Appetite, 1992, 18, 193-206

Chemoreception for Fat: Do Rats Sense Triglycerides Directly?

ISRAEL RAMIREZ Monell Chemical Senses Center, Philadelphia

Rats given a choice between fluid containing O.l-0.5% triglyceride oil and the same fluid without oil, generally preferred the fluid containing oil. Several experiments indicate that this preference is based on the detection of impurities rather than triglycerides per se. Rats preferred crude triolein to a greater degree than they did highly purified triolein or corn oil. Rats did not show any preference for or aversion to tristearin, a fat that does not decompose as readily as triolein. Rats that have been trained to avoid a dilute suspension of triolein, also avoided an aqueous extract of triolein. Since rats that had been trained to avoid triolein oil also avoided corn oil, it seems likely that different oils may possess similar impurities. Since training rats to avoid mineral oil did not reduce preference for triolein, these substances may have different flavors. It is proposed that rats use fat decomposition products to detect the presence of fats in foods.

INTRODUCTION

The sensory properties of fats and oils (triglycerides) have been getting an

increasing amount of attention in recent years [e.g. Drewnowski, 1990; Mindell et al., 1990; Naim et al., 1987; Ramirez, 19851. However, much of this work has focused on textural aspects of fats [Ackroff et al., 1990; Drewnowski & Schwartz, 1990; Elizalde 8z Sclafani, 1990; Hamilton, 1964; Larue, 1978; Mela, 19881; less is known about human or animal sensitivity to other attributes of fats.

It is well known that enzymes are capable of binding to and acting on triglycerides. One could therefore hypothesize that animals possess oral chemorecep- tors for triglycerides that are similar to such enzymes. Alternatively, one could hypothesize that animals make use of impurities normally present in fats and oils [Dugan, 19761. Some of these impurities originate from the plants the oils were derived from; other impurities originate from partial oxidation of unsaturated fats upon exposure to oxygen [Frankel, 1985; Lillard, 19781. Such compounds might be easier for animals to sense than triglycerides, despite their low concentration.’

These hypotheses are difficult to test because it is difficult to obtain triglycerides that are free from such impurities. Food grade oils contain more than 98% triglyceride [Gunstone & Norris, 19831 but humans can detect impurities merely by

This work was supported by a National Science Foundation grant (BNS 87-19309) an NIH grant (1 ROl DCO1107), and the Monell Chemical Senses Center.

Highly skilled technical assistance was provided by Samantha Keener. I thank D. Greenberg for the helpful comments she provided about an early draft.

Address for correspondence: Monell Chemical Senses Center, 3500 Market Street, Philadelphia, PA 19104-3308, USA

0195-6663/92/030193+ 14 %03.00/O 0 1992 Academic Press Limited

194 I. RAMIREZ

sniffing neat vegetable oil. High purity triolein is commercially available but it is too expensive to use in conventional behavioral tests which require several grams of fat per animal. Alternatively, one could attempt to extract impurities with water and determine whether aqueous extracts of fat taste like fat. Unfortunately, impurities must be poorly soluble in water; otherwise they would be removed by conventional processing. Saturated aqueous solutions of impurities are unlikely to have as strong a flavor as neat oil or fat.

The present report describes some improvements in behavioral technology that overcome such obstacles. It is shown that rats can detect very small amounts of vegetable oil suspended in water. This finding made it practical to test preference for purified triglycerides. Even greater sensitivity could be obtained by training rats to respond to dilute oil suspensions via the conditioned flavor aversion technique. This made it practical to test rats’ responsiveness to aqueous oil extracts.

METHODS

A total of 78 female CD rats (Charles River Breeding Labs., Wilmington, MA) were used. The rats were housed individually in hanging stainless steel cages, maintained on a 12/12 h light/dark cycle at 23-27°C and given food (Purina Laboratory Chow #5001) and water ad libitum except where indicated. The rats were maintained in the laboratory for at least 1 week after arrival to permit them to acclimate to their surroundings before testing.

General Procedure

The procedures used are similar to those employed in previous studies of starch preference [Ramirez, 1991(a)]. Experiments 1, 2 and 3 employed the 24-h two-bottle preference technique. Experiments 4 and 5 employed 30-60 min preference tests. Rats were offered a choice of lipid suspended in water vs. a control mixture lacking the test lipid. The fluids were offered in Richter tubes (Experiment 1) or glass jars (all other experiments). Bottles were placed in the front of the cages 9-12cm apart. All mixtures were formulated on a weight basis because intake was measured by weighing. De-ionized water was used to make up suspensions.

All fluids used in preference tests contained 0.3% xanthan gum (Sigma Chemical Co., St Louis, MO), a suspending agent that does not affect food intake or growth of rats when fed in moderate amounts (Ramirez, 1987). Since xanthan gum is a carbohydrate polymer (Kelco Company, undated), it probably interacts minimally with lipids. Suspensions were made by stirring gently for approximately 1 h, using a motor-driven propeller. Visual inspection indicated that the suspensions were stable for 48 h, although they were made fresh daily.

The viscosity and slipperiness imparted to water by xanthan gum masks the textural effects of small amounts of oil in water. This was verified by measuring viscosity with a Brookfield RVT Micro Viscometer (0.8” cone, Stoughton, MA) over a range of shear rates (shear rate may be regarded as the amount of force exerted on a fluid). As may be seen in Figure 1, the presence of 0.5% triolein had no detectable effect on viscosity.

CHEMORECEPTION FOR FAT 195

3 IO 100

Shear rate (s-‘) IO00

FIGURE 1. Viscosity of xanthan gum vehicle as a function of shear rate with and without 05% triolein. Measurements were taken with a Brooktield micro viscometer using the CP-40 spindle (04” cone). The viscosity of plain water is one centipoise. Each point is the mean of two independent samples. Vertical bars are standard errors (some are too small to be visible). + , vehicle; 0, 0.5% h oil.

Statistical methods. Preference was computed in the usual way, intake of the oil suspension divided by total fluid intake (Ramirez, 1991a). A value of 50% indicates no preference and 100% indicates complete preference for carbohydrate suspension. In most cases, the question involved whether rats could discriminate between two solutions. Therefore, simple t-tests were used to determine whether the preference statistic differed from 50%. The results of these t-tests are given in the figures. Results are always given as the mean*SEM.

EXPERIMENT I

The first experiment demonstrates the high sensitivity of rats to commercial grade corn and olive (triolein) oils.

Method

Ten rats weighing 220 f 1 g were given a choice of 0.5, 0.25, or 0.1% corn oil (Mazola, CPC International, Englewood Cliffs, NJ) vs. vehicle on 3 successive days. The following week, the test was repeated except that “practical” grade triolein (65% triolein = refined olive oil, Sigma Chemical, St Louis, MO) was used instead of corn oil. The mixture containing oil was always on the left side of the cage. On the following week, the rats were given a choice of 0.25% corn oil vs. 0.25% triolein.

Results and Discussion

As may be seen in Figure 2, the rats showed a significant preference for all oil suspensions except for 0.1% corn oil. The preference threshold for oil in water is

196 I. RAMIREZ

0.25% 0.1% Oil concentration

FIGURE 2. Mean preference for corn oil (8) and practical triolein (W) +SEM in Experiment 1. Rats were given a choice of suspended oil vs. vehicle. The dashed line represents 50% preference (i.e. no evidence for discrimination). ‘Preference different from 50% pcO.05. bPreference different from 50% p<O.Ol.

equal to or better than the preference thresholds for sucrose (Ramirez, 1990) and starch (Ramirez, 1991a) on a weight basis. This means that rats can detect oil in mixtures having negligible caloric value. Total fluid intake was 26 f 2 to 3 1 f 2 g/day during the three corn oil tests, and 36+4 to 39 f 5 g/day during the triolein tests (data not shown).

When given a choice of the two oils, rats preferred triolein by 66% [data not shown in figure, t(9) = 2.2, p = 0.051. Since the two oils have similar textural effects, nutritive value and triglyceride content, the greater preference for triolein over corn oil suggested that preference might be influenced by impurities in the oils.

EXPERIMENT 2

The next experiment examined preference for purified triolein and the practical grade triolein used in the first experiment. The unsaturated bonds in vegetable oils react with atmospheric oxygen to produce a variety of compounds that may be sensed (Lillard, 1978). Prolonged contact with oxygen may produce rancidity but flavors resulting from lesser degrees of oxidation may be preferred; many humans regard the oxidation products resulting from deep fat frying, to be appealing. The simplest way to prevent auto-oxidation is to use a fully saturated fat. Tristearin was chosen because it is the saturated form of triolein.

Method

Ten rats weighing 234&-2g were used. In order to habituate the rats to the oil, they were given a series of six preference tests involving 0.25% practical grade

C~O~CEP~ON FOR FAT 197

triolein vs. vehicle. Each preference test was followed by a rest period of l-4 days. After these habituation trials, the rats were randomly divided into two groups and tested on two consecutive days. One group got a choice of practical triolein vs. vehicle on the first day, and 95% pure triolein (Sigma Chemical Co.) vs. vehicle on the second day. The other group was given the same choices but in the reverse order. These tests were repeated the following week, except that 99% pure triolein (synthetic, Sigma Chemical Co.) was used instead of 95% pure triolein. A few days later, the rats were given a choice of 0.25% powdered tristearin (Sigma Chemical Co., 90+ %) vs. vehicle on two tests and 1% tristearin vs. vehicle on a third test. As with the habitation trials, these tests were separated from each other by l-4 days. The bottle containing oil was always placed on the left side of the cage.

The stated purities of the triglyceride preparations are conservative estimates. The supplier classified any triglyceride that was not triolein or tristearin as an “impurity”; most of the “impurities” in these preparations consist of triglycerides ~ntai~ng various fatty acids.

An additional check of purity was done by measuring light absorption at 234 and 268nm (Gray, 1985). Oxidation of unsaturated fats results in compounds having conjugated double bonds that absorb ultraviolet light. Therefore, the lower the absorption of ultraviolet, the purer the oil (Gray, 1985). The 95% triolein prep- aration had 9 and 10% of the optical density of practical triolein at these wavelengths respectively. The 99% triolein had 4 and 5% of the optical density of triolein at these wavelengths. According to the supplier, the tristearin preparation has no, or very low levels of, unsaturated fats. It was not possible to assess light absorption of the tristearin at 234nm because the only solvent in which it could be dissolved, chloroform, does not transmit light at this wavelength. At 268nm, the optical density of tristearin was less than lo/c of that of the practical triolein.

Purified triolein preparations were stored under nitrogen until the first day of each test; the portion of the 95-99% triolein that was used on the second day, was stored frozen at -35°C under air for 24 h. Tristearin is a solid whereas triolein is a liquid at room temperature.

Oil preference did not become reliable until after the third habituation trial; thereafter, oil preference stabilized at 79-81% (data not shown). Rats were able to detect practical and 95% pure triolein (Figure 3). Preference for 99% pure triolein did not reach a conventional level of significance [f(9) = 1.98, p = 0.08 two-tailed, see Figure 31. This result suggested that rats might be responding to trace levels of impurities. A more definitive test is that the preference statistic was lower for rats ingesting 99% pure triolein than for rats ingesting practical triolein [t(9)= 2.53, p=O*O2); preference for 95% triolein was not significantly different from that for practical triolein (p>O*2). Evidently, the rats are capable of detecting the low levels of impurities present in the 95% triolein, even when the con~ntration of oil is only 0.25%.

The rats appeared to be indifferent to tristearin; preference ratios were near to 50% on all three tests (Figure 4). Since tristearin lacks unsaturated bonds, this result suggests that rats’ preference for triolein was influenced by products resulting from the oxidation of the double bonds in triolein.

198 I. I&‘WIREZ

0

I.00 0.25% triolein

b

b T

0.80

i ; 0.60 & xl & 3 0.40

i- I.. : j ':,' .: '. .',

rracnca~

b I : : ;..: ‘.:.:

a

T T _,‘.. ._., ,.. ,. . . . . . . ;. . . . .

;:,; .: ::,.:., ::: .>

_.+L+;_;_____ ____--- .: . . .: :i : .: ,; ; ,, .., . . . ., .:. .’ :

.., : :: .:..;.. ::. .::. .-.-..: ;... : . :..: :;:y.,, :. ,_ ,-.. : ..:. . : : . . . . y .,: :, (: :.: ,.. 1.

., ._ : ;.,... : :,,,, ,:,;:,;;.1:. .. 95% Practical 99%

Oil purity

FIGLXE 3. Mean preference for practical and purified triolein + SEM in Experiment 2. Rats were given a choice of suspended oil vs. vehicle. The dashed line represents 50% preference. “Preference different from practical triolein p<O-05. bPreference different from 50% p<O,Ol.

0.25% 0.25% Concentration

FIGURE 4. Mean preference for tristearin -)-SEM in Experiment 2. Rats were given a choice of suspended fat vs. vehicle. The dashed line represents 50% preference.

CHJ2iORECEPTION FOR FAT 199

The results of Experiments I and 2 were inconsistent in one respect; preference for oil appeared immediately in Experiment I whereas it took several exposures to appear in Experiment 2. This might have been due to using oil that had aged a couple of months since it had been purchased. Another potential problem with the previous experiments is that the oil emulsions were always presented on the same side of the cage. It is appropriate to check for the possibility that side preferences influenced the results. In Experiment 3, a group of rats was given a choice of 0.25% triolein vs. vehicle over several days. On most days, the oil used was a couple of months old. On the test day, the rats were given a new batch of oil,

Inspection of indi~d~l data in Experiment 2 suggested that some of the rats were initially avoiding the oil emulsion. Indi~dual data from Experiment 3 were examined in order to confirm this impression.

Animals

A total of nine rats were used. Body weights were not recorded, but the rats were the same age as those used in other experiments (8 weeks).

Method

The methods used are similar to those used in the previous experiments. The rats were given a choice of 0.25% triolein vs. vehicle on 6 cons~utive days. The oil used on most days was approximately 2 months old; on the 4th day, the oil used had been recently purchased and not opened until that day. In order to determine whether off- taste components could be easily stripped from the oil, a sampIe of oil was incubated overnight with activated charcoal (50% of the volume of the oil) and then centrifuged to separate the charcoal from the oil [Gunstone & Norris, 19831. This oil was used in the sixth preference test. The rats were given plain water for 5 days and then tested one last time for their oil preference.

The position of the oil suspension was alternated (left-right) daily, except that the oil was on the left-hand side of the cage on the sixth and seventh tests.

Individual data are shown in Figure 5 because this provides the most compelling picture of what happened. It may be seen that after the first day, two rats consistently avoided the oil, one rat showed slight preference, and the remaining rats showed strong oil preference. The two oil-avoiding rats showed reduced aversion to oil on the seventh trial. Statistical corroboration of these individual differences was obtained by calculating the correlations among preference scores across the 7 days. A total of 21 independent correlations were computed and all had absolute values greater than 0*67 (all ~~0.05).

These results demonstrate that the absence of a significant group preference for oil does not imply that the rats cannot sense the oil.

Inspection of Figure 5 also indicates that changing to a fresh batch of oil or to charcoal-treated oil, had no obvious effect on preference. This result suggests that oil freshness is not a crucial determinant of preference.

200 1. RAMIREZ

Daily preference of Individual rats

I 2 3 4 5 6 7

FIGURE 5. Preference for triolein by individual rats on successive tests in Experiment 3. The oil used on most days was 2 months old; on the fourth day, a fresh bottle of oil was used. The oil used in the sixth preference test had been treated with charcoal to reduce impu~ties. The position of the oil suspension was alternated (left-right) daily, except that the oil was on the left-hand side of the cage on the sixth and seventh tests.

Because the auto-oxidation of oil yields diverse chemical products, having diverse flavors to humans (Frankel, 1985), it seems likely that oils possess both appealing and aversive flavor notes. Perhaps indi~dual differences in oil preference arise because some rats respond more strongly than others to the aversive flavor notes.

EXPERIMENT 4

Rats were trained to avoid dilute oil suspensions in order to examine the role of the water soluble impurities in oil. Two experiments were conducted. The first experiment was designed to determine whether rats are capable of detecting water soluble impurities. The second experiment included a control group that made it possible to make inferences about whether the water soluble impurities taste like oil. One test included mineral oil because this substance is sometimes assumed to have sensory properties similar to those of real fats (Ackroff et ai., 1990; Elizalde & Sclafani, 1990; Hamilton, 1964; Larue, 1978; Mindell et al., 1990).

Animals

A total of nine rats weighing 202 _t 2 g were used in Experiment 4A and 16 rats weighing 216+ 1 g were used in Experiment 4B.

Method

The procedure was modified from that used in previous studies involving starch (Ramirez, 1991b). Unlike previous experiments, rats were tested during the first hour

CHEMORECEPTION FOR FAT 201

of darkness of the dark part of the light-dark cycle under dim red illumination; this modification permitted rats to be tested with much less water deprivation. Periods of habituation to the xanthan gum vehicle (24 h availability of vehicle) were alternated with training trials. Beginning 1 week after the rats arrived in the laboratory, the rats were given the xanthan gum vehicle instead of plain water for 2 days.

On training trials, the rats were water deprived for 24 h, and allowed to drink an aqueous suspension of triolein for 1 h. At the end of the hour, the experimental rats were injected intraperitoneally with 1 ml 0.15 M lithium chloride/lOOg body weight (Nissenbaum, & Scalafani, 1987). In Experiment 4B, but not in Experiment 4A, a control group was given vehicle to drink and injected with lithium chloride in the same way as the experimental rats. Training trials were conducted no more often than twice a week, with at least 2 days recuperation after each lithium chloride treatment. All mixtures contained 0.3% xanthan gum. Rats were allowed to drink 0.3% xanthan gum vehicle for 24-h periods at various times between training trials. After training was complete, rats were given only plain water to drink, except during the tests described below.

In Experiment 4A two training trials were given; the first using 0.5% triolein and the second using 0.1% triolein as the conditioned stimulus. In Experiment 4B three training trials were given, all with 0.1% triolein as the conditioned stimulus. A greater number of training trials were required in Experiment 4B, probably because the weaker 0.1% triolein was less salient than 0.5% triolein.

After training was complete, the rats were given a series of l-h preference tests. The rats were water deprived for 4 h prior to each preference test and then given a choice of 0.3% xanthan gum vehicle and the test suspension for 1 h. Only one preference test was conducted per day. In Experiment 4A, the rats were tested for their preference for 0.1% triolein, 0.5% mineral oil (Squibb, Princeton, N.J.), and the aqueous oil extract. The aqueous oil extract was obtained by storing water with 0.5% triolein in a flask overnight, and then passing the mixture through a separatory funnel to separate the oil and water. A sample of plain water was treated in the same way except that it never contained oil and did not need to go through the separatory funnel; this water was used to prepare the comparison mixture (vehicle). In Experiment 4B, the rats were tested for their preference for 0.1% triolein, 0.1% corn oil, and aqueous oil extract.

The triolein was the same practical grade used in the previous experiments. The ultraviolet absorption technique (see Experiment 2) indicated that mineral oil did not contain substantial amounts of the fat oxidation products that are present in vegetable oil. The optical density of mineral oil at 234 and 268 nm was < 3% of that of practical triolein.

Results and Discussion

In Experiment 4A, the rats avoided triolein, mineral oil, and the aqueous oil extract (see Figure 6). It is surprising that rats could detect and avoid a 0.5% suspension of mineral oil (p < 0.005); the implications of this finding are examined in Experiment 5.

In Experiment 4B, trained rats showed greater avoidance of the test mixtures than did the control rats (all p-c O-01). Furthermore, in Experiment 4B, rats trained to avoid oil, avoided triolein, corn oil and the aqueous oil extract; the control rats

202 I. RAMIREZ

t All rots trained to avoid triolein 0.80

8 0.60 t

Triolein Mineral oil l-l20 extract

J 0.60 5

& g 0.40

Triolein Corn oil H,O extract

Substance offered

FIGURE 6. Upper panei: mean preference for various mixtures+ SEM in Experiment 4A. Lolver panel mean preference for various mixturesIf:SEM in Experiment 4B. For both experiments, rats were trained to avoid triolein and then given a series of 1 h preference tests involving a choice of suspended oil vs. vehicle. The dashed line represents 50% preference. *Preference different from 50% p < O-05. bPreference diffrent from 50% p < O-01.

did not avoid triolein or the aqueous oil extract, but they did show a very slight avoidance of corn oil (see Figure 6). Because trained rats showed a greater degree of avoidance than did control rats, these results indicate that corn oil tastes like triolein.

In both experiments, rats avoided water that previously contained oil. Because triglycerides are virtually insoluble in water, the present results indicate that rats are capable of detecting compounds that diffuse from the oil and dissolve in the water.

EXPERI~NT 5

Experiment 4A indicated that rats could detect a 03% suspension of mineral oil but it did not make it possible to determine whether mineral oil tastes like vegetable oil because Experiment 4A did not include a control group that was not conditioned to avoid oil. In this experiment, a control group was included to test for non-specific effects of flavor aversion conditioning. This group was injected with lithium chloride

CHEMORECEPTION FOR FAT 203

after drinking xanthan gum vehicle and then tested for their avoidance of oil. The experimental groups were trained to avoid either triolein or mineral oil. These groups were subsequently tested for their avoidance of triolein and mineral oil.

Animals

A total of 24 rats weighing 229 f 2 g were used.

Method

The training procedure was similar to that employed in Experiment 4, but certain changes were instituted because they were found to promote formation of taste aversions. The most important change was that rats were injected with lithium chloride within 10 min after they began drinking the test fluid. The other change was that a more systematic sequence of habituation and training trials was instituted. These habituation and training trials were conducted on Mondays, Wednesdays, and Fridays.

On habituation trials, rats were water deprived for 24 h and given xanthan gum vehicle to drink overnight; this was done twice. The rats were then given either xanthan gum vehicle, 0.5% mineral oil, or 0.5% triolein to drink and injected with lithium chloride within 10 min after beginning drinking. After the lithium chloride injection, the rats were given plain water to drink overnight. The rats were then given two more habituation trials plus one training trial. Other procedural details, including the dose of lithium used and the time of testing, were the same as in Experiment 4.

After the second lithium chloride treatment, the rats were water deprived and tested for their preference for 0*5% triolein vs. vehicle for 30 min. They were given plain water overnight, water deprived again and tested for their preference for 0.5% mineral oil vs. vehicle in a second 30-min test.

Results and Discussion

The control group, rats injected with lithium after ingesting vehicle, avoided mineral oil in the choice test (see Figure 7). Nevertheless, rats trained to avoid mineral oil, showed a stronger aversion to this substance than did control rats [t(14)=5*3, p<O-0011 and rats trained to avoid triolein [t(14)=3*2, p<O*Ol, see Figure 71. Rats that were trained to avoid triolein did not differ from control rats in their aversion to mineral oil [t(14)= l-3, n.s., see Figure 71.

Conversely, rats trained to avoid triolein, showed a stronger avoidance of triolein than did control rats [t( 14) = 2.7, p < O-021, whereas rats trained to avoid mineral oil did not show a significantly greater avoidance of triolein than did control rats [t(14) = 1.5, p>O.15]. However, the groups trained to avoid triolein and mineral oil did not differ by a conventional t-test [t(14)= 1.1, n.s.1. The mean preference for the triolein group is inflated by one rat that preferred triolein but drank only 1*6ml of fluid although all other rats preferred vehicle and drank substantially more fluid, 6.6 to 30ml. In order to minimize the effects of an extreme value, without arbitrarily discarding data, the triolein and mineral oil groups were compared with a non- parametric test (U-test); this produced evidence that the mineral oil and triolein groups differed (U = 14, p =0.06, two tailed). These tests, taken together, indicate

204 I. RAMIREZ

Rats trained to avoid

t

m Nothing 0.80

Mineral oil

triolein

Q 0.60 2 0 _-----___--------

l------ ---_

c; 0.40

0.20

0.00 Mineral oil Triolein

Substance offered

FIGURE 7. Mean preference for 0.5% practical triolein or 0.5% mineral oil vs. vehicle in Experiment 5. The dashed line represents 50% preference. ‘Preference different from 50% ~~0.05. *Preference different from 50% ~~0.01.

that dilute suspensions of triolein and mineral oil are discriminable from each other. However, these results do not prove that mineral oil and triolein taste completely different from each other; they might possibly have some flavor notes in common.

Understanding how rats sense 0.5% mineral oil will require more research than can be justified at present. Perhaps the mechanisms are analogous to those for triglyceride oils (impurities retained during purification or acquired during process- ing and storage). The important point is that mineral oil cannot be assumed to be an odorless, tasteless fat mimic that only stimulates touch sensors.

GENERAL DISCUSSION

The present results show that the sensory and motivational properties of fats may be studied using the same techniques that have been used for carbohydrates. Rats often prefer oil suspensions containing as little as 0.25% oil in water. Even though no attempt was made to assess the lower limits of rats’ sensitivity to triglyceride oil, their ability to detect oil is certainly at least as acute as that for common sugars (see Ramirez, 1990). It has been claimed that people are more sensitive to and more aware of sugars in foods than they are of fats (Drewnowski, 1990; Drewnowski & Schwartz, 1990). This is probably not true for rats. It might be appropriate to re-examine this issue for humans.

The present studies attempted to define the stimuli to which rats respond when they sense fat. The most obvious candidate is triglycerides. After all, commercial grade vegetable oils are more than 98% triglyceride (Gunstone & Norris, 1983). However, impurities are probably more important stimuli than triglycerides per se,

since: rats can apparently discriminate between oils from different species of plants (Experiment 1); rats respond poorly to tristearin and highly purified triolein

CHEMORECEPTlON FOR FAT 205

(Experiment 2); rats can be trained to avoid aqueous extracts of oil that presumably contain no triglycerides (Experiment 4); rats can discriminate between mineral oil and triolein despite their having a similar texture (Experiment 5).

It is not proposed that triglycerides are flavorless; rather, rats use their ability to sense impurities to detect triglycerides. Conventional chemical sensory stimuli are thought to initiate a series of chemical reactions within receptor cells that result in a signal being transmitted to the brain. For fats, and probably other substances (Ramirez, 1991~) some of the chemical reactions occur outside of the cell, indeed, outside of the body, before the oil is ingested. In order for this concept to be useful, it would have to be assumed that most fats and oils possess similar favors because they contain similar impurities. Available data [Experiment 4 and (Larue, 1978)] support this idea. Two mechanisms would produce such an effect: auto-oxidation of unsaturated fatty acids would tend to produce a similar array of compounds from diverse oil types, as long as unsaturated fatty acids are present; and lipid soluble biochemicals, present in many species, tend to accumulate in fats.

The present experiments do not allow one to identify the compounds that rats sense. The lack of preference for tristearin suggests that compounds resulting from the oxidation of unsaturated bonds might be responsible. Isolating and identifying the compounds will be difficult because triglyceride oils contain trace levels of many different substances (Dugan, 1976; Lillard, 1978). According to Dugan (1976), as many as 71 different compounds may contribute to the flavor of oxidatively damaged soybean oil. Humans can detect a few parts per billion of some of these compounds (Dugan, 1976; Lillard, 1978).

The concept that rats detect triglycerides via decomposition products and impurities, leads to the idea that the sensory properties of oils may be more complex than those of conventional tastants. Humans use the word rancid to refer to oils having large amounts of oxidized compounds. Since the compounds responsible for rancidity are present, in small concentrations, in any oil exposed to air (Frankel, 1985) it is probable that the oils used in the present studies contained some aversive flavor components, in addition to preferred flavor compounds. In other words, oils always contain a mixture of preferred and avoided flavors. This probably accounts for the occasional avoidance of oil by some rats (Experiments 2 and 3).

Textural cues have been emphasized by previous investigators (e.g. Ackroff et al., 1990; Drewnowski & Schwartz, 1990; Elizalde & Sclafani, 1990; Hamilton, 1964; Larue, 1978; Mela, 1988). The most common argument for this idea rests on evidence that rats are attracted to mineral oil (e.g. Ackroff et al., 1990; Elizalde & Sclafani, 1990; Hamilton, 1964). This is weak evidence since it is tantamount to claiming that if two substances are preferred, they must share a similar flavor. Other data from previously published studies suggest that textural cues may not fully account for the perception of fats. Mela (1988) reported that addition of a texture- modifying agent to dairy products did not fully reproduce the sensory effects of fat. Larue (1978) found that ostensibly anosmic rats that had been trained to avoid Vaseline, showed a greater aversion to that substance than to genuine fats, whereas rats trained to avoid genuine fats showed a greater aversion to fats than to Vaseline.

The present results bolster the idea that non-textural cues may play an important role in fat perception. At low concentrations, the textural effects of oil are so slight that it is hard to imagine that rats could use them to discriminate between corn, practical triolein, purified triolein, and mineral oil. Psychophysical difference thres- holds for fluid texture have not been studied in animals or humans, but available

206 I. RAMIREZ

data suggest that humans are not able to make fine discriminations in solution viscosity (Christensen, 1984). If rats can sense the impurities present in very dilute suspensions, they must also be able to sense them when offered high levels of fat. Therefore, fat substitutes which mimic fat’s textural effects will not be completely satisfactory unless they include some of the compounds normally present in real fats.

REFIXENCES

Ackroff, K., Vigorito, M. & Sclafani, A. (1990) Fat appetite in rats: The response of infant and adult rats to nutritive and non-nut~tive oil emulsions. Appetite, fS, 171-188.

Christensen, C. M. (1984) Food texture perception. Advances in Food Research, 29, 159-198. Drewnowski, A. (1990) Dietary fats: perceptions and preferences. Journal of the American

College of Nutrition, 9, 431-435. Drewnowski, A. & Schwartz, M. (1990) Invisible fats: sensory assessment of sugar/fat

mixtures. Appetite, 14, 203-211. Dugan, L. Jr (1976) Lipids. In: 0. Fennema (Ed.) Principles of Food Science. Part 1 Food

Chemistry, pp. 139-203. N.Y.: Marcel Dekker. Elizalde, G. & Sclafani, A. (1990) Fat apetite in rats: Flavor preferences conditioned by

nutritive and non-nut~tive oil emulsions. Appetite, f-5, 189-197. Frankel, E. N. (1985) Chemistry of autoxidation: mechanism, products and flavor signifi-

cance. In: D. B. Min & T. H. Smouse (Eds) Flavor Chemistry of Futs and Oils, American Oil Chemists’ Society, pp. l-37.

Gray, J. I. (1985) Simple chemical and physical methods for measuring flavor quality of fats and oils. In: D. B. Min & T H. Smouse (Eds) Flavor Chemistry of Fats and Oils, American Oil Chemists’ Society, pp. 223-239.

Gunstone, F. D. & Norris, F. A. (1983) Lipids in Foods: Chemistry, Biochemistry and ~chnofo~gy. Oxford: Pergamon Press.

Hamilton, C. L. (1964) Rats’ preference for high fat diets. JournuI of Comparative and Physio~ogicaf Psychology, 58, 459-460.

Kelco Company. Kelco Xanthan Gum, 2nd edition, revised. San Diego: Kelco Co., undated. Larue, C. (1978) Oral cues involved in the rat’s selective intake of fats. Chemical Senses and

Flavour, 3, 1-6. Lillard, D. A. (1978) Chemical changes involved in the oxidation of lipids in foods. In: M. K.

Supran (Ed.) Lipids as a Source of Flavor, pp. 68-80. Washington D.C.: American Chemical Society.

Mela, D. (1988) Sensory assessment of fat content in fluid dairy products. Appetite, IO, 37-44. Mindell, S., Smith, G. P. & Greenberg, D. (1990) Corn oil and mineral oil stimulate sham

feeding in rats. Physiology and behavior, 48, 283-287. Naim, M., Brand, 3. G. & Kare, M. R. (1987) The preference-aversion behavior of rats for

nutritionally-controlled diets containing oil or fat. Physioiogy and Behavior, 39, 285-290. Nissenbaum, J. W. & Sclafani, A. (1987) Qualitative differences in polysaccharide and sugar

tastes in the rat: a two-carbohydrate taste model. Neuroscience & Biobehavioraf Reviews, II, 187-196.

Ramirez, I. (1985) Oral stimulation alters digestion of intragastric oil meals in rats. American Journal of Physiology, 17, R459-R463.

Ramirez, I. (1987) Diet texture, moisture and starch type in dietary obesity. PhysjoZogy and behavior, 41, 149-154.

Ramirez, I. (1990) Why do sugars taste good? Neuroscience and Biobehavioraf Reviews, 14, 125-134.

Ramirez, I. (1991a) Chemoreception for an insoluble nonvolatile substance: starch taste? American Journal of Physiology, 260, R192-R199.

Ramirez, I. (1991b) Does starch taste like Polycose? Physiology and Behavior, 50, 389-392. Ramirez, I. (1991~) Chemoreception for cellulose. Chemical Senses, 16, 691-698.

Received I.2 barbs 1991. revision 13 Nouember 1991