Reversal of Innate Aversions: Attempts to Induce a ... · Reversal of Innate Aversions: Attempts to...

Journal of Comparative and P1979, Vol. 93, No. 6, 1001-1014

ive and Physiological Psychology

Reversal of Innate Aversions: Attempts to Inducea Preference for Chili Peppers in Rats

Paul Rozin, Leslie Grass, and Geoffrey BerkUniversity of Pennsylvania

Although humans frequently develop preferences for innately unpalatable bit-ter or irritant substances, such preferences are extremely rare in animals. Anattempt was made to understand the nature of this difference by systematicexperiments with laboratory rats, with chili pepper as the unpalatable sub-stance. In parallel with major aspects of the human experience with chili pep-per, rats were exposed to it as a flavoring in all their food for periods up to 11mo from birth, without significant preference enhancement. Gradual intro-duction of chili into the diet also had no effect, nor did a series of poisoningand safety experiences designed to teach the rats that only chili-flavored foodswere safe to eat. A sequence of seven pairings of chili-flavored diet withprompt recovery from thiamine deficiency did significantly attenuate the in-nate aversion and may have induced a chili preference in at least one case.Extensive experience with chili did not reliably make rats much less sensitiveto its oral effects. The only reliable way to eliminate chili aversion in rats isto destroy their chemical irritant sense, which was accomplished in one groupof rats. It is concluded that in contrast to humans, it is extremely difficult toreverse innate aversions in rats.

Almost all animals have receptors thatcause them to avoid potentially noxiouschemicals (Beidler, 1975; Garcia & Hankins,1975; Rozin, 1976). Many substances thatare innately rejected by mammals producebitter (taste mediated) or chemical irritant(trigeminally mediated) sensations in hu-mans. Chemicals that stimulate either ofthese senses tend, in nature, to be harmful.Although there are a few species of verte-brates that may specialize in ingesting somesubstances with the chemical properties thatlead to these sensations (e.g., Hume, 1874),and/or that possess special detoxifyingmechanisms for associated toxins, almost allomnivorous vertebrates that have beenstudied avoid these innately unpalatabletastes.

This research was supported by National ScienceFoundation Grant BNS 76-80108 to Paul Rozin. Weexpress thanks to H. J. Grill and R. L. Solomon forcomments on the manuscript, to the Avoca Division ofthe R. J. Reynolds Tobacco Co. for providing some ofthe chili pepper, and to P. H. Todd of the KalamazooSpice Extraction Co. for determination of Scoville levelsof the chili pepper.

Requests for reprints should be sent to Paul Rozin,Department of Psychology, University of Pennsylvania,Philadelphia, Pennsylvania 19104.

Only one omnivorous vertebrate, our ownspecies, ingests large amounts of innatelyunpalatable substances, and obviously rel-ishes them. Widespread consumption ofcoffee, alcohol, tobacco, and chili peppertestifies to this, as does the more geograph-ically circumscribed consumption of manyother substances (e.g., ginger, quinine water,horseradish). Human infants, young chil-dren, and "uninitiated" adults typicallybehave like animals in rejecting these sub-stances (see Rozin, 1978, for a review andfurther discussion). Most adult humansreverse their natural rejection and end upwith strong positive affective responses to atleast one innately unpalatable substance.Gradual "mere exposure" to increasing levelsof these substances may often be sufficientto reverse the aversion (Rozin, 1978; Rozin& Schiller, Note 1). Both the apparentqualitative difference between humans andother omnivorous animals and the social andmedical importance of some of these innatelyunpalatable substances motivate furtherinvestigation.

Clarification of the distinction betweenpreference and affective value is prerequisiteto analysis of this problem. Animals or hu-

Copyright 1979 by the American Psychological Association, Inc. 0021-9940/79/9306-1001$00.75

1001

1002 P. ROZIN, L. GRUSS, AND G. BERK

mans may refuse foods because of antici-pated negative consequences of ingestionand/or because of hedonically negative sen-sations produced by the food. It has beenargued (Garcia, Kovner, & Green, 1970;Rozin & Kalat, 1971) that in taste aversionlearning, the poisoned food actually comesto taste bad, a hedonic change. Direct ob-servation of animals suggests that such tasteshave become aversive; behavior on exposure(spillage and facial gestures) resembles be-havior toward innately unpalatable quinine(Grill, 1975; Rozin, 1967). Similarly, on thepositive side, increased preference can beproduced by anticipated positive conse-quences of ingestion (what we call the"medicine" effect) and/or taste enhance-ment, a positive hedonic effect. Young(1948) emphasized a distinction similar tothis, with his concept of palatability tied tohedonic effects in need-free animals. Thefact is that nonhuman animals can rarely beinduced to show any sort of preference for aninnately unpalatable substance, and there isno current evidence that convincingly showsa hedonic shift, from negative to positive, forsuch substances.

It would be reasonable to appeal to theeffects of strong sociocultural forces that areuniquely human as an explanation of theease with which humans come to prefer in-nately aversive substances. Social pressuremay induce sufficient exposure to unpalat-able foods to allow changes of preference tooccur in humans. Such pressures would notnormally exist in nonhumans and hencewould deny them the exposure necessary toproduce the preference. This explanationis faulty on two counts. First, it it not anexplanation at all. It does not identify whatexposure might do to reverse the unpalat-ability of these substances. Social pressuremay cause people to try unpalatable foods,but what makes them come to like thesefoods? At this time, there are no convincinganswers. Second, it is contradicted by animpressive amount of animal data suggestingthat forced exposure to unpalatable sub-stances does not lead to preference.

It has proved difficult to turn even a neu-tral taste stimulus into a highly preferredone, though this can be done (see Rozin &Kalat, 1971; Zahorik, 1977, for general dis-cussions). This contrasts with the ease in

establishing acquired aversive tastes. Theacquired positive preferences reported inanimals have usually been weak. They re-quire a number of conditioning trials, andthey extinguish or dissipate easily. Whatbetter evidence for this than the well-knownfact that rats raised in the laboratory on ratchow, and experiencing many repletionsfrom this chow after deprivation, will aban-don it in favor of a new, more palatable dietafter but a few days of choice. The weakacquired preferences that have been dem-onstrated could be interpreted to be me-diated by "anticipation" of consequences(medicine effects) rather than hedonicshift.

There have been few experimental at-tempts to produce preferences for innatelynegative tastes. Mere exposure of infantrats to bitter tastes did produce a temporarypreference (Warren & Pfaffmann, 1959). Amore marked preference following mere ex-posure of rats to a garlic solution has beenreported (Capretta & Rawls, 1974). How-ever, this solution was only slightly aversivein comparison with water. Long exposureto some strong (irritant) spices did not sig-nificantly enhance the preference of rats inanother study (Hilker, Hee, Higashi, Ike-hara, & Paulsen, 1967). Two studies ex-plicitly attempted to induce preferences byassociation of mildly aversive substanceswith positive events. Pairing of an aversiveHC1 solution with food in hungry rats ledto a transitory reduction in the aversion(Siqueland, 1965). Pairing of a slightlyaversive bitter-sweet solution with recoveryfrom morphine withdrawal produced asubstantial preference for this solution(Parker, Failor, & Weidman, 1973). Thispreference was demonstrated under condi-tions that would suggest a medicine effectrather than a hedonic shift. A few inbredstrains of mice (McClearn, 1973) and someindividual rats (Richter, 1941) showed apreference for moderate levels of ethanol inwater over plain water, which suggests thepossibility of a preference for the bittercomponent of ethanol. In these cases, thereis some question whether the preferences areacquired or innate. If acquired, they couldrepresent hedonic shifts, but they might alsobe medicine effects.

There are a few suggestive instances of

REVERSAL OF INNATE AVERSIONS 1003

acquired preferences for innately unpalat-able substances in nonhuman primates.Weiskrantz and Cowey (1963) reported thatrhesus macaques can acquire a preference forinitially rejected black currant juice. Kluver(cited in Kalmus, 1969) described voluntaryingestion of quinine powder by rhesusmonkeys.

It is possible that the many failures to es-tablish oral preferences for addictive sub-stances in animals, especially alco.hol (Mello,1973; Richter, 1956), have foundered on thefact that most of these substances are in-nately unpalatable. A recent study dem-onstrated a rather substantial preference fora neutral tasting solution that was pairedwith intragastric alcohol repletion in alco-hol-withdrawn rats, which suggests that thetaste of alcohol is a significant barrier topreference development (Deutsch & Walton,1977).

This study represents a concerted attemptto establish a preference for an innately un-palatable substance in rats. The substanceselected is chili pepper, a chemical irritant.In spite of the fact that chili pepper is one ofthe most widely consumed spices in theworld and is probably consumed on a dailybasis by about one fourth of the adults in theworld, there have been no studies on thenature or development of this preference(see Rozin, 1978). Chili peppers are rejectedby young children and uninitiated adultsaround the world and, according to anec-dotal evidence, by virtually all omnivorousanimals. Foods seasoned with chili may beeaten by some omnivorous household ani-mals, but there is no convincing evidencethat they like or prefer foods so seasoned.Details on human consumption of chilipepper and the pharmacological effects ofcapsaicin, the substance that causes the ir-ritation, can be found in a chapter by Rozin(1978; see also Rozin & Schiller, Note 1).

This series of studies is designed either toproduce an animal "model" for a pervasivehuman activity or to indicate some of thereasons for this major difference betweenhumans and animals. We also speculateabout possible reasons for chili ingestion andpreference in humans and on the long-termeffects of chili ingestion. It is hoped thatunderstanding of the development of apreference for this particular innately un-

palatable substance will contribute to ourunderstanding of the general process of ac-quisition of food preferences and, in partic-ular, the reversal of innate rejections.

General Method

All subjects were albino laboratory rats (CharlesRiver CD strain), either raised in the laboratory or ob-tained at ages 6 mo to 1 yr. They were housed in atemperature-controlled animal room, with a 14:10 hrlight/dark cycle. They were occasionally housed incommunity cages after they were individually marked.All testing was done in standard 25 X 20 X 18 cm wirecages, with water ad lib and food choices as indicated.The basal diet for all animals unless otherwise indicatedwas powdered Purina Rat Chow. The major additivein these experiments was ground chili pepper.1 Allfood-choice measurements were based on daily orhourly weighings of the contents of two glass 2-oz. (57-ml)food cups, attached to the front wall of the cage by wireloops. Paper under the cages was used to collect spil-lage. Both cups and spillage were weighed to thenearest .5 g. Food-cup positions were alternated daily.Prior to any preference test, rats had at least 2 days toadapt to the two-cup presentation format, with the same(basal) food in both cups. Animals were weighed dailyduring preference test periods and occasionally in pe-riods between tests.

Experiment 1: Exposure, Learned Safety,and Chili Preference

The simplest interpretation of the genesisof human chili preference is that it resultsfrom long-continued exposure to the spice.Although this type of procedure has notworked in prior studies with animals withother unpalatable substances, it is so basicthat it was repeated here with longer expo-sures than in previous studies. The chililevel in the diet used produced a piquancycomparable with that of dishes in manychili-eating cuisines (e.g., Mexico, west Af-rica, Ethiopia, or parts of India and China).In one group of animals, the exposure effectswere supplemented by a "safety" procedure.Sickness was induced in rats raised on chilidiets whenever they consumed a diet that

1 The chili pepper used in Experiments 1,2,4, and 5was purchased locally, in bulk, under the name of"cayenne pepper." This pepper had a pungency of40,000 Scoville units. The pepper used in Experiment3 came from the Avoca Division of the R. J. ReynoldsTobacco Co. and was rated with a pungency of 100,000Scoville units. One Scoville unit is the human detectionthreshold under a set of specified conditions.


was not flavored with chili. In this way, chiliflavor might have become preferred as asafety signal. On the one hand, this proce-dure seemed like one that would successfullyinduce preferences for a neutral diet, on thebasis of the learned aversion literature andprinciples of generalization. On the otherhand, it is conceivable that distinctive fla-vorings that are used repetitively and per-vasively in particular cultures may serve thefunction of labeling food as safe (Rozin, 1976,1978).

Method

All subjects were born in the laboratory within a10-day period. Control rats (n = 13) were from twolitters, and exposure (n = 11) and safety (n = 11) ratswere randomly selected from three other litters. Twoweeks before mating, exposure and safety mothers wereplaced on a diet of 15% chili pepper and 85% powderedchow. Pilot studies indicated that rats would maintainthemselves and grow normally on this diet as the onlysource of food, after a brief rejection period lasting, atmost, a few days. We wished to have the mothers ad-justed to the diet before fertilization. All pups wereweaned and marked at 30 days of age and placed incages with their littermates. They were transferred toindividual cages a few days before testing periods.Exposure rats were offered only the 15% chili pepperdiet, except in choice tests, through the 11-mo durationof the experiment. Thus, except for mother's milk, theyhad no food other than "hot" chow. They may con-ceivably have experienced "hot" milk, if any of thecapsaicin (the source of piquancy) from the chili ap-peared in the milk. Safety rats had a similar experienceexcept for the addition of one or two periods of 1-2 wkeach during which they received the safety treatmentsdescribed below. Controls were raised on powderedchow without chili and experienced chili only duringchoice tests.

At 8 wk of age, somewhat more than half of the ratsin each group experienced a 4-day preference test be-tween powdered chow and this same chow with 15% chilipepper. After this test, these rats were returned togroup cages with other members of their group. Allanimals were tested at 7 mo of age (see Table 1 for a planof the experiment). The 7-mo test was like the 8-wktest except that it included all the rats, some that wereexperiencing chili (or plain powdered chow) for the firsttime and others that had had the choice for 4 days at theage of 8 wk. After the 7-mo test, rats were returned tothe group cages and maintained on their original diets.Some participated in other preference tests or obser-vational studies, some of which are described below. At11 mo of age, seven control, four exposure, and foursafety rats were retested on a 6-day test of preferencebetween powdered chow and the same with 1% chilipepper. This much lower concentration was clearlydifferent from plain chow to the human palate but wasonly slightly piquant. The 15 rats retested with thelower concentration experienced, at most, 10 days with

diets other than their basic rearing diet in the periodbetween 7 and 11 mo.

The safety groups were treated identically to theexposure groups, except for periods of safety trainingbefore the preference tests at 8 wk (if administered) and7 mo (see Table 1). The first safety period occupiedDays 38-46 for the seven animals that were tested at 8wk of age. The second series was given during Days185-198 to all 11 animals prior to the 7-mo preferencetest. The basic idea of the chili-safety experience wasto give slightly food-deprived animals experience witha number of distinctively different novel diets. Thosecontaining chili led to normal repletion, whereas thosewithout it were followed by LiCl-induced illness. Thesecond safety-training sequence (Days 185-198) is de-scribed below in some detail, and variations from it inthe first sequence (Days 38-46) are mentioned.

Three entirely new diets were employed. Each dietcame in two forms: 15% chili added or "plain." Thechili diet always contained 1% NaCl (a taste control forLiCl), and the nonchili diet contained 1% LiCl, whichreliably produced learned taste aversions at this level.The diets can be briefly described by their most dis-tinctive ingredients, as bouillon, Crisco, and butter-milk.2

For each type of diet, the procedure lasted 4 days.Rats were placed on a food-deprivation cycle that al-lowed them 1-hr access to food twice each day, ap-proximately 6 hr apart. On the first day of the cycle,they were offered a new diet with 15% chili. On thesecond exposure that day, the alternate nonchili andpoisoned diet was offered. On the second day, recoverywas permitted, while the standard 15% chili chow wasavailable during both feeding periods. Day 3 was atraining session equivalent to Day 1, but with the orderreversed: The LiCl version of the same new diet wasoffered in the first meal of the day, and the nonchiliversion later in the day. Day 4 was a recovery day, likeDay 2, and completed the cycle. Food intakes weremeasured to the nearest .5 g on all training days. This4-day cycle was carried out in succession for each ofthree diets in the order: buttermilk, Crisco, bouillon.The rats were returned to the 15% chili diet for a fewdays until the subsequent preference test.

The first safety training procedure for the seven ratstested at 8 wk was identical to the procedure describedabove except that recovery days were not inserted be-tween each of the training days. As a result, in somecases rats did not ingest much of the safe chili diet, orthe LiCl diet, on the last of the three new diets, on ac-count of an apparent accumulation of the effects of LiClpoisoning. Thus, a few of the rats in the first safetytraining cycle may have received four or five effectivepoisoning-safety experiences instead of six. For thisreason, the recovery days were inserted in the secondround of safety-poisoning experiences. Overall, then,seven safety rats received two safety sequences, and fourreceived only the second sequence.

2 The diets were constituted as follows: Bouillondiet—15% chicken bouillon, 35% powdered milk, 35%corn oil, 15% sucrose; Crisco diet—40% Crisco, 50%wheat flour, 10% sucrose; buttermilk diet—100% but-termilk.


Table 1Plan of Experiment 1

Group

Age (in wk) Control Exposure Safety

0-67

9-2627-28

29

30-46

47

chowchow

chowchow

chow

15% chili chow15% chili chow

15% chili chow(12 safety-poison trials8)

(Preference tests: chow vs. 15% chili chow [4 days]8)

15% chili chow 15% chili chow15% chili chow 12 safety-poison trials

Preference tests: chow vs. 15% chili chow [4 days]

15% chili chow 15% chili chow

Preference test: chow vs. 1% chili chow [6 days]

1 Animals first tested at 7 mo (Week 29) remained on their basal diets through Weeks 7 and 8.

Results

Exposure rats grew well on the 15% chilidiet. Their weights lagged a little behindthose of controls from before the time ofweaning (see Table 2). By 10 mo of age, 3 ofthe 13 control rats had died, as against 1 of11 exposure rats. These results speakagainst significant toxicity of chili pepper (asdoes the existence of over 60 million Mexi-cans). The total of 6 or 12 poisoning expe-riences did not have a marked effect on thehealth of the safety animals, even thoughcoupled with virtually exclusive exposure tothe 15% chili diet (Table 2). Only 1 of the 11safety rats died by 10 mo of age.

The LiCl was clearly effective in bothpoisoning-safety cycles, as judged by de-pressed food intake on both the first andsecond day of exposure to each LiCl diet.

Table 2Growth of Rats in the Three Groups ofExperiment 1

M weight (in g)

Group Age; 1 mo Age: 10 mo

MaleControlExposureSafety

FemaleControlExposureSafety

658

763

78.771.462.2

68.458.664.0

463407387

303239262

Because of the imposed recovery days in thesecond cycle, the effect was clearer here.The effectiveness of the procedure is mostobvious in the increased relative avoidanceof the nonchili diet the second time it waspresented (Day 3 vs. Day 1 of the cycle foreach novel diet). For the first (buttermilk)diet, 33% of the food on the first training daywas from the nonchili diet, and only 8% onthe second training day. For the Crisco diet,the corresponding numbers are 49% for thefirst trial and 19% for Day 3, and for thebouillon diet, the figures are 66% and 16%,respectively. Although an aversion is clearlyestablished to the nonchili version of eachdiet, these data do not provide direct evi-dence that the rats were gradually learningto avoid any diet not containing chili and/orto treat chili as a safety signal.

The 8-wk preference test (Table 3) re-vealed a strong avoidance of the chili diet by

Table 3Eight-Week Preference Test: 15% Chili DietVersus Powdered Chow

M% of 15% chili diet eaten

Group Dayl Days 3 and 4a

Control 6Exposure 7Safety 7 13.

Note. All significance estimates are based on a Mann-WhitneyU test (one-tailed).• Mean intake per day for Days 3 and 4.*p<.05. »*p<.01. ***p<.001.


Table 4Seven-Month Preference Test: 15% ChiliDiet Versus Powdered Chow



ControlExposureSafety

131111

19.4 J22.6 I-

2.72.35.3

Note. All significance estimates are based on a Mann-WhitneyU test (one-tailed).a Mean intake per day for Days 3 and 4.* p < .01.

Table 5Eleven-Month Preference Test: 1 % ChiliDiet Versus Powdered Chow


Group

ControlExposureSafety

n

744

Dayl

6.1 -Ll20.8 J *31.8 J

Days 5 and 6a

3.6 18.3 *

18.5 J

Note. All significance estimates are based on a Mann-WhitneyU test (one-tailed).fl Mean intake per day for Days 5 and 6.*p <.01.

all groups. The greater acceptance of thisdiet on Day 1 by exposure and safety groupsmay be entirely attributable to a neophobicreaction to the plain chow, which thesegroups had never experienced before.However, in spite of this, on their first daywith this new diet, these rats still consumedless than 50% of their food in the form of thechili diet. By Days 3 and 4, there was nosignificant difference between exposure andcontrol rats (Table 3). However, there issome evidence for attenuation of the aver-sion of the safety group (Table 3), althoughthey still showed a substantial aversion tochili (13% of the total eaten in the last 2 dayswas from the chili diet).

The 7-mo test included both these 8-wkanimals and those not tested at 8 wk. Thedata from both of these subgroups were

70

50

- 7 Month Test: Combined GroupsD- -o Exposure N = 11

•—« Control N = 13

o—o Safety N-11

•= 10.co~ 30

5 20Q_

10

Days

Figure 1. Daily intake of powdered chow with 15%chili pepper versus plain powdered chow over 4 days,with daily readings. (Mean percentage of the chili dieteaten for each day for all rats in Experiment 1, at 7 moof age.)

combined to evaluate the results, into threemajor groups: control, exposure, and safety.As indicated in Table 4 and Figure 1, thereis a strong chili aversion in all three groupsand a virtually identical avoidance of thechili diet after the first day. On the first day,exposure and safety rats consumed morechili diet than did controls, an effect proba-bly attributable to neophobia.

In the final retest at 11 mo (Table 5), witha much lower concentration of chili, the samepattern of results is seen. There was ahigher intake of the slightly piquant diet inexposure and safety rats on Day 1, thoughnot an absolute preference for it. The ex-posure rats avoided this diet strongly (8%)and at roughly the same level as controls byDays 5 and 6, the last 2 days of testing. Aswith the 8-wk data, there was a strong butsignificantly weaker aversion in the safetyanimals at Days 5 and 6 of testing. In gen-eral, exposure to chili for 8 wk to 11 mo hadno effect on chili preference after 1 day ofchoice. Exposure plus 6-12 safety-poisonexperiences led to a significantly reducedaversion in two of the three tests. However,in all three tests, the safety group showed asubstantial aversion to the chili diet at theend of the testing periods.

Experiment 2: Effectsof the Gradual Introduction of Chili

The exposure procedures used in Experi-ment 1 do not parallel the exposure humanchildren normally receive in chili-eating cul-tures, because these children are exposed togradually increasing amounts of chili (Rozin,


1978; Rozin & Schiller, Note 1). In this ex-periment, we incorporated this feature in therearing of a group of rats. They were rearedon very low levels of chili pepper and weresubjected to gradually increasing concen-trations.

Method

Fourteen rats, born in the laboratory, were weanedat 28 days of age. Powdered Purina Rat Chow with 1%chili pepper was the only food available to the motherand to the offspring. The rats were divided at weaninginto two groups of seven rats each. The rats of eachgroup were individually marked and housed as separategroups in two community cages. The constant groupreceived 1% chili diet until testing began on Day 60.The gradual group received 1% chili diet for their first6 days postweaning and then an increased chili con-centration every 6 days, covering the values of 2%, 4%,10%, and 15%. Two days prior to testing, on Day 58, allanimals were placed in individual cages with two cupsof the appropriate diet (1% chili for the constant group,15% chili for the gradual group). On Day 60, a 6-dayalternating side preference test was begun, with a choicebetween 1% chili and plain ground chow. Note that thechili concentration used in the test was much lower thanthe highest level experienced by the gradual group.

Results

As can be seen in Table 6, neither groupshowed evidence of a preference for chili, norwas there a significant difference betweenthem at any time during the 6-day tests.There was, as we have seen before, a muchweaker avoidance than in control animals(from Experiment 1, 11-mo test with 1%chili) on the first day. For the last 2 days ofthe 6-day test, gradual and constant groupsshowed slightly weaker avoidances thancontrols. All groups showed a strong aver-

Table 6Effect of Gradual Versus Constant Exposureto Chili Pepper onl% Chili Diet Preference



ControlGradualConstant

777

41.4 J***28.9

1 3'61* 1U * 10.7 J *

J 7.6 J

Nate. Control data are from Experiment 1. All significanceestimates are based on a Mann-Whitney U test (one-tailed).a Mean intake per day for Days 5 and 6.* p < .05. ** p < .01. *** p < .001.

sion to a mildly piquant diet, with a signifi-cantly weaker aversion in both groups pre-viously exposed to chili. The absence of anydifference between gradual and constantgroups is particularly striking when it is re-alized that the gradual groups came downfrom a 15% chili diet to the 1% chili choicewhereas the constant group had never hada diet with more than 1% chili.

Experiment 3: Association of PositiveVisceral Effects With Chili Ingestion

The most straightforward method for in-creasing a preference for a substance wouldbe to follow its ingestion by positive visceraleffects. One could claim that this is exactlywhat occurs each time the rat on an exposureregimen eats a meal of chili diet. However,it might be that the repletion effects are notsufficiently salient and/or that rats eat mealsbefore they experience significant hungerand hence experience few direct positivevisceral effects. In this experiment, a con-certed attempt was made to induce a positivepreference for chili pepper by associating itwith recovery from thiamine deficiency, avisceral event that has been successfully usedto increase food preferences (Garcia, Ervin,Yorke, & Koelling, 1967; Zahorik, 1977;Zahorik, Maier, & Pies, 1974). Althoughrecovery from deficiency does not corre-spond to any obvious recurring event thatcan be assigned to most human chili eaters,the occurrence of significant postingestiveeffects in these cases cannot be ruled out.

Method

Four male and eight female rats, aged 90 days, wererandomly assigned to one of two basic groups: recoveryand control. The basic design, as initially described byRozin and Kalat (1971) and first successfully employedby Zahorik et al. (1974), involved the use of three dis-tinctively different diets: safe, deficient, and recovery.The safe diet (S) was eaten with normal repletion effectsin the first stage of the experiment. Animals then wentthrough repeated cycles in which they became thiaminedeficient on a deficient diet (D) and recovered fromdeficiency while eating a recovery diet containing chili(Re). The critical test for preference acquisition wasbetween the recovery and the safe diets. The criticaltest for chili preference (chili being one feature of thedistinctive recovery diet) was a choice between two dietsthat were identical except that one contained chili.


Table 7Effects of Association of Chili Pepper With Recovery From Deficiency

Test No.

Variable 10 11 12

Choice of dietsTest lengthStateRecovery group"Control group"SignificanceBest ratc

Rc-SI drec5419**

89

Rc-S4 hdef7522*

89

Rc-S4drec6417**

99

Rc-R2drec3413*

20

Pc-P2drec2410—52

Sc-S2drec345*

75

Rc-S2drec298***

47

Rc-R2drec143—18

Sc-S2drec246*

46

Rc-SI hdef46b

36—77

Sc-SI hdef34b

6—95

Rc-RI hdef36b

17—88

Note. A 2-mo break followed Test 6. R = recovery; S = safe; P = powdered chow; lower case c indicates containing chili; d =days; h = hours; rec = recovered; def = deficient. All significance estimates are based on a Mann-Whitney U test (one-tailed).a Data are mean percentage of chili diet eaten. The n = 6 for each group.b One recovery rat died. These values represent five rats.c All data are from the rat in the recovery group with the highest overall chili preference. All numbers represent data from thesame rat.*p<.05. **p<.01. ***p<.001.

The diets3 used in this experiment were all thiaminedeficient, with the exception of a few days of preferencetesting with powdered chow in the later part of the ex-periment. For control and recovery rats, the deficientdiet was always a powdery, starch-based diet. For halfof the rats in each group, the safe diet was a granular,sugary diet and the recovery diet was a Crisco-basedbland diet with 2% chili pepper added (see Footnote 1).For the other half of the rats in each group, the situationwas reversed: The safe diet was the Crisco-based diet,and the recovery diet was the sugar-based diet with 2%chili pepper. The two critical diets were approximatelyequal in palatability, when the recovery diet was withoutchili pepper.

In the first stage of the experiment, all rats received7 days of exposure to the appropriate safe diet. Sincethis diet was thiamine deficient, there was the possi-bility that over this 7-day period, some subtle signs ofthiamine deficiency might appear (though no overt signsappear for over 2 wk). As assurance against this pos-sibility, each rat was injected with 100 ̂ g of thiamineHC1 on Day 3 of this regimen.

The second stage of the experiment consisted of aseries of seven deficiency-recovery cycles. Recoveryrats were placed on the deficient diet and continued onthis regimen until all rats in the group had lost weightfor 2 consecutive days. When this criterion wasreached, these rats were all offered the recovery diet4

in a single cup (alternated in side from one cycle to thenext). Ten minutes after this introduction, each rat wasinjected sc with thiamine HC1, at a dose level of 200Mg/kg. Rats were allowed access to the recovery diet forthe following 24 hr. They were then returned to dietD and entered the next deficiency-recovery cycle. Onat least one occasion during each deficiency period, therats were injected sc with approximately .5 ml of waterso that they could not learn to associate the injectionprocedure with recovery. Control rats received thesame diet sequence as recovery rats but were not al-lowed to become thiamine deficient. They were in-

jected with 100 /ig of thiamine three times every week.At the same time that recovery rats qualified for a re-covery trial, the control rats were offered the same re-covery diet and were injected with thiamine in corre-sponding doses 10 min after this diet was made avail-able. This cycle continued six times.

The third stage involved preference testing. Pref-erence testing might have been ideally carried out whenthe animals were deficient, since it was in this state thatthe chili diet produced recovery. However, a preferencetest in this situation would also be an extinction trial,since thiamine could not be injected during the test (itmight follow ingestion of diet S). For this reason, onecompletely "clean" preference test was carried out afterthe rats had recovered from the sixth deficiency period.The day following recovery, rats were offered a 24-hrchoice between diet Re and diet S. They were thenallowed to become deficient again (the seventh standardcycle) on diet D. However, when they reached the de-ficiency criterion, they were given a 4-hr choice, whiledeficient, between diet Re and diet S. They were thenreturned to diet D for the remainder of the following24-hr period, and they experienced a standard recoveryprocedure on the next day, which constituted the sev-enth recovery trial.

3 The basal deficient diet contained, in grams perkilogram, sucrose, 20; starch, 600; Mazola, 50; casein,100; egg-white solids, 189; Hegsted salt mix, 40; GeneralBiochemicals thiamine-deficient vitamin mix, 1. Bothother diets contained the same levels of salt and vitaminmixes. In addition, the sugar-based diet containedsucrose, 659; Mazola, 50; casein, 250. The Crisco-basedbland diet contained sucrose, 20; starch, 434; Mazola,5; Crisco, 250; casein, 250.

4 In the first cycle a lower chili pepper level of .5% wasused on the recovery day. This was increased to 1% inthe second cycle and to 2% in the third and all subse-quent cycles.


Following this seventh cycle, the rats were extensivelytested for diet preferences, while recovered. Duringthis period, all rats were injected sc with 100 jig of thi-amine three times each week. The sequence of pref-erence tests is indicated in Table 7. All tests involveda choice between a chili-free diet and a diet with 2%chili. The first test was a 4-day choice between diet Reand diet S. This was followed by a 2-day choice be-tween the recovery diet vehicle (R) and the same dietwith chili added (Re) to determine the extent to whichthe taste of chili per se, as opposed to other propertiesof the recovery diet, acquired positive value. The pal-atability of chili was further tested in 2-day choicesbetween powdered chow (P) and the same with 2% chilipepper (Pc), and then diet S with (Sc) or without (S)chili pepper. Following these tests, the rats were kepton Purina Checkers for 2 mo. They were then retestedon three of these tests: Re versus S, Re versus R, andSc versus S. Finally, all rats were placed back on dietD. Recovery rats were allowed to become thiaminedeficient, while control rats were injected (as before)with thiamine. When the recovery rats met the defi-ciency criterion, all rats were given 1-hr preference testson 3 successive days between the same three basicchoices: Re versus S, Re versus R, and Sc versus S.The recovery rats were thiamine deficient during allthree brief tests.

Results

The deficiency-recovery cycles were ef-fective in every case. Every recovery ratshowed a large weight gain on every recoveryday. The cycles on diet D varied from 5 to7 days. Because experiments of this typehave shown weak, if any, effects, we inten-tionally used as a primary test, pairings ofsafety and recovery diets that differed alongmany sensory dimensions, only one of whichwas chili pepper. On this test, whether ratswere briefly tested while deficient or forperiods of a day or more while recovered,there was a substantial and significant in-crease in preference for diet Re (see Table 7for these and all other results). On all threetrials with this pairing of diets, before the2-mo break in the experiment, average in-take of the recovery diet was above 50% oftotal intake, which suggests an actual pref-erence. This effect was clearest in the de-ficient situation (75% Re preference in re-covery rats vs. 22% in controls). This is oneof the largest effects observed in the acquiredpreference literature: Its size is enhancedbecause of the unpalatability of diet Re, andhence the low intake of this diet in con-trols.

Subsequent testing in recovered rats iso-

lated the effects of chili per se, pairingidentical diets except for the presence orabsence of chili. Results were similar in allthree tests in the initial round of testing(Tests 4, 5, and 6 in Table 7): There was anattenuation of the aversion to the chili dietbut far from an absolute preference. Thiseffect was notably weaker than the Rc-Sbasic choice and suggests that much of theenhanced preference in the basic tests is re-lated to the nonchili distinctive componentsof the recovery diet. Retesting after 2 mo(Tests 7-12, Table 7) revealed the samepattern of results but with a further atten-uation of chili-diet preferences in all tests.As before, the basic choice (Re vs. S) pro-duced the greatest attenuation in aversionto chili, although not an absolute preference(29%). Testing during deficiency (Tests10-12) enhanced the preferences or, morecorrectly, further attenuated the chili aver-sion in the recovery group. Although moststatistical tests for recovery-control differ-ences (Mann-Whitney U, one-tailed) do notyield acceptable significance in the 2-moretests, all six tests show a superiority of atleast 10 percentage points for the recoveredgroup.

Variability was high in both recovery andcontrol groups. Table 7 shows the data fromthe recovery animal having the strongestacquired preference for chili. Note thatunder deficiency conditions some monthsafter training, this rat shows an absolutepreference for the chili choice, even when thepairing involved only a difference in chili (Revs. R or Sc vs. S).

In summary, (a) seven pairings of chili dietwith recovery from thiamine deficiency at-tenuated the aversion to chili pepper; (b) thisattenuation was even stronger when testingoccurred during a deficiency state, when therecovery properties of the chili diet would beexpected to be manifested; (c) a major por-tion of the enhanced preference effect is at-tributable to distinctive aspects of the re-covery diet other than chili; and (d) one an-imal showed an absolute preference for chiliwhen in the deficient state.

We conclude that it is possible to signifi-cantly weaken a chili aversion. A positivepreference may appear when the animal is inthe negative physiological state that chili


diet has, in the past, alleviated. It is quitelikely that longer experiments, involving tensor hundreds of recovery cycles, could pro-

.duce more stable, absolute acquired prefer-ences. The effect might be further en-hanced by limiting the difference betweenrecovery and safe diets to merely the pres-ence of chili in the former. The questionremains whether these predicted prefer-ences, or the more modest preferences ac-tually demonstrated here, represent hedonicshifts or medicine effects. The considerableenhancement of preference during defi-ciency suggests a medicine-effect compo-nent.

Experiment 4: Effects of Long Exposureto Chili on Sensitivity to Chili

It would be surprising if up to 1 yr of in-gestion of food containing significant.amounts of a chemical as potent as capsaicinshould have no measurable effects (Jancso-Gabor & Szolcsanyi, 1969; Maga, 1975). Onthe basis of estimates from the literature(Osol & Farrar, 1955), these rats were con-suming, based on daily intakes of 20 g (andassuming a capsaicin level of from .15 to 1.5mg/g of diet), 3-30 mg/day of capsaicin.Since this diet was comparable in hotness tochili-flavored foods in chili-eating countries,it can be assumed that these rats were con-suming more chili (per kilogram) than hu-mans do.

Capsaicin has significant effects onchemical irritant sensitivity in rats and hu-mans (Jancso, 1960; Jancso-Gabor & Szolc-sanyi, 1969). Work from the laboratory ofJancso and Jancso-Gabor in Hungary indi-cates that topical applications of solutionscontainin capsaicin in the range of .5%-!%to rat or human skin or tongue results in adesensitization that lasts more than a day,as measured by markedly raised thresholdsto several different chemical irritants.These concentrations are 5-60 times higherthan the 15% chili diet used here, but expo-sure was over much greater periods in ourstudies. Furthermore, systemic injectionsof as little as 50 mg/kg capsaicin into a ratcan totally desensitize it to all chemical ir-ritant substances for weeks or months.These rats showed no response to strong ir-

ritants, such as capsaicin, mustard oil, orammonia, placed on their eyes or ears. Therats in this experiment were consumingcapsaicin at the rate of about 9-90 mg/kg/day over hundreds of days. Of course, we donot know how much is absorbed, but it is, onthe face of it, surprising that so little changein avoidance is seen. One might well expecta loss of sensitivity, as one might well expectto see the same in humans living on chilidiets. Indeed, loss of sensitivity to the irri-tant effects of chili is a possible partial ex-planation for the development of chili pref-erence in humans.

In this experiment, sensitivity to chili wasstudied in terms of avoidance threshold: thelowest chili concentration that a rat wouldavoid, in comparison with the same dietwithout chili. These data provided sug-gestive data on desensitization and a verysensitive test for preference changes thatchili exposure might produce. Rats fromExperiment 1, raised on 15% chili for at least10 mo, were compared with rats of about thesame age that had never experienced chilipepper.

Method

Twelve rats from Experiment 1,11 mo of age, servedas experimental subjects. With the exception of par-ticipation in a few preference tests, lasting less than atotal of 20 days, during which they had access to non-chili-flavored diets, the rats had spent their entirepostweaning life on a diet of 15% chili pepper in pow-dered chow. Twelve rats of approximately the same age(6-12 mo) were taken from the rat colony and used ascontrols. These rats had been raised on PurinaCheckers. One experimental rat died during the ex-periment. All rats were housed individually and testedin the standard two-cup choice situation.

The basic unit of preference in this study was the2-day choice between two diets, with intakes measureddaily and cup position switched daily. Rats were testedon a staircase method, with a choice between plainpowdered chow and the same diet with a measuredamount of chili pepper. If a rat avoided the chili diet(defined as less than 40% of total intake from the chilichoice), the concentration of chili was halved for thenext 2-day cycle, and this procedure was continued untilan indifference point was reached. An indifferencepoint was defined as the higher of two consecutiveconcentrations at which the rat consumed at least 40%of the chili diet. The avoidance threshold was then setas the lowest concentration above the indifference point.Thus if a rat consumed 25% of the chili diet at 4 mgchili/g, 44% at 2 mg/g, and 43% at 1 mg/g, 4 mg/g wouldbe taken as the avoidance threshold. (In a few instances


a rat would eat more than 40% at one concentration anddip below this at the next lowest. In these few cases wecontinued to drop the concentration until the criterionof two successive concentrations above 40% was met.)

All animals began with a choice between pure pow-dered chow arid chow with 4 mg/g chili pepper. In 6 of23 cases, rats did not avoid this concentration. Underthese circumstances, on the next 2 days the concentra-tion of chili was doubled, instead of halved, and this was

0)

1II

^1*T3~0

"I

«

1

|

:=

o

16

8

4

2

1

5 x 10-1

2.5 x 10'1

1.25x 1C'1

6.25 x 1C-2

3.12 x 1(T2

1.6x ID'2

7.8 x 1(T3

3.9 x 10'3

2.0 x 10'3

1.0x 10-3

4.9 x 10-«

- • o

- •• 00

oo

- ••• o

••- •• o

o

-

. 0

-

-

o

o

o

-

o

1 1

11 months Chili15% chili Naive

Figure 2. Chili pepper avoidance threshold for 11animals raised on 15% chili pepper for 11 mo and for 12chili-naive controls. (Each point represents thethreshold for one rat.)

continued until the preference dropped below 40% fortwo consecutive 2-day periods. The avoidancethreshold was then defined, as before, as the lowestconcentration avoided above the indifference point.Avoidance thresholds were determined for 11 experi-mental and 12 control rats.

Results

The avoidance thresholds for the 23subjects are graphically displayed in Figure2. The distributions of chili-naive andchili-exposed rats overlap considerably.Four chili-naive rats show by far the lowestavoidance thresholds, but the other eight areindistinguishable from the 11-mo-exposedanimals. Overall, the distributions do notdiffer significantly (Mann-Whitney U = 54,p > .05, one-tailed). The geometric meanavoidance threshold for the exposed groupswas 1.55 mg/g and .27 mg/g for the naivegroup. Insofar as avoidance thresholds canbe taken to be a measure of sensitivity, theevidence for desensitization here is weak.

Experiment 5: Capsaicin Desensitizationand Chili Preference

It is presumed that rats find chili aversivebecause of the piquant effects of capsaicin.These effects are mediated by the chemicalirritant sense. If this is so, then densitiza-tion to the sensory effects of capsaicin shouldeliminate the chili aversion; in some sense,the chili diet would then taste like chowflavored with mild paprika. This experi-ment tested this presupposition.

Method

Six 55-day-old rats served as subjects. They werehoused in individual cages. For the first 8 days of theexperiment, each received sc a daily, increasing dose ofcapsaicin. In accord with the procedure of Jancs6(1960), .1 g of capsaicin was dissolved in 2-3 drops ofethanol and about 20 drops of Tween 80. Water wasthen added to make 10 ml of injection solution. Overa period of 8 days, rats were injected with the followingsequence of capsaicin doses (in milligrams per kilo-gram): 24,31,31,48,65,80,100, and 120. Five minutesbefore each dose, .7 ml of atropine sulfate solution (.54mg/ml) was administered to counter some of the acuteeffects of capsaicin. On Day 9, the animals received notreatments. On Day 10, they were tested for the ef-fectiveness of desensitization: A drop of 1% ammoniawas placed in one eye. None of the six capsaicin-treatedanimals showed any paw-wiping response to this stim-


Table 8Preference for Chili in Capsaicin-Desensitized Rats

Diet choice

Desensitized rat( « = 6 )

Preference8

Control rat

Experiment Preference8

Chow vs. 15% chili chowChow vs. 1% chili chowSugar diet vs. 15% chili sugar diet

35.865.744.5

13 17 1

3.55.9

3.0Sugar diet vs. 1% chili sugar diet 53.2

Note. Experiment 3 controls faced a choice between 2% chili in the sugar diet and plain sugar diet. The sugar diet contained,in grams per kilogram, sucrose, 659; Mazola, 50, casein, 250; Hegsted salt mix, 40; General Biochemicals thiamine-deficient vitaminmix, 1.a Mean percentage of chili diet eaten over 2 days.

ulus. Normal rats average about five paw wipes in the15-sec period following drop introduction. Duringthese first 10 days, rats were fed ad lib on a starch-baseddiet (see Footnote 3). This diet was not used in sub-sequent preference tests since it was possible that thecapsaicin injections could support aversive condi-tioning.

Four consecutive 2-day preference tests were per-formed on these animals. In each case cup position wasalternated across the 2 days to provide a balanced total.The sequence of choices is indicated in Table 8. Re-testing of the paw-wiping response to ammonia, afterthe preference tests, confirmed the stability over timeof the desensitization procedure.

Results

As indicated in Table 8, capsaicin desen-sitization virtually eliminates the aversionto chili. Comparable control data from ratsin Experiments 1 and 3 are included in thetable. Overall, there is neither a preferencenor an aversion to chili-flavored diet, butrather indifference. The difference betweencapsaicin-treated rats and control rats needsno formal statistical verification since thereis no overlap.

Discussion

One cannot fail to be impressed by theresistance shown by laboratory rats to theacquisition of a preference for chili pepper.This contrasts to the apparently effortlessacquisition of this preference by tens ofmillions of humans each year by a processthat seems little more than "mere exposure."Mere exposure over 11 mo, or this same ex-posure coupled with poisoning and safety

experiences, did almost nothing to attenuatethe innate aversion in rats, let alone establisha preference. Simulation of the human ex-perience of gradual introduction led toequally negative results. Explanation ofhuman preferences as resulting, in large part,from desensitization to the irritant effects ofchili pepper was not supported here. Elevenmonths on a piquant diet produced littlechange in the rat's chili-avoidancethreshold.

In contrast, two procedures did signifi-cantly modify or eliminate chili aversions.One was desensitization with systemic in-jections of capsaicin. When the sensory ir-ritant response to chili is abolished, theaversion disappears. This result confirmsthe unpalatability of the irritant propertiesof chili pepper and establishes them as suf-ficient to account for chili aversion. How-ever, there is little evidence that such a de-sensitization actually occurs consequentupon ingestion of chili in concentrationstypically consumed by humans. The secondeffective procedure was a series of seven ex-periences of recovery from illness followingchili ingestion. This might, if extended, leadto an absolute preference. Although thefindings reported here are among the mostsubstantial in the learned-preference liter-ature, they can still be taken to support thedifficulty in establishing learned preferencesin rats, and the major asymmetry betweenthe strong and rapidly acquired learnedaversions and learned preferences. Thepositive findings reported here do not speakdirectly to the fundamental distinction be-


tween hedonic shifts and medicine effects.Given that the major change was producedusing recovery effects and that the smallerpoison-safety effect was based on manipu-lation of the consequences of chili ingestion,it is reasonable to guess that the preferenceshifts shown were motivated by anticipatedconsequences (medicine effects). This isfurther supported by the marked enhance-ment of chili preference during the recoveryexperiment, when testing was carried out inthe deficiency state that chili had in the pastrelieved.

Both the safety-poison and the recoveryprocedures were carried out a rather smallnumber of times when viewed in the contextof a lifetime of experiences (but a largenumber of times in comparison with those inother laboratory studies on preferences andaversions). If such experiences did play arole in human preference, it might reason-ably be proposed that tens or hundreds ofsuch experiences preceded preference ac-quisition, so that the results reported heremight indicate an adequate mechanism.However, both poison-safety and recoveryprocedures seem quite artificial with respectto the actual experience of the human chiliuser (Rozin, 1978; Rozin & Schiller, Note 1).The mere exposure and gradual exposuretreatments, which gave no promising results,are much more in line with the human ex-perience.

The data presented do argue that underlaboratory conditions with domesticatedrats, it would be difficult to establish a modelfor human chili preference. However, onecannot jump from these data to other speciesor other circumstances. It might be quitepossible to establish such preferences inprimates or other domesticated animals,such as dogs. Furthermore, the laboratoryexperiences provided here are a far cry fromthe kinds of experiences that humans typi-cally have with chili pepper. It is introducedand eaten in social settings, in conjunctionwith an elaborated cuisine. Perhaps a moreappropriate test of the hypothesis that chilipepper and similar preferences are almostuniquely human would be to study domes-ticated animals, such as dogs, that are raisedaround the home in chili-eating cultures andconsume chili-flavored food as table left-

overs. Such a study is presently underwayin Mexico (Rozin).

The resistance to acquisition of a chilipepper preference parallels data on labora-tory rats for alcohol and a variety of bittercompounds, and it contrasts with stronghuman preferences for chili, alcohol, coffee,and other innately unpalatable substances.The problem remains as to why such a"simple" preference acquisition should be soclosely limited to humans. Perhaps thecorrect strategy would be to study humans,because this is where the phenomenon re-sides. The use of chili pepper to explore thisfundamental issue in preference and affecthas much to recommend it. Unlike many ofthe other substances in the innately unpal-atable category, chili pepper is safe andnonaddictive.

Reference Note

1. Rozin, P., & Schiller, D. The nature and acquisitionof the preference for chili pepper by humans.Manuscript submitted for publication, 1979.

References

Beidler, L. Phylogenetic emergence of sour taste. InD. A. Denton & J. P. Coghlan (Eds.), Olfaction andtaste V, New York: Academic Press, 1975.

Capretta, P. J., & Rawls, L. H., III. Establishment ofa flavor preference in rats: Importance of nursingand weaning experience. Journal of Comparativeand Physiological Psychology, 1974,86, 670-673.

Deutsch, J. A., & Walton, N. Y. A rat alcoholism modelin a free choice situation. Behavioral Biology, 1977,19, 349-360.

Garcia, J., Ervin, F. R., Yorke, C. H., & Koelling, R. A.Conditioning with delayed vitamin injection.Science, 1967, 755, 716-718.

Garcia, J., & Hankins, W. G. The evolution of bitterand the acquisition of toxiphobia. In D. A. Denton& J. P. Coghlan (Eds.), Olfaction and taste V. NewYork: Academic Press, 1975.

Garcia, J., Kovner, R., & Green, K. F. Cue propertiesversus palatability of flavors in avoidance learning.Psychonomic Science, 1970,20, 313-314.

Grill, H. J. Sucrose as an aversive stimulus. A/euro-science Abstracts, 1975,1, 525.

Hilker, D. M., Hee, J., Higashi, J., Ikehara, S., & Paul-sen, E. Free choice consumption of spiced diets byrats. Journal of Nutrition, 1967,97, 129-131.

Hume, A. The islands of the Bay of Bengal. StrayFeathers, 1874, 2, 29-324.

Jancso, N. Role of the nerve terminals in the mecha-nism of inflammatory reactions. Bulletin of theMillard Fillmore Hospital, Buffalo, New York, 1960,7, 53-77.


Jancso-Gabor, A., & Szolcsanyi, J. The mechanism ofneurogenic inflammation. In A. Bertelli & J. C.Houck (Eds.), Inflammation biochemistry and drugintegration. Amsterdam: Excerpta Medica,1969.

Kalmus, H. The sense of taste of chimpanzees andother primates. Chimpanzee, 1969, 2,130-141.

Maga, J.A. Capsicum. CRC Critical Reviews of FoodScience and Nutrition, 1975, 7, 177-199.

McClearn, G. E. The genetic aspects of alcoholism. InP. G. Bourne & R. Pox (Eds), Alcoholism: Progressin research and treatment. New York: AcademicPress, 1973.

Mello, N. K. A review of methods to induce alcoholaddiction in animals. Pharmacology Biochemistryand Behavior, 1973,1, 89-101.

Osol, A., & Farrar, G. E. The dispensatory of theU.S.A. (25th ed.). Philadelphia: Lippincott,1955.

Parker, L., Failor, A., & Weidman, K. Conditionedpreferences in the rat with an unnatural need state:Morphine withdrawal. Journal of Comparative andPhysiological Psychology, 1973, 82, 294-300.

Richter, C. P. Alcohol as food. Quarterly Journal ofStudies on Alcohol, 1941,1,650-662.

Richter, C. P. Production and control of alcoholiccravings in rats. In H. A. Abramson (Ed.), Trans-actions of the Third Conference, on Neurophar-macology, Princeton, N.J. New York: JosiahMacy, Jr. Foundation, 1956.

Rozin, P. Specific aversions as a component of specifichungers. Journal of Comparative and PhysiologicalPsychology, 1967, 64, 237-242.

Rozin, P. The selection of foods by rats, humans, and

other animals. In J. Rosenblatt, R. A. Hinde, C.Beer, & E. Shaw (Eds.), Advances in the study ofbehavior (Vol. 6). New York: Academic Press,1976.

Rozin, P. The use of characteristic flavorings in humanculinary practice, hi C. M. Apt (Ed.), Flavor: Itschemical, behavioral and commercial aspects.Boulder, Colo.: Westview Press, 1978.

Rozin, P., & Kalat, J. Specific hungers and poisonavoidance as adaptive specializations of learning.Psychological Review, 1971, 78, 459-486.

Siqueland, E. R. Experimental modification of tastepreference. Journal of Comparative and Physio-logical Psychology, 1965, 59, 166-170.

Warren, R. P., & Pfaffmann, C. Early experience andtaste aversion. Journal of Comparative andPhysiological Psychology, 1959, 52, 263-266.

Weiskrantz, L., & Cowey, A. The aetiology of food re-ward. Animal Behaviour, 1963,11, 225-234.

Young, P. T. Appetite, palatability, and feeding habit:A critical review. Psychological Bulletin, 1948,45,289-320.

Xahorik, D. M. Associative and non-associative factorsin learned food preferences. In L. Barker, M. Best,& M. Domjan (Eds.), Learning mechanisms in foodselection. Waco, Tex.: Baylor University Press,1977.

Zahorik, D. M., Maier, S. F., & Pies, R. W. Preferencesfor tastes paired with recovery from thiamine defi-ciency in rats: Appetitive conditioning or learnedsafety? Journal of Comparative and PhysiologicalPsychology, 1974, 87, 1083-1091.

Received January 17,1979 •

Correction to Sandell, Gross, and Bornstein

In the article "Color Categories in Macaques" by Julia H. Sandell, CharlesC. Gross, and Marc H. Bornstein (Journal of Comparative and PhysiologicalPsychology, 1979, Vol. 93, No. 4, pp. 626-635), in Table 1 under the sectionExperiment 2, the entry in the far right column for the Munsell designationshould read "2.5YR 7/10" instead of "2.5R 7/10."

Reversal of Innate Aversions: Attempts to Induce a ... · Reversal of Innate Aversions: Attempts to...

Documents

Transcript of Reversal of Innate Aversions: Attempts to Induce a ... · Reversal of Innate Aversions: Attempts to...