Journal of Personality and Social Psychology Copyright 1586 by the American Psychological Association, Inc.

1986, Vol. SO, No. 4, 703-712 0022-3514/86/$00,75

Operation of the Laws of Sympathetic Magic

in Disgust and Other Domains
Paul Rozin, Linda Millman, and Carol Nemeroff
University of Pennsylvania

Two laws of sympathetic magic were described by Frazer and Mauss at the beginning of this century
to account for magical belief systems in traditional cultures. In this study, we show that these laws fit
well with a variety of behaviors in American culture, in responses to disgusting, dangerous, or valued
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

objects. The first law, contagion, holds that "once in contact, always in contact." That is, there can be
This document is copyrighted by the American Psychological Association or one of its allied publishers.

a permanent transfer of properties from one object (usually animate) to another by brief contact. For
example, in this study we show that drinks that have briefly contacted a sterilized, dead cockroach
become undesirable, or that laundered shirts previously worn by a disliked person are less desirable
than those previously worn by a liked or neutral person. The second law, similarity, holds that "the
image equals the object," and that action taken on an object affects similar objects. In this study, we
demonstrate this law by showing, for example, that people reject acceptable foods (e.g., fudge) shaped
into a form that represents a disgusting object (dog feces), or that people are less accurate at throwing
darts at pictures of the faces of people they like. With these and other measures, we found a great
deal of evidence for the operation of the laws of sympathetic magic in all 50 of the subjects we studied.
The laws of sympathetic magic correspond to the two basic laws of association (contiguity and similarity}.
We discuss the parallel and report a disgust conditioning study to develop this parallel.

In this article, we argue that the principles of sympathetic The second law, similarity, holds that things that resemble one
magic, thought to be a characteristic of many "primitive" belief another share fundamental properties ("the image equals the
systems and rituals, are also operative in some aspects of daily object"). As with contagion, the image is believed to contain the
life in modern Western culture. essence of its "source," so that action on the image can produce
The two laws of sympathetic magic were propounded most similar effects on the source ("like produces like"). In a special
clearly by Sir James Frazer (1890/1922/1959), in "The Golden case of similarity, the name of the object (written or spoken),
Bough," and Marcel Mauss (1902/1972), in "A General Theory being an attribute of the object, is held to contain the object's
of Magic." They were proposed to describe widespread magical essence. The two laws together are illustrated in the Malay custom
practices and rituals in traditional cultures. According to the (Frazer, 1890/1922/1959) in which a clay figure of an enemy
first, the law of contagion, things that once have been in contact (similarity) is constructed, incorporating residues (e.g., hair, fin-
with each other may influence each other through transfer of gernail parings) from that person (contagion). The figure is then
some of their properties via an "essence." This influence remains scorched, causing harm to the enemy by the action of both laws.
after the physical contact has ceased, and may be permanent By our analysis, the laws of sympathetic magic have two basic
(hence, "once in contact, always in contact"). The "contact" dimensions. First, the "magic" can be positive or negative; thus,
may be directly between an offensive or revered person or animal in contagion, contact of an object with a loved or respected person
and a previously neutral object, as when a person grows, cooks, can enhance the value of the object (positive contagion), whereas
or touches a food. In magical practices, contact is frequently contact with a disliked or despised person (or an offensive sub-
manifested as a personal residue; fingernail parings, spittle, or stance such as feces) can devalue the object (negative contagion).
other personal residues retain essential properties of their original Second, transmitted essence can mediate effects either in its
owner (source). The "essence" remains in these residues in some source or in its recipient. In forward causation, the essence in-
form of nonphysical contact with its source. This allows for the fluences the entity it has contacted (the recipient). This is a pow-
possibility that action taken against the essence contained in a erful force in Hindu India, where the past history of a food (who
residue or recipient can affect the original source, and is the basis has contacted it) determines whether it is acceptable to an in-
for a major form of sorcery. dividual and serves as a major way of maintaining the social
(caste) structure (Appadurai, 1981). Another example of forward
contagion, widespread in traditional cultures (Frazer, 1890/1922/
This research was supported by funds from the John D. and Catherine 1959; Crawley, 1902), is the belief that one takes on the properties
MacArthur Health Related Behaviors Network. Thanks to Jonathan
of the food one eats ("You are what you eat"). In backward
Baron, Muriel Bell, April Fallen, and Bonnie Ross for critical comments
causation, action on the residue (essence) reflects back on its
on the manuscript, and MiYung Yoon for assistance in carrying out the
studies. source, as in the practice of scorching the hair or fingernail parings
Correspondence concerning this article should be addressed to Paul of an enemy. Although there may be a rational basis for forward
Rozin, Department of Psychology, University of Pennsylvania, 3815 Wal- causation (e.g., true microbial contamination), because the res-
nut Street/T3, Philadelphia, Pennsylvania 19104. idue is in physical contact with the recipient, backward causation

703
 704 P. ROZIN, L. MILLMAN, AND C. NEMEROFF

cannot be accounted for in terms of known physical principles. stimuli present. In the second part, they completed a questionnaire. The
We shall examine both the positive-negative and forward-back- first part lasted about 40 min, the second about 15 min. The first part
ward dimensions in this study. consisted of a series of ministudies always presented in the same order:
roach contamination (negative contagion, disgust), roach conditioning
The phenomenon of disgust offers a special opportunity to
(negative contagion and similarity, disgust), dart throwing (positive and
study the laws of sympathetic magic, because disgusting stimuli
negative similarity, interpersonal), cyanide labels (negative similarity,
produce strong effects that are in accord with these laws. Fur-
danger), and imitation dog feces and vomit (negative similarity, disgust).
thermore, these effects are easy to produce in the laboratory
Each procedure is described later.
under ecologically valid but controlled conditions. This is much
harder to accomplish with interpersonal relations.
Part I: Direct Measurements
Expanding on a definition of Angyal (1941), we define disgust
as a revulsion at the prospect of (oral) incorporation of an of- The same rating scale was used for all studies in Part 1. The scale was
fensive substance (Rozin & Fallon, 1985). The objects of disgust a line 200 mm long, labeled dislike extremely at the left end and like
vary from culture to culture, but include feces universally, and extremely at the right end. There were no other marks on the fine. Subjects
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

could not see the lines with their previous ratings.

in most cultures, other body products as well. They also include
This document is copyrighted by the American Psychological Association or one of its allied publishers.

During each of the sequences in Part 1, the experimenter maintained

the meat of most animal species (e.g., in our culture, insects and
a somewhat formal, pleasant, and calm demeanor, and was unexpressive
worms) and certain parts (e.g., eyes) of some edible species (An-
even during the presentation of unusual (e.g., disgusting or dangerous)
gyal, 1941; Rozin & Fallon, 1981, 1985). A striking feature of stimuli.
disgusting substances is that they can render a perfectly good Roach contamination. In this procedure, the effects of contact between
food inedible by brief contact, even if there is no detectable trace juice and a sterilized, dead cockroach were assessed. The experimenter
of the offensive item. The idea (history) of contact is sufficient. said:
We have called this the principle of contamination (Fallon, Rozin,
I'm going to offer you some juices. I'd like you to taste them and
& Pliner, 1984; Rozin & Fallon, 1981, 1985). So far as we know,
rate how much you'd like to have another sip of the same juice (show
this effect is universal among adults. rating line). Notice that this line goes from dislike extremely to like
In this study, we examine the operation of the two laws of extremely. Just mark the line at the place that indicates how much
sympathetic magic in American adults, using both measurements you'd like to drink some more of the sample we offer. [Experimenter
demonstrates marking with a pencil slash mark in the middle of the
in the laboratory and questionnaire responses. We include re-
line.] This would mean that you were neutral about this item. After
sponses to disgusting stimuli, dangerous stimuli, and interpersonal
each rating, we'll turn the rating sheet face down and move on to
situations. We recognize that findings that are in accord with the the next sheet
laws of sympathetic magic may also be accounted for in other
ways. We will deal with alternative accounts, particularly the Two distinctively shaped, transparent plastic disposable glasses were
each removed from their new plastic wrappings and placed in front of
laws of association, in the Discussion section. Our aim is to dis-
the subject. New, unopened 8-oz cartons of a standard brand of grape
cover whether the set of phenomena that gave rise to the for-
and apple juice were opened in front of the subject, and a few ounces of
mulation of the laws of sympathetic magic are present in Amer-
each were poured into separate glasses. The cartons were placed behind
ican culture. In the event of positive findings, a next step is to
the appropriate glasses, so that the subject could see them and their brand
determine the extent to which these laws of magic are an effective markings. The experimenter then said:
and economical way of accounting for the phenomena.
Take a sip of this (the apple juice) and indicate your rating of how
much you'd like to drink some more on this rating sheet. [Wait]
Method Now the same for this other juice. [Wait] Now please tell me which
of the two juices you prefer; that is, the one you would prefer to
Subjects drink some more of. [Wait for subject's response.) Now please take
another sip of the juice you prefer. [After each statement of pref-
erence, in this choice and all that follow, the subject was asked to
Subjects were 50 people, approximately half students, from the Uni-
sample the preferred item. This was done to produce the expectation
versity of Pennsylvania community. There were 17 men and 33 women,
in the subject that the consequences of his stated preference would
whose ages ranged from 17 to 50 years (mean 23.6). They were recruited
be real.]
primarily from advertisements for studies of food preferences. The 50
subjects represent approximately half of the subjects who agreed to par- Apple juice was always tested first. For 30 subjects, the preferred juice
ticipate; the remaining half failed to come to their experimental session. was selected as the target for the roach, and for 20 subjects, the roach
All who came completed the session. was placed in the nonpreferred juice. One juice was contacted by a plastic
birthday candleholder (the control), the other by a dried, sterilized cock-
roach (Periplaneta americana), about 4 cm in length. The apple juice
Procedures
was always contacted fiist. We will describe the sequence in which the
The study took place in a laboratory room ( 4 X 4 m). Care was taken roach contamination occurred first.
to give the room and the experimenter a clean appearance. All glasses, The experimenter placed a tray covered with paper towels in front of
plates, and utensils used were disposable, were conspicuously taken out her. She then removed part of the paper towel cover of the tray, exposing
of their original commercial packaging at the beginning of each trial, and the roach in a small plastic cup, lifted the roach with a pair of forceps
were discarded promptly after use. The subject sat at a table covered with lying next to it, and said: "Now I'm going to take this sterilized, dead
a tablecloth, in the center of the room. The experimenter sat at the table cockroach, it's perfectly safe, and drop it in this juice glass." The roach
at a 90 degree angle to the subject. All experimental materials not in was dropped into the glass, and stirred with the forceps for 5 s. The
current use were shielded from subject's view by a .5-m high barrier subject was then asked to count the roach's legs (to assure his attention).
across the table. The study consisted of two parts. The first part was in The roach was then removed with a new plastic spoon that came from
vivo, with subjects participating in ratings or activities with appropriate a new spoon container, and placed back in its original cup. The spoon
 SYMPATHETIC MAGIC 705

was discarded. This sequence was repeated with the plastic candleholder The experimenter set out two different colored plastic cups, one in front
in the other juice, using a different forceps. Subjects then repeated the of each bottle, and poured water from a glass pitcher into both, until
ratings of each juice, stated their preference, and sipped from the preferred they were about half full. Now, using separate, new plastic spoons for
juice. (For some subjects the first contact, always with apple juice, was each bottle, the experimenter put a half spoonful of powder from the
with the candleholder.) "sugar" bottle into one cup, and stirred it. The spoon was discarded, and
Roach conditioning. This continuation of the previous manipulations the same was done with the sugar in the cyanide bottle, with a new spoon.
was designed to determine whether a new sample of the kind of juice The subject then rated, on the 200 mm line, how much he would like to
contaminated by a roach on the previous trial would display any negative drink from each of the cups, and stated a preference between the two.
properties. (We call this roach conditioning because, although it can be The subject was then asked to take a sip of the sugar water from the
accounted for by a sequential application of the laws of contagion [roach preferred cup, and subsequently to account for his or her choice.
in juice] and similarity [test with similar juice], it also fits well with a Cyanide label—2. We thought avoidance of the cyanide-labeled bottle
classical conditioning paradigm.) might be motivated by doubts about the real contents of the bottle (though
The roach, candleholder, and two glasses of "contaminated" juice were it seems absurd that the experimenter would try to poison the subject by
removed from subject's view. Two new plastic glasses, of the same types offering a poison-labeled bottle). For this reason, in the second sequence,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

used initially, were filled with juice from the apple and grape cartons, the subject himself labels the bottles. Initially, we performed this second
This document is copyrighted by the American Psychological Association or one of its allied publishers.

such that the same shape and size of glass held the same juice as before. test only for subjects who indicated a substantial preference for the sugar-
Subjects then went through the standard rating and preference measure- labeled bottle's contents. The last 20 subjects, however, were run on this
ment. procedure independent of their results on the first sequence. The total JV
Dart throwing. We measured subjects accuracy in throwing darts at for the second sequence was 38; that is, 12 subjects were eliminated on
photographs of liked or disliked people and neutral (blank) targets. Because the basis of their performance on the first cyanide test.
a person's picture is similar to the real person, by the law of similarity, Previously used bottles and glasses were taken away, and two similar,
people should be more disturbed and hence less accurate in throwing empty bottles were brought out. The covers were removed, and sugar
darts at a picture of a liked person than at a neutral target (see the Dis- from a 5-lb box of locally sold "Domino" sugar (sucrose) was poured
cussion section). into each bottle (to a level of about one-third full). The subject was then
Subjects stood at a mark on the floor. A dart board was hung on a given two peel off labels on a piece of paper. One read "Sucrose (table
wall, with its midline roughly at shoulder level. The board was 182 cm sugar)," the other read "Sodium Cyanide." (Through an error, this cyanide
from the subject. Targets on pieces of standard 8'A X 11 in. pieces of label did not have a red "Poison" sticker affixed to it.) The subject was
white paper were pinned to the center of the board. For each target, asked to put one label on each bottle, in any way he wanted. Then, the
subjects threw four darts twice. After the first four darts were thrown, procedure used for Cyanide Label I was repeated (mixing sugar water,
they were removed from the target, and four more were thrown. The rating of both solutions, indicating a preference, and sipping the preferred
holes left in the paper by the darts were used later to measure accuracy. solution).
The first target was a black dot about 1 cm in diameter, in the center of Similarity: Dogfeces. Subjects were offered a piece of high-quality
an otherwise blank piece of paper. Subjects got eight "practice" throws chocolate fudge, in a square shape, on a paper plate. They ate the piece,
at this target. It was then removed, and a second, identical target was and rated on the standard line, their desire to eat another piece. (If the
used (Blank 1). Five experimental targets followed, in a random order subjects said they were dieting, they were asked to ignore this fact in
that varied from subject to subject (discussed later). Finally another black rating their desire for another piece). Two additional pieces of the same
dot target (Blank 2) was presented to allow correction for improvement fudge were presented, each on its own paper plate. One piece was shaped
due to practice. The five experimental targets were high quality xeroxed in the form of a disc or muffin, and the other in the shape of a surprisingly
copies of front views of the faces of Adolf Hitler, John F. Kennedy, and realistic piece of dog feces. The pieces were of approximately the same
the experimenter (each approximately 20 cm high), and two additional size. The experimenter said: "Here are two more (present both) pieces
blank targets with black dots, on which subjects were asked to project of the same fudge. One (pointing) is moulded in the shape of a disc or
images of the faces of the person they liked and disliked most in the muffin, and the other (pointing) in the shape of a dog doo." The subjects
world, so that the black dot fell between the eyes. For the faces, subjects rated their desire to eat more of each (disc first, then dog feces), indicated
were instructed to aim for the spot "between the eyes," marked by the the one they preferred, and were then asked to take a bite from the pre-
black dot for the imaged faces. For the blanks, subjects were told to aim ferred piece.
for the black dot. In summary, subjects were presented with one practice Similarity: Vomit. A similar procedure compared a flat rubber sink
blank target, one blank, five experimental targets, and another blank. stopper (about 13 cm in diameter) and a piece of rubber imitation vomit,
Subjects threw eight darts at each. of about the same size, purchased in a novelty store. This procedure,
Cyanide label—1. In this procedure, we ascertain whether the label unlike the procedure with fudge, involved no prior exposure to a "blank"
"sodium cyanide" imparts its quality to the substance it labels (by the piece of rubber. The experimenter said:
principle of similarity; see the Introduction and Discussion sections).
Here is a clean rubber sink stopper. Here is a rubber imitation vomit,
Subjects returned to their seat at the table and were presented with
sold by novelty stores. Both are clean. Rate how much you would
two brown glass 500 ml "chemical" bottles, each about one-quarter filled
like to hold the sink stopper between your lips for 10 seconds. [Wait]
with a white powder, which was, in fact, sucrose. One had a typed label
Now rate how much you would like to hold the rubber vomit between
on it that said "Sucrose (Table sugar)," the other a typed label that said your lips for 10 seconds. [Wait] Now indicate which you would
"Sodium Cyanide" with a red printed "Poison" sticker below it. The prefer to hold between your lips.
experimenter said:

Here we have two bottles with powder in them. The powder in both Part 2: Questionnaire
bottles is sucrose, that is, table sugar. These are brand new bottles
that we just bought. They never had anything in them but sugar.
This part of the study consisted of a 53-item questionnaire dealing
This bottle (on the subjects' left) has a sucrose label that we put on
primarily with disgust and contamination. In this article, we will discuss
it. It's a brand new label, that was never on any other bottle. This
other bottle (on the subjects' right) has a brand new sodium cyanide only a small number of items that deal directly with magical issues. For
label on it. This label was never on any other bottle and was never most of the questions, subjects were asked to use a scale that ran from
even near cyanide. Remember, sugar is in both bottles. -100 (most unpleasant experience you can imagine) through 0 (neutral)
 706 P. ROZIN, L. MILLMAN, AND C. NEMEROFF

to +100 (most pleasant experience you can imagine). They were en- CONTAGION
couraged to select intermediate values. In contrast to Part 1, ratings here
were numerical, rather than on a line, and subjects could see previous
ratings on the questionnaire. Subjects responded to a series of questions
by placing the appropriate number next to each question. Questions 1 BRACK CANDLE
to 5 dealt with interpersonal contamination (positive and negative), and 140 R ACH
° HOLDER
Questions 6 to 10 with contamination in disgust. The questions were as
120
T
follows: T T ,T
(For Items 1-3, the subject was asked to think of a blouse/shirt of a
style she or he liked, and to consider each of the questions to be about 100 '
another instance of that same blouse style.)
5 80
1. Wearing a cleaned blouse (shirt) of a style you like that comes from Of

the racks of a used clothing store. 2 60

2. Wearing a cleaned blouse (shirt) of a style you like that belonged 2
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

to someone you dislike. 40

This document is copyrighted by the American Psychological Association or one of its allied publishers.

3. Wearing a cleaned blouse (shirt) of a style you like that belonged

to someone you like. 20 T
4. Using a toothbrush that belongs to (and has been used by) the
person you are most sexually attracted to.
5. Using a brand new toothbrush. - § ? g i g1
6. Blowing your nose using a clean, new piece of facial tissue. ill s 'f 1
7. Blowing your nose using soft toilet paper from a brand new roll.
For Items 8-10, the subject was asked to "consider a bowl of your favorite " I " I
soup" and then rate:
8. Eating the soup. Figure 1. Mean rating by 49 subjects (on 200 mm scale) of juices before
9. Eating a new bowl of fresh soup after you have spit into it. (pretest) and after (contagion) contact with a cockroach or birthday
10. Eating a new bowl of fresh soup poured into a brand-new bedpan. candleholder, and of new instances (conditioning) of these same juices.
Subjects also completed a brief disgust-contamination sensitivity ques- (Error bars represent standard errors.)
tionnaire that we had used in previous studies (Rozin, Fallon, & Mandell,
1984), and that reveals large individual differences in this population.
This questionnaire contains 10 items, all asking for ratings on a 9-point
hedonic scale from like extremely (9) to dislike extremely (I). Eight of see Table 1) was -98.8 points (p < .001). This net change was
the questions deal with contamination of a bowl of favorite soup by a negative for 46 of 49 subjects.
thoroughly washed used fly swatter, a brand-new fly swatter, a thoroughly
washed used comb, a brand-new comb, a grasshopper (after it was re- Roach Conditioning
moved), new soup in the same (grasshopper) bowl after the bowl was
washed in a dishwasher, a leaf from a houseplant, and new soup from A new glass of the juice that was previously "reached" (called
the same, unwashed (leaf) bowl. For the last 2 items, subjects rated a new reached) is rated a mean of 10 points lower than the original
highly preferred (9 rating) cookie after a bite had been taken by a waiter juice, in contrast to a 2-point increase in the rating of the control
in a restaurant, and after a new instance of it had fallen on a lawn while juice (Figure 1). Again, there was no evidence for generalization
picnicking. to the control juice. The overall conditioning effect, compensating
for change in the control juice, is —11.33 points (t = 1.74, p <
Results .05, one-tailed). An effect in the directions predicted by condi-
tioning (drop in new reached juice preference in comparison to
We begin by describing the results from the direct measure- controls) took place in 30 of 49 cases (ns). Ratings of the new
ments of Part 1 of the study. We then consider the direct mea- reached juice dropped substantially (either 50 points or 50%
surement and questionnaire data relating to each law of sym- with respect to original levels) in 6 subjects.
pathetic magic, in negative and positive forms, in turn. Finally,
we consider the relations among the various measures and in- Similarity in Disgust
dividual differences in magical "behaviors." Unless otherwise
indicated, probabilities reported for findings are based on two- Both the dog feces and vomit imitations were rated much
tailed t or binomial tests. lower than the same substance in more innocuous form (Figure
2, panels A and B). The drops in ratings (compared with the
Direct Measurement Studies control disc fudge or rubber sink stopper) were -47.18 for dog
feces fudge (p < .001) and -49.26 for rubber vomit (p < .001)
Roach Contamination (Table 1). Almost all subjects showed both effects (Table 1).

Not surprisingly, contact with a sterilized roach had a massive Similarity in Danger
effect on the acceptability of a particular type of juice (Figure
1). The mean drop on the 200-point rating scale was 102 points. The sugar labeled as "Sucrose" was preferred to the sugar
The contagion does not generalize to the juice contacted with labeled as "Sodium Cyanide" by 41 of 50 subjects (p < .001).
the candleholder (drop of 3 points). The net change of the The mean difference between the cyanide- and sugar-label ratings
"reached" juice (compensating for changes in the control juice; was -30.58 (p < .001) (Figure 2, panel C; Table 1). When asked
 SYMPATHETIC MAGIC 707

Table 1
Evidence for Different Types of Sympathetic Magic
Category Measure* N M" SD Cases

Contagion
Negative Roach overall13 49 -98.82** 71.62 46**
Soup: spit - plain 50 -93.26** 57.87 49.S**
Blouse: dislike — used 50 -36.44** 42.90 43.5**
(Toothbrush, see Footnote c)
Positive Blouse: used - like 49 -12.86ft 37.06 35*
Toothbrush0: new - like 50 19.68* 46.56 36.5*"
Similarity
Negative Fudge: dog feces - disc 50 -47.18** 49.42 45**
Rubber: vomit — mat 50 -49.26** 38.14 49**
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Label: cyanide - sugar 50 -30.58** 44.52 41**

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Soup: bedpan - bowl 49 -92.78** 57.88 49**

Dart*: Hitler - blank 47 2.11 30.48 23
Dart11: disliked - blank 47 -4.91 29.69 28
Tissue: toilet - facial 50 -4.30tt 12.00 30.5
Positive Darf1: blank - Kennedy 47 -11.32ft 31.84 30
Darf1: blank - liked 47 -11.02tt 28.22 34*

t p < .05. ttf < -02. * p < .01. ** p < .001.

' All scores are calculated so that a negative score means a magical effect in the predicted direction. b Roach overall refers to the change in the
"reached" juice minus the change in the control juice. ° Predicted to be an instance of positive contagion, but results show a significant negative
contagion effect. d The dart score is the median distance from the target for 8 throws. Hence, larger numbers mean poorer accuracy. Here, as opposed
to Figure 3, all dart scores are presented so that a negative score means a magical effect in the predicted direction.

to explain their choice, the most common responses were ref- Similarity in Response to Photographs
erence to the label (JV = 23), and no response (N = 17). Only 1
Accuracy of dart throwing was determined by measuring (in
subject suggested the possibility that there might be cyanide in
mm) the distance between the hole made by each of the eight
the cyanide-labeled bottle. The second cyanide manipulation, in
which the subject put the labels on herself (with a smaller N, see darts and the target center. Inaccuracies of more than 200 mm
the Methods section), showed a much smaller but still significant were scored as 200 mm. The median distance for the eight throws
at each target was computed for each target. On the basis of the
effect, with a net difference of 16.5 points between cyanide and
accuracy for the first and last blank targets, a linear estimate was
sugar (Figure 2, panel D) (/ = 2.60, p < .02).
made of expected accuracy, assuming another blank target was
present for each subject, for each of the five order positions of
SIMILARITY the face targets. The accuracy score used was the subjects' ac-
curacy for the target in question minus the predicted accuracy
based on our linear projection. There was a significant drop in
160 mean accuracy for Kennedy (11.32 mm) and liked person (11.02
140 mm), p < .02, in both cases (Figure 3; Table 1). The more neutral
(experimenter) or negative stimuli (Hitler or disliked person)
120
showed no significant effects (Figure 3; Table 1).
i 100
2
5 80
LU
Evidence for Laws of Sympathetic Magic
60
Negative Contagion
40
Table 1 lists results from all measures that could show a neg-
20
ative contagion effect. Significant negative contagion effects were
0 seen for all three predicted cases (roach in juice, disliked person's
blouse, and spit in soup; Table 1).

Positive Contagion
Figure 2. Mean rating (on 200 mm scale) by 50 subjects of four pairs of
Two questionnaire items were selected to show positive con-
items. [(A) Fudge shaped in the shape of dog feces or a disc. (B) A flat
piece of rubber in the shape of a sink stopper (mat) or vomit. (C) Bottles
tagion. There was a significant positive contagion effect for a
filled with sugar, and labeled either "Sodium Cyanide" or "Sugar." (D) laundered blouse previously worn by a liked person (Table 1).
Same as (C), but subject put the labels on the bottle. Error bars represent For 7 subjects, there was an enhancement in value of at least 50
standard errors.] points, whereas 8 subjects reported a more negative rating for
 708 P. ROZIN, L. MFLLMAN, AND C, NEMEROFF

SIMILARITY-DARTS Negative Similarity

Of seven measures predicted to show negative similarity, five
(dog feces fudge, rubber vomit, cyanide label [1 or 2], bedpan,
and toilet tissue) showed significant decreases in ratings in com-
parison to more neutral equivalents; Table 1). Neither of the
negative photographs (Hitler or disliked person) significantly al-
tered dart throwing accuracy (Table 1).

-2 Positive Similarity
-4 Both measures (picture of Kennedy and liked person image)
showed the predicted effect, a significant decrease in accuracy
-6
(Table 1).
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Patterns of Response and Individual Differences

-10

-12
The relations among the measures, expressed as Pearson r&,
JFK liked Hltlir disliked are displayed in Table 2, (We leave out the conditioning data and
the toilet tissue results because in both cases most subjects showed
no effects; and the second cyanide label data, because it was
Figure 3. Mean accuracy (in mm), compared to expected performance
carried out with only 38 subjects). For convenience, correlations
with a blank target, for 47 subjects throwing darts at face targets. [The of .3 or greater (significant at p < .02, one-tailed with N = 50)
target was a spot between the eyes. The targets were (in order along the are starred in the table, and will be the focus of the discussion.
A'axis) a xeroxed photograph of John F. Kennedy, the imagined face of We examine first whether there is a tendency for an individual
the person the subject likes most, xeroxed photograph of the experimenter, to be generally influenced by a particular law of sympathetic
xeroxed photograph of Adolf Hitler, the imagined face of the person the magic. Two of the three intercorrelations of the three negative
subject dislikes most. Error bars represent standard errors.] contagion items (roach, spit-in-soup, and blouse-of-disliked per-
son; Table 2) are substantial (/• = .42 and .52) and one (between
disliked person's blouse and roach) is small and not significant
the liked person's blouse than for a used blouse of unknown (.07). The mdn r — .42. If we add into our calculations the one
history. Rather than a net enhancement of one's lover's tooth- item (blouse-of-liked-person) of positive contagion, the coherence
brush over a new toothbrush (positive contagion), we found a of the category decreases markedly (mdn r = .20; Table 2), be-
significant effect in the opposite direction (Table 1). Although 4 cause correlations across positive and negative contagion are low
subjects showed substantial enhancement effects for the lover's (varying between .17 and .21). (Note, we score instances of pos-
toothbrush of at least 50 points, the majority of the subjects itive contagion or similarity such that more negative scores mean
preferred the new toothbrush (Table 1). more enhancement of value, i.e. more magical effect. Hence, if

Table 2
Correlation (Pearson r) of Basic Measures
Measures' spi bid bll tbr ens dfe vom bpa dtt dgt
d
Roach .42* .07 .17 .02 .18 .27 .06 .48* .10 .49*
Spit .52* .20 -.36 .41* .48* .10 .89* -.12 .53*
Blou-dis .21 -33« .40* .27 .27 .52* -.13 .20
Blou-lik" -.06 .15 .32* .22 .22 -.03 .36*
T-brushb -.44* -.39* -.24 -.34 .20 -.29
Cn-sug .52* .38* .40* -.15 .45*
Dog feces .21 .45* -.11 .43*
Vomit -.11 .06 .25
Bedpan -.07 .61*
Dart"'c: + -.08
Disgust score

Note. a Row measures are abbreviations of column measures, which are abbreviations of the items in the second column of Table 1. The measures
are as follows (row abbreviation in parenthesis): roach (see Footnote d); spit (spi) Soup: spit minus plain; blou-dis (bid) Blouse: dislike minus used;
blou-lik (bll) Blouse: used minus like; t-brush (tbr) Toothbrush: new minus like; cn-sug (ens) Label: cyanide minus sugar, dog feces (dfe) Fudge: dog
feces minus disc; vomit (vom) Rubber: vomit minus mat; bedpan (bpa) Soup: bedpan minus bowl; dart: + (dt+) (see Footnote c); disgust (dgt) Score
on 10 item disgust (contamination sensitivity) questionnaire.b For "positive" items, blouse like, toothbrush, and dart +, scores are calculated so that
positive magical effects yield negative scores. Thus, if positive and negative magic go together (e.g., dart + vs. blouse dislike comparison), the correlation
will be positive." The dart + score is the combined accuracy score for Kennedy and liked person (see Table I). d Roach overall refers to the net negative
effect of roach contamination, corrected for any effects of candleholder contamination. * r i abs(.30), p < .02.
 SYMPATHETIC MAGIC 709

positive and negative magic covary, correlations should be pos- and direct measurements. All three measures of negative con-
itive.) tagion, covering both interpersonal and object contamination,
There are substantial relations among the four negative sim- showed substantial effects. Five of seven measures of negative
ilarity items (rows 6-9 of Table 2). The six intercorrelations similarity showed significant results in the predicted direction.
between them vary from. 11 to .52, with a mdn r = .39. We note These effects include the cyanide label result, in which similarity
with surprise that the correlation between the dog feces and vomit is transmitted by a written word, rather than physical similarity.
scores, using similar types of disgust items and identical proce- The two results that failed to show significance (dart throwing
dures, was only .21, in contrast with a .52 correlation between at Hitler and disliked person) probably suffered from ceiling ef-
dog feces and the cyanide bottle. If we add into the similarity fects. Subject's motivation to perform on the dart task was high,
correlation matrix the positive similarity data (the mean accuracy and they were probably throwing as well as they could at a blank
score on Kennedy and liked person, combined), the coherence target, so that an enhancement with a negative target may not
of the category is weakened; three of the four correlations between have been possible. In addition, there may be some reluctance
dart and negative similarity are negative, and none are significant to throw darts at any pictured face (perhaps based on similarity
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(mdn r = .16 for combined similarity category). to living people, in general), so that there is no clear prediction
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Looking across laws of magic, we compared the three examples from the magical laws of the relative scores on negative faces and
of negative contagion (roach, spit-in-soup, and blouse-of-disliked- blank targets.
person) with the four instances of negative similarity (fudge dog Data illustrating positive manifestations of the laws are weaker.
feces, rubber vomit, cyanide label, and soup-in-bedpan). These Of two predicted cases for the law of contagion, one (blouse worn
correlations varied between .06 and .89, mdn = .40 (Table 1). by liked person) was significant in the predicted direction, and
The largest correlation in the matrix, .89, was between soup-in- the other (lover's toothbrush) was significant in the opposite
bedpan (negative similarity) and spit-in-soup (negative conta- (negative contagion) direction, although some subjects showed
gion). In general, the within-contagion or within-similarity cor- large enhancements in the positive direction. One might expect
relations were not notably higher than the between-contagion this effect to be largest for those currently in a romantic love
similarity measures. There were no significant correlations be- situation; many of our subjects probably did not qualify. The
tween positive and negative measures of contagion or positive positive blouse data can be questioned because the comparison
and negative measures of similarity. situation was a laundered blouse from a used clothing store.
Correlations between each of our measures and our previously Such a blouse has potential negative contagion properties, and
developed contamination sensitivity score (10 items, 8 of which this possibility may have influenced responses.
are clear cases of contagion without similarity, 2 of which include The evidence for positive similarity comes from the drop in
both contagion and similarity) were high for two contamination accuracy of dart throwing with the Kennedy and favorite person
measures (roach-in-juice, .49; spit-in-soup, .53) but also for three targets. There were no failures to attain positive similarity, but
of the four nonpersonal negative similarity measures (cyanide, we did not even try an example of possible nonpersonal similarity
.45; dog feces, .43; and soup-in-bedpan, .61) (Table 2). (e.g., enhancement in the value of mud as food if presented as
Anecdotal reports and our own measurements (Rozin, Fallon, chocolate pudding) because it seemed obvious that such manip-
& Mandell, 1984) indicate substantial individual differences in ulations would fail. In our culture, at least, it seems that the
disgust and contamination sensitivity, based on questionnaire interpersonal domain is where one would expect to find positive
results. We examined the issue of individual differences in magical magic effects. This is true of our meager data on positive con-
"thinking" in this study, by looking at the 11 measures displayed tagion and similarity. A possible exception is the positive con-
in Table 2, plus the degree of conditioning in the roach juice tagion effect associated with "lucky charms."
study, across all 50 subjects. We divided each measure into quin- Even within the interpersonal domain, negative effects (of
tiles, and assigned a quintile to each subject for each measure, contagion) appear more potent than positive effects. This posi-
with the lowest quintile always assigned to the scores representing tive-negative asymmetry seems to be part of a more general
the most magical outcomes. Three subjects were in the bottom tendency in humans and other animals to learn more rapidly
quintile on 8 (of 12) categories. The exact probability of 8 or and respond more strongly to negative events (Rozin & Zellner,
more bottom quintiles for any individual is .0006. One subject 1985; Rozin, in press). Indeed, it is striking that although disgust
had 7 bottom quintiles (exact p = .0039). The probability that is generally considered a basic "negative" emotion, there is no
4 or more out of 50 subjects would achieve an event of .0039 basic "positive" emotion that represents its opposite (Rozin &
probability is less than .0001. Hence we conclude that there is a Fallon, 1985; Rozin, in press). In the words of what we have
tendency for some people to be particularly sensitive to the laws been told is a Nebraska car mechanic, "A drop of sewage spoils
of sympathetic magic. As to insensitivity, only 1 subject had as a barrel of wine, but a drop of wine does nothing for a barrel of
many as 7 top quintile scores. The probability of one or more sewage." It is also possible that modern Western cultures devalue
such event out of 50 subjects is .177, a nonsignificant effect. the personal history of objects and hence selectively reduce pos-
itive magic effects, given that positive effects seem to depend
more on interpersonal factors. Positive contagion is prevalent in
Discussion some cultures, such as the Hua of New Guinea (Meigs, 1978,
1984), where the personal history of objects, in terms of contact
We have demonstrated results consistent with the operation with people with positive or negative relations to an individual,
of the laws of negative similarity and negative contagion across powerfully influences behavior. For example, the value of a food
a number of domains, including results from both questionnaires for a particular person is enhanced if it is spat on by someone
 710 P. ROZIN, L. MILLMAN, AND C. NEMEROFF

in a positive relation to that person. However, even among the experimenters) and people we discuss these findings with find
Hua, negative effects seem more potent. the patterns of thinking we report compelling in terms of personal
Our weakly significant results on "roach conditioning," allow experience; and (d) in the real world, events corresponding to
for interpretation in either Pavlovian or magical terms (discussed many of those measured do occur, and they seem to be responded
later). In either case, they offer a possible experimental approach to as indicated in our study. It would be desirable to collect solid
to the study of the transfer of value. Our finding of a negative evidence on this point. Finally, in a recent study (Nemeroff &
evaluative shift in a neutral object "paired" with a disgusting Rozin, 1985), we report evidence that American college students
object is supported by a substantial collection of anecdotal reports act as if they believe in the "you are what you eat" principle (an
of real-life situations in which aversions were established by the instance of contagion). In this study, we used an Asch (1946)
pairing of neutral items with disgusting stimuli (Rozin, in press). impression formation technique that cannot be faulted on the
The roach conditioning finding also resembles a response to a grounds of demand characteristics.
questionnaire item in a previous study (Rozin, Fallon, & Mandell. A second problem with this study has to do with interpretation.
1984). About 50% of adults reported that they would dislike to Although the results were predicted by the laws of sympathetic
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

eat some of their favorite soup if it had been stirred by a brand magic, they may well be consistent with other formulations. The
This document is copyrighted by the American Psychological Association or one of its allied publishers.

new fly swatter. This can be analyzed in conditioning terms. Along most likely candidate is the laws of association. Both Frazer and
with the roach findings, it can also be analyzed in magical terms: Mauss noted the parallel between the laws of sympathetic magic
a combination of the similarity principle (the brand new fly and the laws of association as propounded by the British empiri-
swatter looks like a used fly swatter) and contagion (the fly swatter cist philosophers. The laws of association vary in number de-
contacts and is then separated from the soup). Whatever the pending on the author, but the two laws propounded most fre-
interpretation, we have three lines of evidence suggesting a simple quently and consistently (Warren, 1921) are contiguity and sim-
transmission of negativity by some sort of contact with disgust. ilarity (Hume, 1748/1902; Mill, 1843/1930). According to Mill,
The experimental effect we report is small. However, the other the law of similarity holds that "similar ideas tend to excite one
data suggest that this type of process is important in the spread another" and the law of contiguity holds that "when two impres-
of disgust, and perhaps other evaluative changes. In our dem- sions have been frequently experienced (or even thought of) either
onstration, only one trial (pairing roach with juice) occurred, simultaneously or in immediate succession, then whenever either
and the charged element (or US), a dried, sterilized roach, may of these impressions or the idea of it recurs, it tends to excite the
not have been potent enough to have been maximally effective idea of the other."
under these conditions. We are examining this paradigm as a The implication of this "analogy" is that both sets of laws are
possible model system for the study of evaluative conditioning descriptions of fundamental patterns of human thought. But,
in humans. whereas the laws of association describe ways of linking thoughts
Results of our analysis of clusters of beliefs and individual to one another, and so remain inside the head, the laws of magic
differences proved somewhat disappointing, but correlations were describe practices that go a step further: people behave as though
generally comparable in magnitude to those found in many other they believe the world to be organized in the same way as their
personality domains. Although there is some tendency for magical thoughts. The idea that one can harm an enemy by burning his
thinking to extend across domains, there is no relation between fingernail parings does not follow from the laws of association.
positive and negative magical thinking, and the occurrences of The phenomena we have demonstrated can be explained by
negative contagion or negative similarity taken separately do not the laws of association. We have no direct evidence for the belief
cohere together any more than occurrences that cross the two by our subjects that their behaviors actually produced effects at
categories. There is great variation in response to almost all our a distance (e.g., that their dart throws actually hurt the targets).
measures of magical thinking, but individuals do not strikingly The law of contiguity can account for most instances that we
sort into consistently strong or weak magical thinkers. On the have attributed to contagion. The association of negative items
other hand, the distribution is far from random. Perhaps a broader (USs; e.g., dog feces or a disliked person) with objects (CSs, e.g.,
range of subjects, not restricted to a University community, would fudge or a sweater) can lead, by a process of evaluative condi-
reveal more striking and consistent individual differences in tioning (Martin & Levey, 1978; Rozin & Zellner, 1985), to de-
magical beliefs. valuation of the object. Associative accounts of backward cau-
The measures we have taken were determined by our interest sation contagion (not demonstrated in this study; but see Rozin,
in the laws of sympathetic magic. The results, although consistent Nemeroff, Wane, & Sherrod, 1985, for suggestive data) are subject
with these laws, need not be interpreted as strong support for to a similar, but more forced account. One's own sweater is de-
our claim that the laws operate in a number of domains in our valued when worn by a disliked person because that person (US)
culture. First, the data themselves can be questioned. It is very is associated with the sweater (CS). But the critical cognition
difficult to make the type of measurements we wish to make that it is your sweater is essential, and this already carries some
while disguising intent from the subjects. Some subjects may of the mark of sympathetic magical formulations. And, of course,
have perceived our interest in these measures, so that demand the idea (if present) that harm will come to you because your
characteristics might account for their results. We do not think enemy wears your sweater is not derivable from association prin-
demand characteristics are substantial factors in our results be- ciples.
cause: (a) all subjects showed some clear violations of magical In similarity, the rejection of fudge dog feces results from the
thinking; (b) in general, subjects are somewhat embarrassed about negative associations produced by the negative appearance of
the way they behave or the questionnaire responses they provide; the feces (US). However, the poorer dart performance with the
they recognize that what they are doing is irrational; (c) we (the positive faces is not easily explained by an associative account.
 SYMPATHETIC MAGIC 711

Most congenial to an associative account is the phenomenon tions of illness appear at about this same age (Bibace & Walsh,
we have described as roach conditioning. The procedure used is 1980).
Pavlovian: the CS is the juice, the US is the roach, the CS-US At a minimum, we have called attention to some patterns of
pairing is the roach in juice, and the new juice test represents thinking that have generally been considered to be restricted to
the CS alone. By the magical account, the negativity of the preliterate or Third World cultures. (See Shweder, 1977, for a
reached juice arises from contagion, and the negativity of the consideration of the possibility that the principle of similarity
new juice is accounted for as similarity (to the actually contam- holds in our judgments of personality, and Jahoda, 1969, for
inated juice). In this case, unlike some of the others, the condi- evidence that superstition, another possible manifestation of
tioning (associative) account seems less clumsy. "primitive" belief systems, is widespread in Western culture.)
Heider's (1958) balance principle provides another account We believe these patterns of thinking are not uncommon in adults
of at least some of the phenomena we have described. For the in our society, and that they can influence individual economic
case of fudge dog feces, the account is subject dislikes dog feces decisions (e.g., holding on to an old, malfunctioning car because
(negative sentiment), the piece of chocolate fudge is similar to of one's history with it), health decisions (e.g., avoiding a food
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

dog feces (unit relationship), hence subject dislikes the piece of because it looks like something offensive, or because it is asso-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

fudge (negative sentiment). There is a similar formulation for ciated with an undesirable name; or being reluctant to receive
contagion, if one can assume that the roach in the juice can be blood from a donor of another race, or being reluctant to give
described as a unit relation. As with associative accounts, the blood that might go to an AIDS victim; or practicing homeopathic
belief (or behavior) that suggests backward causation presents medicine), and tastes in things like clothing, food, or music. The
the greatest difficulty for balance theory. type of nonrational thinking we have described may extend recent
We do not think it is possible, with existing data, to decide descriptions of other failures of rationality in everyday thinking
definitively among the magical, associative, balance, or perhaps (e.g., Kahneman, Slovic, & Tversky, 1982; Nisbett & Ross, 1980).
other formulations. The associative framework is the oldest, and However, unlike the phenomena that are the focus of recent work
perhaps the simplest. The balance framework is the newest, but on biases and heuristics (e.g., the availability heuristic), our set
may be less complex than the sympathetic magic formulation. of phenomena seem to operate at a "gut level" that is not sus-
However, we believe three arguments favor a magical formulation. ceptible to the types of educational interventions that might re-
First, the phenomena in question flow directly from the laws of verse these other biases.
sympathetic magic, although they are retrospectively accounted We are now examining the important dimensions on which
for by the other theories. In spite of the great importance of an analysis of this type of thinking should proceed: the distinc-
associative theory in psychology, and the significant impact of tions between similarity and contagion, positive and negative
balance theory, the phenomena we have described have received effects, and forward and backward causation. At the same time,
almost no attention from psychologists, and are essentially un- it will be necessary to draw sharper distinctions between magical
known in modern psychology. Second, only the laws of sympa- and other accounts of the feelings and behaviors that we have
thetic magic can account for the projection of thoughts into the demonstrated.
world ("you are what you eat," or scorching an enemy's hair or
putting pins in his image to harm him). This clearly occurs in
References
traditional cultures, where the associative or balance accounts
will fail. It seems uneconomical to invoke different accounts for Angyal, A. (1941). Disgust and related aversions. Journal of Abnormal
that subset of these behaviors that we have demonstrated in and Social Psychology. 36, 393-412.
American culture. Third, the potency of animals and interper- Asch, S. E. (1946). Forming impressions of personality. Journal of Ab-
sonal contact in contagion follows easily from the magical idea normal and Social Psychology, 41, 258-290.
that "essence" is more abundant and powerful in animals. The Appadurai, A. (1981). Gastro-politics in Hindu South Asia. American

animal predominance finding can be handled in an associative Ethnologist, 8, 494-511.

Bibace, R, & Walsh, M. E. (1980). Development of children's concept
account, but only with an arbitrary additional preparedness as-
of illness. Pediatrics, 66, 912-917.
sumption.
Crawley, E. (1902). The mystic rose: A study of primitive marriage. Lon-
We do not know the origin of the type of thinking that we don: Macnullan.
have described. At one level it seems rudimentary, and hence Fallon, A. E., Rozin, P., & Pliner, P. (1984). The child's conception of
might be expected to be more common in children. This would food: The development of food rejections with special reference to
hold particularly for similarity; "the image equals the object" disgust and contamination sensitivity. Child Development, 55, 566-
can be viewed as a failure in differentiation. On the other hand, 575.
similarity may involve a learned set of conventions. The principle Frazcr, J. G. (1959). The new golden bough: A study in magic and religion
of contagion seems to presume two rather sophisticated notions. (abridged). New York: Macmillan (Edited by T. H. Caster, 1922; Orig-
inal work published 1890).
One is that matter is composed of tiny and invisible particles,
Heider, F. (1958). The psychology of interpersonal relations. New York:
an idea that is not present in children until about age 7 (Piaget,
John Wiley.
1974). The other is that two objects identical to the senses may
Hume, D. (1748). Enquiry concerning human understanding. London
differ significantly because of their individual histories. In this (Reprinted: Oxford, Clarendon Press, Oxford University Press, 1902).
light, it is not surprising that we have found that most children Jahoda, G. (1969). The psychology of superstition. Baltimore: Penguin.
under 7 years of age do not show contagion in disgust (Fallon et Kahneman, D., Slovic, P., & Tversky, A. (Eds.). (1982). Judgment under
al., 1984; Rozin, Fallon, & Augustoni-Ziskind, 1985), or that uncertainty: Heuristics and biases. Cambridge, England: Cambridge
others (e.g., Nagy, 1953) have found that germ-theory explana- University Press.
 712 P. ROZIN, L. MILLMAN, AND C. NEMEROFF

Martin, L, & Levey, A. B. (1978). Evaluative conditioning. Advances in How man chooses what he eats. A symposium (pp. 35-48). Zurich:
Behavior Research and Therapy, 1, 57-102. Forster.
Mauss, M. (1972). A general theory of magic. (Robert Brain, Trans.). Rozin, P., & Fallon, A. E. (1985). A perspective on disgust. Manuscript
New York: W. W. Norton. (Original work published 1902). submitted for publication.
Meigs, A. S. (1978). A Papuan perspective on pollution. Man, 13, 304- Rozin, P., Fallon, A. E., & Augustoni-Ziskind, M. (1985). The child's
318. conception of food: The development of contamination sensitivity to
Meigs, A. S. (1984). Food, sex, and pollution: A New Guinea religion. "disgusting" substances. Developmental Psychology, 21, 1075-1079.
New Brunswick, NJ: Rutgers University Press. Rozin, P., Fallon, A. E., & Mandell, R. (1984). Family resemblance in
Mill, J. S. (1843). System of logic. London. (Reprinted: London, Long- attitudes to food. Developmental Psychology, 20, 309-314.
mans-Green, 1930) Rozin, P., Nemeroff, C., Wane, M., & Sherrod, A. (1985). Sympathetic
Nagy, M. (1953). The representation of "germs" by children. The Journal magic in everyday life in American culture. Manuscript in preparation.
of Genetic Psychology, S3, 227-240. Rozin, P., & Zellner, D. A. (1985). The role of Pavlovian conditioning in
Nemeroff, C., & Rozin, P. (1985). Evidence for belief in "You are what the acquisition of food likes and dislikes. Annals of the New York Acad-
you eat" in American culture. Manuscript submitted for publication. emy of Sciences, 443, 189-202.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Nisbett, R., & Ross, L. (1980). Human inference: Strategies and short- Shweder, R. A. (1977). Likeness and likelihood in everyday thought:
This document is copyrighted by the American Psychological Association or one of its allied publishers.

comings of social judgment. Englewood Cliffs, NJ: Prentice-Hall. Magical thinking in judgments about personality. Current Anthropology,
Piaget, J. (1974). Understanding causality. New York: W. W. Norton. 18, 637-658.
Rozin, P. (in press). One-trial acquired likes and dislikes in humans: Warren, H. C. (1921). A history of the association psychology. New York:
Disgust as a US, food predominance and negative learning bias. Scribners.
Learning and Motivation.
Rozin, P., & Fallon, A. E. (1981). The acquisition of likes and dislikes Received July 5, 1985
for foods. In J. Solms & R. L. Hall (Eds.), Criteria of food acceptance: Revision received October 28, 1985 •