93.3, whereas the negative animals in each pair ranged from 6.6 to 33.7 on the same scale (all pairs significantly different at p < .001, paired t tests). In terms of disgust, mean disgust for the four positive animals was rated between 6.8 and 7.3, and for the negative animals between 36.9 and 87.6 (all pairs significantly different at p < .001, paired t tests). Therefore, our stimulus selection met the anticipated criteria that we presupposed.

Disgust at animals themselves versus disgust at eating them

Prediction 1: For negative animals, disgust will be high for both the animals themselves, and eating them, whereas for positive animals, disgust will be low for the animals themselves, but high for eating them.

Participants responded very negatively to the idea of eating the positive animals (Table 3, 4th column; Figure 3, second column in each positive triad). For each positive animal, there is a major increase in disgust, comparing disgust at the animal itself with disgust at eating it: 6.8 versus 67.4 for butterfly, 5.8 versus 63.0 for canary, 7.3 versus 72 for koala, and 7 versus 78.7 for dog (d ranging from 2.20 to 2.83; all p < .001; see Table 4). Thus, in all cases, low animal disgust becomes high animal-eating disgust.

In contrast, for the negative animals, disgust at the animals themselves is high (Table 3, 4th column; Figure 3, second column in the negative triads), and the differences are much smaller: 87.6 versus 89.7 for cockroach, 52.8 versus 75.7 for vulture, 67.5 versus 87.0 for rat, and 36.9 versus 71.0 for hyena (d ranging from 0.09 to 1.03; all p < .001, except for cockroach, p = .22; see Table 4). Comparing the difference scores within each set (e.g., butterfly difference score of 60.5 vs. cockroach difference score of 2.0), reveals large and significant differences (d ranging from 0.75 to 1.75; all p < .001). For a representation of all disgust measures across all animals, see Figure 3. These findings strongly confirm Prediction 1.

Eating animals versus eating food they contacted

Prediction 2: For the negative animals, both disgust at eating the animal, and eating a favorite food contacted by the animal, will be substantial, because of contamination. However, disgust at eating a favorite food contacted by a positive animal will be much lower than eating a favorite food contacted by a negative animal. Willingness to eat will show an inverse pattern of results as disgust, as willingness to eat reflects a positive response and disgust reflects a negative response.

Although participants are highly disgusted by eating the positive animals and very unwilling to do so, this negative response is markedly lower for eating a favorite food touched by the positive animal, in accordance with Prediction 2. This occurs for all four positive examples, with a mean drop in disgust of 22 points, and a mean increase in willingness of, as it happens, also 22 points (see Table 3). In comparison to the marked drop in disgust from eating to contamination for the positive animals, there is a surprising (and unpredicted) absence ofany drop between eating and contamination for the negative animals. For the negative animals, the mean disgust “drop” is −1.3. That is, eating the contacted favorite food is slightly and nonsignificantly more disgusting than eating the animal itself. Similarly, the willingness to consume the negative animal averaged 1.6 higher than the willingness to eat the contacted favorite food, in all four cases. Because of the moderate drop in negativity for the contaminated food in the positive animals, and essentially no effect for the negative animals, the differences between the differences, for both disgust and willingness (e.g., the difference score for butterfly minus the difference score for cockroach) are highly significant (all at least p < .005; d ranging from 0.23 to 1.04; see Table 4).

Presence and absence of contamination effects

Is contamination a necessary consequence of strong disgust at eating something (or strong unwillingness to eat it)? Testing this involves making some relatively arbitrary assumptions. The relevant instances are when there is a strong rejection (disgust or unwillingness) to eating the animal itself. In the absence of that, no contamination should be expected. A priori, we elected to define a strong rejection of ingestion as a disgust score greater than 60, or a score of less than 40 for willingness. We selected, a priori, a very strict criterion for absence of contagion: a disgust score of 0 (not at all disgusting), or a willingness score of 100 (completely willing), with respect to the contacted favorite foods. Had we selected a more moderate criteria for no contagion, for example, slight disgust (<5), the effects we report below would be larger. For some people, there is probably some minimal contagion when a neutral object, or even a positive object, touches a favorite item, and we would not consider such cases as examples of contagion in this analysis. The number and percent of participants who meet the criteria for strong rejection of ingestion (disgust >60 or willingness <40) and show no disgust or complete willingness is displayed in Table 5, for all animal targets.

There is no overlap at all between percent showing no contamination for the positive versus negative animals, across the four pairings. For disgust, percent showing no contamination varies across the four positive animals from 11% to 21%; for negative animals, the range was 1% to 3.8%. A similar pattern emerges for willingness to consume the contacted food, ranging from 7% to 20% for the positive animals, and 2% to 4% for the negative animals. Within every positive–negative animal pair, the percent of people showing no contamination is significantly higher for the positive animal (see Table 5). Most striking is that although butterflies are very disgusting to consume, and there is great unwillingness to do so, 31 out of 148 qualifying participants show no contamination from contact, and 40 of 202 qualifying participants are completely willing to consume a butterfly contacted target food. These findings strongly support Prediction 2.

Immorality of killing

Prediction 3: For thepositive but not the negative animals, perceived immorality of killing will predict both disgust at eating and willingness to eat. In contrast, perceived immorality will be unrelated to contamination effects for positive animals, because these do not involve killing or harming the animals.

For the positive animals, the four correlations between Kill Immoral and Disgust Eat are all positive and significant (mean r = .36; ranging from .21 to .47; all p < .001). In contrast, the values for the four negative animals are low (mean r = .06; ranging from −.10 to .18; rat and cockroach ns, vulture and hyena ns (p < .02). Examining the correlation of KILL IMMORAL with WILLING EAT reveals a similar pattern. For the positive animals, there is a weak relationship, such that the more immoral it is to kill them, the less willing people are to eat them (mean r = −.21; ranging from −.12 to −.25; butterfly ns, other ps < .01). In contrast, the values for the four negative animals are low and mostly positive (mean r = .08; ranging from −.08 to .31; vulture, rat and hyena ns, and cockroach p < .001; note that the significant positive correlations are opposite to the predicted direction). There is no significant correlation between Kill Immoral and contamination for any of the positive animals, unlike the case for eating the animal, because there is no harm to the positive animals, as predicted by the harm/kill/immoral hypothesis (Prediction 3).

We report three findings from two studies, with Indians and Americans, which we believe are new. The first is the disjunction between the disgustingness of an entity and disgust at consuming it. The second is that the linkage between disgust and contamination is not necessary and tends not to hold for disgust at eating positive animals. The data supporting the third is weaker than the first two and is that the disgust at consuming positive animals appears to be related to the perceived immorality of harming them.

First, many people like or even love butterflies, canaries, koalas, and dogs but are unwilling to eat them and are very disgusted at the thought of doing so. The strong disgust and unwillingness to consume positive/favored animals is clearly demonstrated in Study 1 for butterflies in India and the United States, and in Study 2, for four positively regarded animals in the United States. We are confident that this would be true for many other positive/admired animals, including, in particular, all primates (and of course, among the primates, especially humans). However, although many Americans like cows, pigs, and chickens, they willingly eat them on a regular basis without disgust. For Americans at least, these may be major exceptions to the general claim we make about “positive” animals. Of course, many omnivores do not believe it is immoral to kill these three species, and a variety of processes have been invoked to explain how the immorality is not manifested in most omnivores (see Bastian & Loughnan, 2017; Piazza et al., 2015; Rothgerber, 2013). The decoupling of inherent disgust toward an animal and disgust at consuming it questions the necessity of “inherent” disgustingness as an account of disgust at ingestion.

The second distinct finding in this study is that the relation between disgust and contamination is contingent, and not necessary. We (Rozin & Fallon, 1987; Rozin et al., 1993, 2017) have proposed that contamination is a defining feature of disgust, and that has held up rather well until this set of findings. The positive animal examples question the necessity of the contamination requirement. We report here much less contamination-based rejection of positive than negative animals in both studies. However, there is still a mean substantial contamination effect. In Study 2, the drop from disgust at eating the animals themselves to disgust at eating a favorite food that was contaminated (touched) by them is much steeper for the positive animals (see Figure 3). Furthermore, we consider the percent of people who rejected eating the positive animals but did not at all reject eating food contacted by them, up to 21% for butterflies, as sufficient to reject a very tight linkage between disgust at eating and disgust contamination, for positive/admired entities.

One could argue that the big difference in contamination potency between the negative and positive animals might be partly due to the fact that the negative animals, in terms of means, elicited more disgust and more unwillingness to eat. However, our second analysis of the contamination effect includes only participants who show strong disgust or unwillingness at the prospect of consumption. In addition, for one of our animal pairs, dog-hyena, disgust is higher, and willingness lower, for the eating of the positive animal (dog; see Table 3).

For all of the negative animals, the negativity of eating a favorite food contacted by the animals was about as high as eating the animal itself. We were surprised by this finding, thinking that brief contact of a favorite food with something bad should be much better than eating a substantial piece of the bad thing. Although contamination and disgust have long been viewed as having clear links to avoidance of pathogens (Rozin & Fallon, 1987), in recent years, avoidance of pathogens has been highlighted as the principal account of much of disgust, and also contagion (Curtis, 2013; Oaten, Stevenson, & Case, 2009; Tybur et al., 2013). This view offers an account of our contamination findings. Our negative animals (especially cockroaches and rats) are generally more likely to carry pathogens harmful to humans than are our positive animals. Our EAT condition in both studies involves roasted animals, which are presumably pathogen-free, although this may not have been clearly understood by participants. In contrast, it is easy to explain the high rejection of the contacted foods, because they are touched by the live animal. From the pathogen perspective, the drop of negativity after contact with the positive animals could be thought to be the normal response, in the absence of pathogen threat. But, the pathogen threat from the positive animals is far from zero. Overall, the pattern of results we support is not fully explained by a pathogen account.

It is also possible that the disgust reaction to consuming positive animals, which is a self-report measure, is not really a disgust response, or not the same type of response as is usually studied. This touches on the important issue of the criteria for assuming disgust. Although disgust is sometimes assessed by facial, physiological, or neural responses, in general, the gold standard has been self-report (reviewed in Rozin et al., 2017). Complicating matters, self-reported disgust may include some unrelated subtypes and may be conflated with anger and/or fear in some cases. We did not measure negative affective responses other than disgust to ingestion of the animals in this study.

Our third finding suggests that disgust at consuming positive animals is of a moral type that is different from the core disgust involved in consuming disgusting animals. In taxonomizing the set of potential foods that people reject (Fallon & Rozin, 1983; Rozin & Fallon, 1987), we identified four categories—(a) distastes: entities rejected primarily on account of their sensory properties; (b) dangers: entities rejected primarily because of anticipated negative consequences following ingestion; (c) inappropriates: entities rejected, without much affect, because they are not culturally classified as foods (e.g., paper, sand); and (d) disgust: entities rejected with strong negative affect, based on knowledge of their nature or origin. We conceived of the disgust foods as having negative “inherent” properties. The pathogen account, of course, does not assume inherent disgustingness, but rather the strong general link between pathogens and animals. We now may have to expand this category to include things rejected because of their positive connotations, in conjunction with the inherently destructive events like chewing, swallowing, and digestion, consequent on eating. Moral vegetarians reject eating meat because of their knowledge of its nature. They have a high regard for most or all animals, and in particular, those typically consumed in most societies. For people who are primarily health vegetarians, meat usually falls into the danger category, but for moral vegetarians it is usually in the disgust category (e.g., Rozin, Markwith, & Stoess, 1997; Ruby, 2012).

There are two different negative aspects of eating. First, eating is the most concrete and common form of incorporation of foreign matter that humans experience. Given the risks of consumption, particularly from animal-borne pathogens, there is typically strong disgust and contamination potency for meat. It is appropriate, for example, to consider the three mammal species commonly eaten by Americans as exceptions, given that there are over 4,000 mammal species. Second, eating is an inherently destructive act (Kass, 1994), especially when the target of ingestion is an animal or part of an animal. An animal either must be killed by someone prior to eating or is killed in the act of eating. One reason some may not want to eat butterflies or dogs is that this requires them to be killed, and most of our subjects felt that this was immoral. This could trigger a feeling of moral disgust, perhaps like the reaction of a moral vegetarian to the idea of consuming meat (Rozin et al., 1997).

The results of both of our studies suggest that the rejection of eating positively regarded animals has a substantial moral component (at least for Americans; the case for Indians in Study 1 is much weaker). For Americans, in both studies, killing the positive animal is rated as much more immoral than killing the negative animal. In both studies, the highest correlation between immorality and willingness to ingest (or disgust at ingestion) was achieved by the positive animals. Finally, in both studies, a favorite food is more acceptable and less disgusting when contacted by a positive animal, in contrast to a negative animal. Contact is a situation in which the animal is not killed or even harmed. These correlational results all suggest a moral disgust component. This possibility could be tested more rigorously in future work by separating the act of ingestion from killing, for example, by assessing disgust at eating butterflies and other positive animals that died from accidental causes and were not killed for the purpose of being eaten.

The Indian results neither support nor oppose the disgust-moral hypothesis. We do not have an account for this, especially because the Indians do rate killing butterflies (as opposed to other insects) as by far the most immoral. In the present study, our other work on insect ingestion (e.g., Ruby et al., 2015; Ruby & Rozin, 2019), and some other research we have done comparing Indians and Americans recruited via Mturk, we have reliably found that a much higher percentage of Indians failed our attention checks, and the great majority of the correlations across variables in those subjects who passed attention checks were lower in Indians (even between such measures as height and weight). The Indian MTurk sample is less nationally representative than the MTurk sample of Americans, because the MTurk samples in India are English-speaking and have access to online studies, representing a minority of Indians. However, this in itself would not account for reliably lower correlations. That might result from the fact that more of the Indians are participating for financial reasons and are less careful in responding (see Kaufmann, Schulze, & Veit, 2011, for a review of Mturk Worker motivations). Or, of course, it is possible that there are some fundamental differences in the moral domain, and/or in moral disgust between Indians and Americans (Rozin et al., 1999; Shweder, Much, Mahapatra, & Park, 1997).

The most problematic area of understanding in disgust is the moral domain. Is disgust a basic general reaction to immorality (Tybur et al., 2013), or a more focused reaction to divinity violations (Shweder et al., 1997; Rozin et al., 1999)? Alternatively, some hold that disgust is not really a moral emotion, that harm is the basis of all moral judgments (e.g., Schein & Gray, 2018), and that the basic moral emotion is anger (e.g., Royzman, Atanasov, Landy, Parks, & Gepty, 2014). If, in fact, what our participants experience is genuine disgust, and if this is because of harming of the to-be-ingested animal, then our results support the supposition of Tybur et al. (2013) that disgust is an emotion that responds to all moral violations. Or, in the view of Royzman et al. (2014); Gray, Schein, and Ward (2014), and Schein and Gray (2018), our subjects are primarily showing anger in response to harm and not disgust per se. We did not measure anger in the present studies, and future research could examine this possibility. An example of how much the area of moral disgust is in flux at the present time is recent work by Kupfer and Giner-Sorolla (2017) that suggests that expressed disgust is deeply moral and involves a more principled form of moral response than anger, with is more self-centered.

With respect to our taxonomy of food rejections (Fallon & Rozin, 1983; Rozin & Fallon, 1987), it seems clear to us that our findings enlarge the disgust category, or at least the set of foods whose potential ingestion is verbally labeled as “disgusting.” The disgust and disliking ratings we report for eating positive animals, which are not themselves disgusting or disliked, are very high. This could be a confusion between disgust and anger, which might leave the positive animals in what we called the disgust (concern about the nature or origin of the animal) category of food rejections. But unlike the case for cockroaches and rats, which might have a major component of fear in the rejection (hence falling partly in the danger category), there is no reasonable way to construe the disgust at eating positive animals as fear (the danger category).

Our results are limited in that they consist only of self-report, and for the moral issues, are primarily correlational. Although results are mixed among our Indian participants, within our American samples, we do find strong support for our basic hypotheses, across four different positive animals. Surely, the moral hypothesis needs further investigation. There are about four times as many Indians as Americans in the world, so with respect to our suggestions about moral disgust, for this to be a general claim, further understanding and investigation of the Indian perspective is necessary.

Our results provide support for some aspects of the pathogen theory of disgust, while simultaneously questioning the clear distinction between core (pathogen) and moral disgust, and whether the link between disgust and contamination is a necessary one. Rejection of eating favored entities touches on important issues in the understanding of vegetarianism and moral psychology. Our moral suggestion (third hypothesis) has the weakest support from our findings, but is most relevant to the currently very active controversy about the nature of moral disgust. Our results in this area are subject to different interpretations, which will mesh differently with different views of moral disgust. These different views are diagrammed in part on our paper about the oral-moral link (Rozin, Haidt, & Fincher, 2009). One thing our results may suggest is that moral disgust, rather than being about either divinity-based or all moral violations, is a reaction to moral violations that have a bodily component, such as eating or sex. In this context, it may be that disgust serves as an enhancer of the basically anger reactions to harm violations.

At this point, the main contribution of this research is to expand the elicitors of disgust to eating admired animals (re: our original taxonomy of food rejections, Fallon & Rozin, 1983), separating disgust as an inherent property of entities from disgust at ingestion, and questioning a tight linkage between disgust and contamination. With the recent great attention to disgust, and the salience of a pathogen view of the most fundamental disgusts, our findings on the sharp reduction in contamination of food for admired animals may have theoretical significance. Our view has been that contact involves a transfer of essence. The essence of a positive animal is not negative and should not produce any contamination. But then, of course, we have to explain why eating positive animals with direct incorporation of their essence, is so negative and disgusting. This is where harm-based morality may enter the picture.

Bastian, B., & Loughnan, S. (2017). Resolving the meat-paradox: A motivational account of morally troublesome behavior and its maintenance. Personality and Social Psychology Review, 21, 278–299. 10.1177/1088868316647562

Curtis, V. (2013). Don’t look, don’t touch, don’t eat. The science behind revulsion. Chicago, IL: University of Chicago Press. 10.7208/chicago/9780226089102.001.0001

DeFoliart, G. R. (1999). Insects as food: Why the western attitude is important. Annual Review of Entomology, 44, 21–50. 10.1146/annurev.ento.44.1.21

Fallon, A. E., & Rozin, P. (1983). The psychological bases of food rejections by humans. Ecology of Food and Nutrition, 13, 15–26. 10.1080/03670244.1983.9990728

Gray, K., Schein, C., & Ward, A. F. (2014). The myth of harmless wrongs in moral cognition: Automatic dyadic completion from sin to suffering. Journal of Experimental Psychology: General, 143, 1600–1615. 10.1037/a0036149

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74, 1464–1480. 10.1037/0022-3514.74.6.1464

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.

Halloran, A., Flore, R., Vantomme, P., & Roos, N. (Eds.). (2018). Edible insects in sustainable food systems. Cham, Switzerland: Springer. 10.1007/978-3-319-74011-9

Kass, L. R. (1994). The hungry soul. Eating and the perfection of our nature. New York, NY: The Free Press.

Kaufmann, N., Schulze, T., & Veit, D. (2011, August). More than fun and money. Worker Motivation in Crowdsourcing-A Study on Mechanical Turk. AMCIS, 11, 1–11.

Kupfer, T. R., & Giner-Sorolla, R. (2017). Communicating moral motives: The social signaling function of disgust. Social Psychological and Personality Science, 8, 632–640. 10.1177/1948550616679236

Looy, H., Dunkel, F. V., & Wood, J. R. (2014). How then shall we eat? Insect-eating attitudes and sustainable foodways. Agriculture and Human Values, 31, 131–141. 10.1007/s10460-013-9450-x

Oaten, M., Stevenson, R. J., & Case, T. I. (2009). Disgust as a disease-avoidance mechanism. Psychological Bulletin, 135, 303–321. 10.1037/a0014823

Paoletti, M. G., & Dreon, A. L. (2005). Minilivestock, environment, sustainability, and local knowledge disappearance. In M. G.Paoletti (Ed.), Ecological implications of minilivestock: Potential of insects, rodents, frogs and snails. Boca Raton, FL: CRC Press. 10.1201/9781482294439

Piazza, J., Ruby, M. B., Loughnan, S., Luong, M., Kulik, J., Watkins, H. M., & Seigerman, M. (2015). Rationalizing meat consumption. The 4Ns. Appetite, 91, 114–128. 10.1016/j.appet.2015.04.011

Rothgerber, H. (2013). Real men don’t eat (vegetable) quiche: Masculinity and the justification of meat consumption. Psychology of Men & Masculinity, 14, 363–375. 10.1037/a0030379

Royzman, E., Atanasov, P., Landy, J. F., Parks, A., & Gepty, A. (2014). CAD or MAD? Anger (not disgust) as the predominant response to pathogen-free violations of the divinity code. Emotion, 14, 892–907. 10.1037/a0036829

Rozin, P., Kurzer, N., & Cohen, A. B. (2002). Free associations to “food”: The effects of gender, generation, and culture. Journal of Research in Personality, 36, 419–441.

Rozin, P., & Fallon, A. E. (1987). A perspective on disgust. Psychological Review, 94, 23–41. 10.1037/0033-295X.94.1.23

Rozin, P., Haidt, J., & Fincher, K. (2009). From oral to moral. Science, 323, 1179–1180. 10.1126/science.1170492

Rozin, P., Haidt, J., & McCauley, C. R. (1993). Disgust. In M.Lewis & J.Haviland (Eds.), Handbook of emotions (pp. 575–594). New York, NY: Guilford Press.

Rozin, P., Haidt, J., & McCauley, C. R. (2017). Disgust. In M.Lewis, J.Haviland, & L.Barrett (Eds.), Handbook of emotions (4th ed., pp. 815–834). New York, NY: Guilford Press.

Rozin, P., Lowery, L., Imada, S., & Haidt, J. (1999). The CAD triad hypothesis: A mapping between three moral emotions (contempt, anger, disgust) and three moral codes (community, autonomy, divinity). Journal of Personality and Social Psychology, 76, 574–586. 10.1037/0022-3514.76.4.574

Rozin, P., Markwith, M., & Stoess, C. (1997). Moralization: Becoming a vegetarian, the conversion of preferences into values and the recruitment of disgust. Psychological Science, 8, 67–73. 10.1111/j.1467-9280.1997.tb00685.x

Rozin, P., Millman, L., & Nemeroff, C. (1986). Operation of the laws of sympathetic magic in disgust and other domains. Journal of Personality and Social Psychology, 50, 703–712. 10.1037/0022-3514.50.4.703

Rozin, P., & Nemeroff, C. (1990). The laws of sympathetic magic: A psychological analysis of similarity and contagion. In J. W.Stigler, R. A.Shweder, & G.Herdt (Eds.), Cultural psychology: Essays on comparative human development (pp. 205–232). New York, NY: Cambridge University Press. 10.1017/CBO9781139173728.006

Ruby, M. B. (2012). Vegetarianism. A blossoming field of study. Appetite, 58, 141–150. 10.1016/j.appet.2011.09.019

Ruby, M. B., & Rozin, P. (2019). Disgust, sushi consumption, and other predictors of acceptance of insects as food by Americans and Indians. Food Quality and Preference, 74, 155–162. 10.1016/j.foodqual.2019.01.013

Ruby, M. B., Rozin, P., & Chan, C. (2015). Determinants of willingness to eat insects in the USA and India. Journal of Insects as Food and Feed, 1, 215–225. 10.3920/JIFF2015.0029

Schein, C., & Gray, K. (2018). The theory of dyadic morality: Reinventing moral judgment by redefining harm. Personality and Social Psychology Review, 22, 32–70. 10.1177/1088868317698288

Shweder, R. A., Much, N. C., Mahapatra, M., & Park, L. (1997). The “big three” of morality (autonomy, community, and divinity), and the “big three” explanations of suffering. In A.Brandt & P.Rozin (Eds.), Morality and health (pp. 119–169). New York, NY: Routledge.

Tybur, J. M., Lieberman, D., & Griskevicius, V. (2009). Microbes, mating, and morality: Individual differences in three functional domains of disgust. Journal of Personality and Social Psychology, 97, 103–122. 10.1037/a0015474

Tybur, J. M., Lieberman, D., Kurzban, R., & DeScioli, P. (2013). Disgust: Evolved function and structure. Psychological Review, 120, 65–84. 10.1037/a0030778

van Huis, A., Van Itterbeeck, J., Klunder, H., Mertens, E., Halloran, A., Muir, G., & Vantomme, P. (2013). Edible insects: Future prospects for food and feed security. Rome, Italy: Food and Agriculture Organization of the United Nations. Retrieved from http://www.fao.org/docrep/018/i3253e/i3253e.pdf

Yen, A. L. (2009). Edible insects: Traditional knowledge or Western phobia?Entomological Research, 39, 289–298. 10.1111/j.1748-5967.2009.00239.x

Submitted: July 30, 2017 Revised: January 10, 2019 Accepted: January 14, 2019